Electromagnetic wave sensing device

Abstract

To provide an electromagnetic wave sensing device with which it is possible to quantitatively measure the percentage of water content of a sample that would change depending on position.SOLUTION: A relationship between the transmitted wave measurement data of amplitude and phase change amounts of a transmitted wave measured by an electromagnetic sensing unit 10 and a measurement unit 20 and reflected wave measurement data of amplitude and phase change amounts of a reflected wave and the water content of a sample is regressively analyzed, and a regression equation composed of the water content of the sample as an objective variable and the transmitted wave measurement data and reflected wave measurement data as explanatory variables is determined by an analysis unit 30. Therefore, it is made possible to quantitatively measure the water content of the sample that would change depending on position.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an electromagnetic wave sensing device capable of quantitatively estimating the amount of a substance contained in a sample. [Background technology]

[0002] Patent Document 1 discloses an analytical device that irradiates an electromagnetic wave onto a measurement object, creates a calibration curve showing the relationship between the amount of change in phase or amplitude of the reflected wave or transmitted wave of the electromagnetic wave between a state in which an object is not present and a state in which an object is present, and the amount of a specific substance contained in the measurement object, and determines the amount of the specific substance contained in the measurement object from the calibration curve. Patent Document 2 discloses a moisture content measuring device that detects the moisture content of a granular or amorphous bulk material by irradiating the material with electromagnetic waves, determining the slope of a straight line that linearly approximates multiple relationships between pairs of amplitude changes and phase differences detected, and applying the slope to a calibration curve.

[0003] Fig. 55 is a functional block diagram showing the configuration of the moisture content measuring device disclosed in Patent Document 2. The moisture content measuring device shown in Fig. 55 includes electromagnetic wave transmitting means for irradiating an object with electromagnetic waves having a frequency of 100 kHz to 1 THz, electromagnetic wave receiving means for receiving the reflected or transmitted electromagnetic waves, a measuring device for measuring the phase difference and amplitude of the transmitted and received waves, and an analyzing device for comparing the amplitude and phase difference of the transmitted and received waves to detect and analyze changes in amplitude and phase difference. Prior to measuring the object to be inspected, rice with a moisture content of 24.62% is passed through the processing line, and electromagnetic waves with a frequency of 4 GHz are irradiated from the electromagnetic wave transmitting means. The transmitted electromagnetic waves are received by the electromagnetic wave receiving means, and the amplitude change and phase difference are detected by the measuring device. When the phase difference detected by the measuring device is plotted on a graph with the horizontal axis and the detected amplitude change on the vertical axis, a straight line is formed, and the slope is obtained from the equation of the straight line by the analyzing device. Similarly, the slope of the straight line is obtained by the analyzing device for the cases of moisture content of 22.03% and 20.22%. Next, the slope obtained on the horizontal axis of the moisture content is plotted on a graph with the vertical axis of the slope obtained and the moisture content is plotted on the graph shown in FIG. 56. This graph can be used as a calibration curve. Next, the rice to be inspected is fed into the processing line, and electromagnetic waves with a frequency of 4 GHz are irradiated from the electromagnetic wave receiving means, the transmitted electromagnetic waves are received by the electromagnetic wave receiving means, the amplitude change and phase difference are detected by a measuring device, and the slope of a straight line graph of the phase difference and amplitude change is found by an analyzing device, and the moisture content of the rice to be inspected can be found by applying this to the previously obtained calibration curve of slope and moisture content, which is the graph in Figure 56. For example, if the slope found is 0.2, the moisture content can be found to be approximately 13.4% from the calibration curve shown in Figure 56. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-188995 [Patent Document 2] Patent No. 6253096 Summary of the Invention [Problem to be solved by the invention]

[0005] For example, when the moisture content of a sample such as a piece of wood or lumber is to be measured using the conventional analysis device disclosed in the cited document 1, the dielectric constant and dielectric loss of the inside of the piece of wood or lumber are not uniform, so when the sample is irradiated with electromagnetic waves, the phase and amplitude changes of the reflected and transmitted waves from the sample vary depending on the position of the sample to be measured. This causes the measured values ​​to vary, making it difficult to create a highly accurate calibration curve. Furthermore, the conventional moisture content measuring device disclosed in Patent Document 2 is an effective means when the object to be measured is a material with a uniform dielectric constant and dielectric loss; however, with conventional moisture content measuring devices, for objects such as pieces of wood or lumber whose internal dielectric constant and dielectric loss are not uniform and which do not yield quantitative values, it is difficult to obtain a correlation between the slope of the linear approximation, making it difficult to create a calibration curve for accurately determining the content of a specific substance.

[0006] Here, in the case of samples such as pieces of wood or lumber, the moisture content varies depending on the position on the sample and is not a constant amount, but in reality it is necessary to quantitatively estimate the moisture content of the sample as a constant amount. However, conventional analytical devices and conventional moisture measurement devices have a problem in that it is difficult to quantitatively estimate the moisture content of a sample, the moisture content of which varies depending on the position. Therefore, the present invention aims to provide an electromagnetic wave sensing device capable of quantitatively estimating the content of a substance contained in a sample, the content of which varies depending on the position. Specifically, when the sample is lumber or wood chips, an electromagnetic wave sensing device capable of quantitatively estimating the moisture content of the lumber or wood chips is provided, and when the sample is a solution, an electromagnetic wave sensing device capable of quantitatively estimating the concentration of a solute in the solution is provided. [Means for solving the problem]

[0007] The electromagnetic wave sensing device of the present invention, which can achieve the above-mentioned object, comprises an electromagnetic wave sensing section which irradiates an electromagnetic wave on a sample which is moved at predetermined intervals and outputs a transmitted wave and / or a reflected wave, a measurement section which receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing section every time the sample moves a predetermined interval and measures either the amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined interval or the amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined interval, first transmitted wave measurement data of the amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined interval and / or first reflected wave measurement data of the amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined interval, which are previously measured by the measurement section, and and an analysis unit which performs a regression analysis of the relationship between the amount of the inclusion contained in the sample and the amount of the inclusion contained in the sample to obtain a regression equation with the amount of the inclusion in the sample as a response variable and the first transmitted wave measurement data and the first reflected wave measurement data as explanatory variables, wherein the amount of change in amplitude and phase is the amount of change in amplitude and phase from when the sample is not present, and the sensing unit and the measurement unit substitute second transmitted wave measurement data of the amount of change in amplitude and phase of the transmitted wave measured every time the measured sample moves a predetermined distance and second reflected wave measurement data of the amount of change in amplitude and phase of the reflected wave measured every time the measured sample moves a predetermined distance into the regression equation obtained by the analysis unit, thereby estimating the amount of the inclusion contained in the measured sample. Another electromagnetic wave sensing device of the present invention, which can achieve the above-mentioned object, includes an electromagnetic wave sensing unit that irradiates an electromagnetic wave on a sample that is moved at predetermined intervals and outputs a transmitted wave and / or a reflected wave, a measurement unit that receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing unit every time the sample moves a predetermined distance and measures either the amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined distance or the amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined distance, first transmitted wave average measurement data of the average amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined distance and / or first reflected wave average measurement data of the average amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined distance, which are previously measured by the measurement unit, and and an analysis unit which performs a regression analysis of the relationship between the amount of inclusion contained in the sample and the amount of inclusion contained in the sample, and determines a second regression equation with the amount of inclusion in the sample as a response variable and the first transmitted wave averaged measurement data and the first reflected wave averaged measurement data as explanatory variables, wherein the changes in amplitude and phase are the changes in amplitude and phase from when the sample is not present, and the sensing unit and the measurement unit substitute second transmitted wave averaged measurement data, which is the average amount of change in amplitude and phase of the transmitted wave measured every time the measured sample moves a predetermined distance, and second reflected wave averaged measurement data, which is the average amount of change in amplitude and phase of the reflected wave measured every time the measured sample moves a predetermined distance, into the second regression equation determined by the analysis unit, thereby estimating the amount of inclusion contained in the measured sample. Furthermore, still another electromagnetic wave sensing device of the present invention that can achieve the above-mentioned object includes an electromagnetic wave sensing unit that irradiates an electromagnetic wave on a sample that is moved at predetermined intervals and outputs a transmitted wave and / or a reflected wave, a measurement unit that receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing unit every time the sample moves a predetermined distance and measures either an amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined distance or an amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined distance, and a VSWR every time the sample moves a predetermined distance, first transmitted wave average measurement data of an average value of the amount of change in amplitude and phase of the transmitted wave every time the sample moves a predetermined distance and / or first reflected wave average measurement data of an average value of the amount of change in amplitude and phase of the reflected wave every time the sample moves a predetermined distance, which are previously measured by the measurement unit, first VSWR average measurement data of an average value of the VSWR every time the sample moves a predetermined distance, and and an analysis unit which performs a regression analysis of the relationship with the amount of inclusion contained therein, and determines a third regression equation in which the amount of inclusion in the sample is used as a response variable, and the first transmitted wave averaged measurement data and / or the first reflected wave averaged measurement data and the first VSWR average measurement data are explanatory variables, and the changes in amplitude and phase are the amplitude and phase from when the sample is not present. The sensing unit and the measurement unit estimate the amount of inclusion contained in the sample by substituting, into the third regression equation determined by the analysis unit, second transmitted wave averaged measurement data which is the average amount of change in amplitude and phase of the transmitted wave measured every time the measured sample moves a predetermined distance, second reflected wave averaged measurement data which is the average amount of change in amplitude and phase of the reflected wave measured every time the measured sample moves a predetermined distance, and second VSWR averaged measurement data which is the average value of VSWR measured every time the measured sample moves a predetermined distance.

[0008] In the electromagnetic wave sensing device of the present invention, in the measurement unit and the analysis unit, the amount of change in the amplitude and phase of the transmitted wave may be represented by the amount of change in the real part and imaginary part of the complex representation of the transmitted wave, and the amount of change in the amplitude and phase of the reflected wave may be represented by the amount of change in the real part and imaginary part of the complex representation of the reflected wave. In addition, in the electromagnetic wave sensing device of the present invention, the electromagnetic wave sensing unit includes a metallic sensing housing in which at least one antenna capable of receiving electromagnetic waves from the sample or the sample to be measured is provided, a metallic input section which feeds the sample or the sample to be measured into the inside of the sensing housing, and a metallic output section which feeds the sample or the sample to be measured from the sensing housing, and radio wave absorbers may be attached to the inner wall surfaces of the sensing housing excluding the surface in which the antenna is provided, and to the inner wall surfaces of the input section and the output section. Furthermore, in the electromagnetic wave sensing device of the present invention, the sample and the measured sample are lumber, and the amount of a substance contained in the sample and the measured sample is taken as the moisture content of the lumber. Furthermore, in the electromagnetic wave sensing device of the present invention, the sample and the measured sample are wood chips stored in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is the moisture content of the wood chips stored in the container. Furthermore, in the electromagnetic wave sensing device of the present invention, the sample and the measured sample are solutions placed in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is the concentration of the solute in the solution placed in the container. Furthermore, in the electromagnetic wave sensing device of the present invention, the sample and the measured sample are aqueous solutions placed in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is either the sugar content, salt concentration, or alcohol content of the aqueous solution placed in the container. Effect of the Invention

[0009] In the electromagnetic wave sensing device of the present invention, a regression analysis is performed on the relationship between the transmitted wave measurement data of the amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance and the reflected wave measurement data of the amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, and the content of the ingredients contained in the sample, to obtain a regression equation in which the content of the sample is the objective variable and the transmitted wave measurement data and the reflected wave measurement data are explanatory variables, making it possible to quantitatively estimate the content of a sample whose content changes depending on the position. Furthermore, by performing a regression analysis on the relationship between the transmitted wave average measurement data, which is the average amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance, and the reflected wave average measurement data, which is the average amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, and the content of the inclusions contained in the sample, and deriving a second regression equation in which the content of the sample is the objective variable and the transmitted wave average measurement data and the reflected wave average measurement data are explanatory variables, it becomes possible to make a highly reliable quantitative estimation when estimating the content of a sample whose content changes depending on the position. Furthermore, by performing a regression analysis on the relationship between the transmitted wave average measurement data of the average amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance, the reflected wave average measurement data of the average amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, and the VSWR average measurement data of the average VSWR each time the sample moves a predetermined distance, and the amount of inclusions contained in the sample, a third regression equation is obtained in which the amount of inclusions in the sample is the dependent variable and the transmitted wave average measurement data, the reflected wave average measurement data, and the VSWR average measurement data are explanatory variables, which makes it possible to make a more reliable quantitative estimation when estimating the amount of inclusions in a sample whose amount changes depending on the position. Furthermore, the electromagnetic wave sensing device of the present invention can estimate the moisture content of lumber or wood chips, as well as the concentration of solutes in a solution. In this case, it becomes possible to estimate the sugar content, salt concentration, or alcohol content of the aqueous solution, assuming that the solution is an aqueous solution. [Brief description of the drawings]

[0010] [Figure 1] 1 is a functional block diagram showing a configuration of an electromagnetic wave sensing device according to a first embodiment of the present invention. [Diagram 2] 1A and 1B are a front view and a right side view showing the configuration of an electromagnetic wave sensing unit of an electromagnetic wave sensing device according to a first embodiment of the present invention. [Diagram 3] 1 is a partially cutaway front view showing the configuration of an electromagnetic wave sensing unit of an electromagnetic wave sensing device according to a first embodiment of the present invention. [Figure 4] 1A and 1B are perspective views showing the configuration of an electromagnetic wave sensing unit of an electromagnetic wave sensing device according to a first embodiment of the present invention, and a perspective view showing the configuration of a radio wave absorber. [Diagram 5] FIG. 1 is a diagram showing a configuration for measuring a piece of wood as a sample by an electromagnetic wave sensing device according to a first embodiment of the present invention. [Figure 6] 1A to 1C are another diagram and further another diagram showing the configuration for measuring a piece of wood as a sample with the electromagnetic wave sensing device according to the first embodiment of the present invention. [Figure 7] 1 is a graph showing the moisture content versus data on the amount of change in the real part of a transmitted wave measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 8] 1 is a graph showing the moisture content versus data on the amount of change in the imaginary part of a transmitted wave measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 9] 1 is a graph showing the moisture content versus data on the amount of change in the real part of reflected wave A measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 10] 1 is a graph showing the moisture content versus data on the amount of change in the imaginary part of reflected wave A measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 11] 1 is a graph showing the moisture content versus data on the amount of change in the real part of reflected wave B measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 12]1 is a graph showing the moisture content versus data on the amount of change in the imaginary part of reflected wave B measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation determined from the data. [Figure 13] 1 is a graph showing the moisture content versus data on the average value of the change in the real part of the transmitted wave measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation obtained from the data. [Figure 14] 1 is a graph showing the moisture content versus data on the average value of the change in the imaginary part of a transmitted wave measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation calculated from the data. [Figure 15] 1 is a graph showing the moisture content versus data on the average value of the change in the real part of the reflected wave A measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and the regression equation obtained from the data. [Figure 16] 1 is a graph showing the moisture content versus data on the average value of the change in the imaginary part of reflected wave A measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation calculated from the data. [Figure 17] 1 is a graph showing the moisture content versus data on the average value of the change in the real part of the reflected wave B measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation calculated from the data. [Figure 18] 1 is a graph showing the moisture content versus data on the average value of the change in the imaginary part of reflected wave B measured by the electromagnetic wave sensing device of the first embodiment of the present invention, and a regression equation calculated from the data. [Figure 19] 4 is a graph showing the moisture content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the first embodiment of the present invention in comparison with the actual moisture content. [Figure 20] 4 is a graph showing an error in the moisture content calculated by a multiple regression equation obtained by the electromagnetic wave sensing device according to the first embodiment of the present invention. [Figure 21] 13 is a graph showing the water content versus average VSWR when the distance from the first antenna in the electromagnetic wave sensing device according to the second embodiment of the present invention is defined as a first distance. [Figure 22]13 is a graph showing the water content versus average VSWR when the distance from the second antenna in the electromagnetic wave sensing device according to the second embodiment of the present invention is set to a second distance. [Diagram 23] 1 is a graph showing the moisture content calculated by the multiple regression equation obtained in the electromagnetic wave sensing devices according to the first and second embodiments of the present invention, in comparison with the actual moisture content. [Figure 24] 11 is a graph showing errors in moisture content calculated by multiple regression equations obtained by the electromagnetic wave sensing devices according to the first and second embodiments of the present invention. [Diagram 25] 11 is a graph showing the amount of change in the real part of the transmitted wave versus frequency in an electromagnetic wave sensing device according to an embodiment of the present invention, with and without a radio wave absorber in the electromagnetic wave sensing section. [Figure 26] 11 is a graph showing the amount of change in the imaginary part of a transmitted wave versus frequency in an electromagnetic wave sensing device according to an embodiment of the present invention, with and without a radio wave absorber in the electromagnetic wave sensing portion. [Figure 27] FIG. 13 is a diagram showing a configuration for measuring wood chips used as samples by an electromagnetic wave sensing device according to a fourth embodiment of the present invention. [Figure 28] FIG. 13 is a diagram showing a configuration for measuring wood chips used as samples by an electromagnetic wave sensing device according to a fourth embodiment of the present invention. [Figure 29] 13 is a graph showing the moisture content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fourth embodiment of the present invention, in comparison with the actual moisture content. [Diagram 30] 13 is a graph showing an error in the moisture content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fourth embodiment of the present invention. [Diagram 31] 13 is a graph showing the moisture content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fourth embodiment of the present invention and by other multiple regression equations, in comparison with the actual moisture content. [Diagram 32] 13 is a graph showing errors in moisture content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fourth embodiment of the present invention and by other multiple regression equations. [Diagram 33]FIG. 13 is a diagram showing a configuration for measuring a solution serving as a sample by an electromagnetic wave sensing device according to a fifth embodiment of the present invention. [Diagram 34] FIG. 13 is a diagram showing a configuration for measuring a solution serving as a sample by an electromagnetic wave sensing device according to a fifth embodiment of the present invention. [Diagram 35] 13 is a graph showing sugar content versus data on the average amount of change in the real part of transmitted waves measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Diagram 36] 13 is a graph showing sugar content versus data on the amount of change in the imaginary part of a transmitted wave measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Figure 37] 13 is a graph showing sugar content versus data on the average value of the change in the real part of reflected wave A measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Figure 38] 13 is a graph showing sugar content versus data on the average value of the change in the imaginary part of reflected wave A measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Figure 39] 13 is a graph showing sugar content versus data on the average value of the change in the real part of reflected wave B measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Diagram 40] 13 is a graph showing sugar content versus data on the average value of the change in the imaginary part of reflected wave B measured by an electromagnetic wave sensing device according to a fifth embodiment of the present invention, and a regression equation calculated from the data. [Diagram 41] 13 is a graph showing sugar contents calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and other multiple regression equations, in comparison with the actual sugar contents. [Diagram 42] 13 is a graph showing errors in sugar content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device of the fifth embodiment of the present invention and by other multiple regression equations. [Diagram 43]13 is a graph showing sugar content versus average VSWR when the distance from the first antenna in the electromagnetic wave sensing device of the fifth embodiment of the present invention is set to a first distance. [Diagram 44] 13 is a graph showing sugar content versus average VSWR when the distance from the second antenna in the electromagnetic wave sensing device of the fifth embodiment of the present invention is set to a second distance. [Diagram 45] 13 is a graph showing sugar contents calculated by another multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and by yet another multiple regression equation, in comparison with the actual sugar contents. [Diagram 46] 13 is a graph showing errors in sugar content calculated by another multiple regression equation obtained by the electromagnetic wave sensing device of the fifth embodiment of the present invention and yet another multiple regression equation. [Figure 47] 13 is a graph showing salinity concentrations calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and other multiple regression equations, in comparison with actual salinity concentrations. [Figure 48] 13 is a graph showing errors in salinity concentration calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and by other multiple regression equations. [Figure 49] 13 is a graph showing salinity concentrations calculated by another multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and still another multiple regression equation, in comparison with the actual salinity concentrations. [Figure 50] 13 is a graph showing another multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and an error in the salinity concentration calculated by yet another multiple regression equation. [Figure 51] 13 is a graph showing the alcohol content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and by other multiple regression equations, in comparison with the actual alcohol content. [Figure 52] 13 is a graph showing errors in alcohol content calculated by the multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and by other multiple regression equations. [Diagram 53]13 is a graph showing alcohol content calculated by another multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and another multiple regression equation, in comparison with the actual alcohol content. [Figure 54] 13 is a graph showing errors in alcohol content calculated by another multiple regression equation obtained by the electromagnetic wave sensing device according to the fifth embodiment of the present invention and another multiple regression equation. [Figure 55] FIG. 1 is a functional block diagram showing the configuration of a conventional moisture content measuring device. [Figure 56] 1 is a graph showing the moisture content versus the slope determined using a conventional moisture measuring device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] <Electromagnetic wave sensing device according to a first embodiment of the present invention> The configuration of an electromagnetic wave sensing device 1 according to a first embodiment of the present invention is shown in Figures 1, 2(a) and 2(b), 3, and 4(a) and 4(b). Figure 1 is a functional block diagram showing the configuration of the electromagnetic wave sensing device 1 according to the first embodiment, Figure 2(a) is a front view showing the configuration of the electromagnetic wave sensing unit 10 of the electromagnetic wave sensing device 1 according to the first embodiment, Figure 2(b) is a right side view showing the configuration of the electromagnetic wave sensing unit 10 of the electromagnetic wave sensing device 1 according to the first embodiment, Figure 3 is a front view showing the configuration of the electromagnetic wave sensing unit 10 of the electromagnetic wave sensing device 1 according to the first embodiment with a part cut away, Figure 4(a) is a perspective view showing the configuration of the electromagnetic wave sensing unit 10 of the electromagnetic wave sensing device 1 according to the first embodiment, and Figure 4(b) is a perspective view showing the configuration of a radio wave absorber provided in the electromagnetic wave sensing unit 10 of the electromagnetic wave sensing device 1 according to the first embodiment.

[0012] The electromagnetic wave sensing device 1 of the first embodiment shown in these figures can quantitatively estimate the moisture content inside a sample such as lumber or wood chips. The electromagnetic wave sensing device 1 of the first embodiment is composed of an electromagnetic wave sensing unit 10 formed by processing a metal plate, a measurement unit 20 that measures the real and imaginary parts of the transmitted wave and the real and imaginary parts of the reflected wave from the electromagnetic wave sensing unit 10, and an analysis unit 30 that performs analysis to quantitatively estimate the moisture content of the sample based on the multiple pieces of data measured by the measurement unit 20. The measurement unit 20 can be composed of, for example, a network analyzer, and the analysis unit 30 equipped with a memory 31 can be composed of a personal computer (PC) on which software capable of performing at least regression analysis and multiple regression analysis is installed.

[0013] As shown in Figs. 2(a)(b), 3, and 4(a)(b), the electromagnetic wave sensing unit 10 includes a sensing housing 11 formed into a rectangular parallelepiped shape by processing a metal plate, and the inside of the sensing housing 11 is hollow. As shown in Fig. 3, a first antenna 11b is attached to the upper side of the hollow sensing housing 11, and a second antenna 11c is attached to the lower side so as to face the first antenna 11b. The first antenna 11b and the second antenna 11c are, for example, rectangular ring antennas, but may be antennas of other types. A radio wave absorber 11a that suppresses reflection of electromagnetic waves is attached to the inner wall surface of the sensing housing 11 except for the surface on which the antenna is provided. Instead of the radio wave absorber 11a, a radio wave absorbing sheet may be attached or radio wave absorbing paint may be applied.

[0014] A rectangular cylindrical feed section 12 is formed integrally with the right side surface of the sensing housing 11, and a send section 13 of the same shape as the rectangular cylindrical feed section 12 is formed integrally with the left side surface of the sensing housing 11. A sample such as rectangular column-shaped lumber or wood chips is fed into the feed section 12, and the sample is sent out from the send section 13 after passing through the sensing housing 11. The feed section 12 and the send section 13 are formed into a rectangular cylindrical shape by processing a metal plate. Inside the electromagnetic wave sensing unit 10, a radio wave absorber having a configuration shown in Fig. 4(b) is provided. That is, a radio wave absorber 12a having a rectangular cylindrical shape as shown in Fig. 4(b) is attached to the entire inner wall surface of the input unit 12, and a radio wave absorber 13a having a rectangular cylindrical shape as shown in Fig. 4(b) is attached to the entire inner wall surface of the output unit 13. Also, a radio wave absorber 11a having a rectangular ring shape as shown in Fig. 4(b) is attached to the inner wall surface of the front inner surface, the inner inner surfaces of both sides, and the inner inner surface of the back inner surface of the sensing housing 11, except for the part where the input unit 12 and the output unit 13 are provided, with a width between the first antenna 11b and the second antenna 11c. The end of the radio wave absorber 12a on the sensing housing 11 side is connected to the radio wave absorber 11a, and the end of the radio wave absorber 13a on the sensing housing 11 side is connected to the radio wave absorber 11a.

[0015] The first antenna 11b attached to the inner surface of the upper surface of the sensing housing 11 is connected to the first coaxial terminal 14a fixed to the upper surface, and the first coaxial terminal 14a is connected to the measurement unit 20. The second antenna 11c attached to the inner surface of the lower surface of the sensing housing 11 is connected to the second coaxial terminal 14b fixed to the lower surface, and the second coaxial terminal 14b is also connected to the measurement unit 20. When the electromagnetic wave sent from the measurement unit 20 is supplied to the first antenna 11b via the first coaxial terminal 14a, the electromagnetic wave is irradiated from the first antenna 11b. Then, the transmitted wave of the irradiated electromagnetic wave is received by the second antenna 11c, and the received transmitted wave is input to the measurement unit 20 via the second coaxial terminal 14b. Also, the reflected wave of the electromagnetic wave irradiated from the first antenna 11b is received by the first antenna 11b, and the received reflected wave is input to the measurement unit 20 via the first coaxial terminal 14a. Conversely, when the electromagnetic wave sent from the measurement unit 20 is supplied to the second antenna 11c via the second coaxial terminal 14b, the electromagnetic wave is irradiated from the second antenna 11c, and the transmitted wave of the irradiated electromagnetic wave is received by the first antenna 11b. The received transmitted wave is then input to the measurement unit 20 via the first coaxial terminal 14a. Furthermore, the reflected wave of the electromagnetic wave irradiated from the second antenna 11c is received by the second antenna 11c, and the received reflected wave is then input to the measurement unit 20 via the second coaxial terminal 14b.

[0016] <Measurement of a sample using the electromagnetic wave sensing device according to the first embodiment of the present invention> In the electromagnetic wave sensing device 1 of the first embodiment of the present invention, in order to quantitatively estimate the moisture content of a rectangular prism-shaped piece of lumber 100, which is a sample whose moisture content varies depending on the position, the electromagnetic wave sensing unit 10 irradiates the lumber 100 with an electromagnetic wave of any frequency within a frequency range of 100k to 50GHz, for example, 4200MHz, and measures the amplitude and phase of the transmitted wave and the reflected wave from the lumber 100. In this case, the amplitude and phase of the reflected wave and the transmitted wave can also be expressed as the real part and the imaginary part of the complex representation of the reflected wave and the transmitted wave, and here, data of the real part and the imaginary part of the reflected wave and the transmitted wave are measured. A configuration for measuring a rectangular columnar piece of lumber 100, which is a sample, in the electromagnetic wave sensing device 1 of the first embodiment of the present invention is shown in Figures 5, 6(a) and 6(b). Figure 5 is a perspective view of the electromagnetic wave sensing unit 10 in the configuration for measuring the lumber 100 with the electromagnetic wave sensing device 1 of the first embodiment, Figure 6(a) is a side view of the electromagnetic wave sensing unit 10 in the configuration for measuring the lumber 100 with the electromagnetic wave sensing device of the first embodiment, and Figure 6(b) is a cross-sectional view of the electromagnetic wave sensing unit 10 cut along line BB in the configuration for measuring lumber with the electromagnetic wave sensing device of the first embodiment. As shown in Figure 5, a rectangular columnar piece of lumber 100 to be used as a sample is fed through the feed-in section 12, and the lumber 100 is positioned within the sensing housing 11 as shown in Figure 6(b), while the tip side of the lumber 100 is fed out from the feed-out section 13.

[0017] Prior to measuring the lumber 100, initial settings are performed. In the initial settings, in a state where no sample is inserted in the electromagnetic wave sensing unit 10, an electromagnetic wave is supplied from the measuring unit 20 to the first antenna 11b via the first coaxial terminal 14a, and the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c. The transmitted wave received by the second antenna 11c is input to the measuring unit 20 via the second coaxial terminal 14b. The amplitude and phase data of the transmitted wave input to the measuring unit 20 are converted into real part data and imaginary part data expressed in complex numbers and supplied to the analyzing unit 30. In the analyzing unit 30, the real part data (hereinafter referred to as "Re12 base") and imaginary part data (hereinafter referred to as "Im12 base") of the transmitted wave are stored in the memory 31. The frequency of the electromagnetic wave is, for example, 4200 MHz.

[0018] Next, an electromagnetic wave is supplied from the measuring unit 20 to the first antenna 11b via the first coaxial terminal 14a, and the reflected wave of the electromagnetic wave irradiated from the first antenna 11b is received by the first antenna 11b. The reflected wave received by the first antenna 11b is input to the measuring unit 20 via the first coaxial terminal 14a. The amplitude and phase data of the reflected wave input to the measuring unit 20 are converted into real part data and imaginary part data that express the reflected wave in complex numbers, and are supplied to the analyzing unit 30. In the analyzing unit 30, the real part data (hereinafter referred to as "Re11 base") and imaginary part data (hereinafter referred to as "Im11 base") of the reflected wave are stored in the memory 31. The frequency of the electromagnetic wave is, for example, 4200 MHz.

[0019] Furthermore, the measurement unit 20 supplies electromagnetic waves to the second antenna 11c via the second coaxial terminal 14b, and the reflected waves of the electromagnetic waves irradiated from the second antenna 11c are received by the second antenna 11c. The reflected waves received by the second antenna 11c are input to the measurement unit 20 via the second coaxial terminal 14b. The amplitude and phase data of the reflected waves input to the measurement unit 20 are converted into real part data and imaginary part data that represent the reflected waves in complex numbers, and are supplied to the analysis unit 30. In the analysis unit 30, the real part data (hereinafter referred to as "Re22 base") and imaginary part data (hereinafter referred to as "Im22 base") of the reflected waves are stored in the memory 31. The frequency of the electromagnetic waves is, for example, 4200 MHz.

[0020] Here, a plurality of pieces of lumber 100 with a moisture content known in advance, for example, in the range of 2% to 110%, are prepared. After the initial setting is completed, a rectangular prism-shaped piece of lumber 100 with a predetermined moisture content is fed from the feed unit 12, and every time the piece of lumber 100 moves a predetermined distance, an electromagnetic wave is supplied from the measurement unit 20 to the first antenna 11b via the first coaxial terminal 14a, and the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c. The transmitted wave received by the second antenna 11c is input to the measurement unit 20 via the second coaxial terminal 14b. In the measurement unit 20, the amplitude and phase data of the input transmitted wave are converted into real part data and imaginary part data of the transmitted wave expressed in complex numbers, and are supplied to the analysis unit 30. In the analysis unit 30, the change in the real part (hereinafter referred to as “Re12”), which is the change from the Re12 base, and the change in the imaginary part (hereinafter referred to as “Im12”), which is the change from the Im12 base, are stored in the memory 31. In this case, Re12 is the change in the real part obtained by subtracting the Re12 base stored in memory 31 from the real part data of the transmitted wave, and Im12 is the change in the imaginary part obtained by subtracting the Im12 base stored in memory 31 from the imaginary part data of the transmitted wave. The frequency of the electromagnetic wave is, for example, 4200 MHz.

[0021] In addition, every time the lumber 100 moves a predetermined distance, an electromagnetic wave is supplied from the measuring unit 20 to the first antenna 11b via the first coaxial terminal 14a, and the reflected wave of the electromagnetic wave irradiated from the first antenna 11b is received by the first antenna 11b. The reflected wave received by the first antenna 11b is input to the measuring unit 20 via the first coaxial terminal 14a. In this case, the reflected wave input to the measuring unit 20 is treated as a reflected wave A in which the electromagnetic wave irradiated from the first antenna 11b is reflected by the lumber 100 in the electromagnetic wave sensing unit 10, and data on the amplitude and phase of this reflected wave A is treated as real part data and imaginary part data that represent the reflected wave A in complex numbers and is supplied to the analyzing unit 30. In the analyzing unit 30, the change amount of the real part (hereinafter referred to as "Re11"), which is the change amount from the Re11 base, and the change amount of the imaginary part (hereinafter referred to as "Im11"), which is the change amount from the Im11 base, are stored in the memory 31. In this case, Re11 is the amount of change in the real part obtained by subtracting the Re11 base stored in memory 31 from the real part data of the reflected wave A, and Im11 is the amount of change in the imaginary part obtained by subtracting the Im11 base stored in memory 31 from the imaginary part data of the reflected wave A. The frequency of the electromagnetic wave is, for example, 4200 MHz.

[0022] Furthermore, every time the lumber 100 moves a predetermined distance, the measurement unit 20 supplies electromagnetic waves to the second antenna 11c via the second coaxial terminal 14b, and the electromagnetic waves irradiated from the second antenna 11c and the reflected waves are received by the second antenna 11c. The reflected waves received by the second antenna 11c are input to the measurement unit 20 via the second coaxial terminal 14b. In this case, the reflected waves input to the measurement unit 20 are treated as reflected waves B, which are electromagnetic waves irradiated from the second antenna 11c and reflected by the lumber 100 in the electromagnetic wave sensing unit 10, and data on the amplitude and phase of the reflected waves B are treated as real part data and imaginary part data that represent the reflected waves B in complex numbers and are supplied to the analysis unit 30. In the analysis unit 30, the change in the real part (hereinafter referred to as "Re22"), which is the change in the Re22 base, and the change in the imaginary part (hereinafter referred to as "Im22"), which is the change in the Im22 base, are stored in the memory 31. In this case, Re22 is the amount of change in the real part obtained by subtracting the Re22 base stored in memory 31 from the real part data of the reflected wave B, and Im22 is the amount of change in the imaginary part obtained by subtracting the Im22 base stored in memory 31 from the imaginary part data of the reflected wave B. The frequency of the electromagnetic wave is, for example, 4200 MHz.

[0023] By carrying out the above-mentioned operation, data of Re12, Im12, Re11, Im11, Re22, and Im22 at each predetermined interval of the lumber 100 having a predetermined moisture content is obtained. In this case, the number of predetermined intervals is, for example, 20, and 20 sets of data each consisting of Re12, Im12, Re11, Im11, Re22, and Im22 are obtained, which is the same number as the number of predetermined intervals. That is, the 20 sets of data are data each consisting of Re12, Im12, Re11, Im11, Re22, and Im22 obtained at each of the 20 predetermined intervals. Next, by carrying out the above-mentioned operation on other lumber 100 having a different moisture content, 20 sets of data each consisting of Re12, Im12, Re11, Im11, Re22, and Im22 of the other lumber 100 having a different moisture content are obtained. The 20 sets of data are data that are groups of Re12, Im12, Re11, Im11, Re22, and Im22 for lumber 100 with different moisture contents. By repeating this, 20 sets of data that are groups of Re12, Im12, Re11, Im11, Re22, and Im22 for lumber 100 with multiple types of moisture contents are stored in memory 31. The 20 sets of data are data that are groups of Re12, Im12, Re11, Im11, Re22, and Im22 acquired at 20 predetermined intervals for lumber 100 with multiple types of moisture contents. Here, Re12, Im12, Re11, Im11, Re22, and Im22 are data of the amount of change, which is the amount of change from the initial setting data measured in a state where no sample is inserted into the electromagnetic wave sensing unit 10. Since the above-mentioned measurements were repeated twice for the same piece of lumber 100, 20×2=40 sets of data were measured for each piece of lumber 100. If the number of pieces of lumber 100 with the multiple types of moisture content prepared was, for example, 12 pieces, 40 sets of data were measured for each piece of lumber 100, resulting in 40×12=480 sets of data being measured.

[0024] FIG. 7 shows a graph plotting data of the change in the real part (Re12) of the transmitted wave measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention and stored in the memory 31 of the analysis unit 30 as described above. In FIG. 7, the horizontal axis is the change in the real part (Re12), and the vertical axis is the moisture content [%]. The graph in FIG. 7 is a graph in the case where 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 7, FIG. 7 shows 40×12=480 Re12 values ​​measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with moisture contents in the range of 2% to 110%. The Re12 values ​​of 40 for each piece of lumber 100 are plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the Re12 values ​​of 40 vary within a predetermined range on the line. This is because the moisture content varies depending on the position in the lumber 100. Here, the regression equation with one explanatory variable x for one objective variable y is as follows: y=b1x+b0 where b0 and b1 are partial regression coefficients. If the objective variable y is the moisture content and the explanatory variable x is Re12, and the partial regression coefficients b0 and b1 are found by the analysis unit 30 using the least squares method based on Re12 of 480, the regression equation of the following equation (1) is obtained. The least squares method is a well-known method for finding the most likely relationship equation by minimizing the sum of the squares of the errors in processing measured values ​​that contain errors. y=-14.616x+59.699 (1) In addition, the coefficient of determination of this regression equation, R 2 teeth, R 2 =0.0149 (2) The coefficient of determination represents the degree to which the dependent variable can be explained by the explanatory variables.

[0025] Next, FIG. 8 shows a graph plotting data of the change amount (Im12) of the imaginary part of the transmitted wave measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention as described above and stored in the memory 31 of the analysis unit 30. In the graph of FIG. 8, the horizontal axis is the change amount (Im12) of the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 8 is a graph in the case where 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared. Referring to FIG. 8, FIG. 8 shows 40×12=480 Im12 values ​​measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with a moisture content in the range of 2% to 110%. The 40 Im12 values ​​for each piece of lumber 100 are plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the 40 Im12 values ​​vary within a predetermined range on the line. One of the reasons is as described above. Here, when the objective variable y is the moisture content and the explanatory variable x is Im12, the regression equation (3) shown below is obtained by performing regression analysis in the analysis unit 30. y=34.94x+52.174 (3) The partial regression coefficient of the regression equation (3) is calculated by the analysis unit 30 using the well-known least squares method based on Im12 of the above 480. The coefficient of determination R 2 teeth, R 2 =0.1809 (4) It becomes.

[0026] Furthermore, FIG. 9 shows a graph plotting data of the change amount (Re11) of the real part of the reflected wave A measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention as described above and stored in the memory 31 of the analysis unit 30. In the graph of FIG. 9, the horizontal axis is the change amount (Re11) of the real part, and the vertical axis is the moisture content [%]. The graph of FIG. 9 is a graph in the case where 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared. Referring to FIG. 9, FIG. 9 shows 40×12=480 Re11s measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with a moisture content in the range of 2% to 110%. The Re11 of 40 for each piece of lumber 100 is plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the Re11 of 40 varies within a predetermined range on the line. One of the reasons for this is as described above. Here, when a regression analysis is performed in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being Re11, the following regression equation (5) is obtained. y=38.924x+44.709 (5) The partial regression coefficient of the regression equation (5) is calculated by the analysis unit 30 using the well-known least squares method based on Re11 of 480. The coefficient of determination R 2 teeth, R 2 =0.0302 (6) It becomes.

[0027] Furthermore, FIG. 10 shows a graph plotting data of the change amount (Im11) of the imaginary part of the reflected wave A measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention as described above and stored in the memory 31 of the analysis unit 30. In the graph of FIG. 10, the horizontal axis is the change amount (Im11) of the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 10 is a graph in the case where 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared. Referring to FIG. 10, FIG. 10 shows 40×12=480 Im11s measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with a moisture content in the range of 2% to 110%. The 40 Im11s for each piece of lumber 100 are plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the 40 Im11s vary within a predetermined range on the line. One of the reasons is as described above. Here, when the objective variable y is the moisture content and the explanatory variable x is Im11, the regression equation (7) shown below is obtained by performing regression analysis in the analysis unit 30. y=47.126x+61.31 (7) The partial regression coefficient of the regression equation (7) is calculated by the analysis unit 30 using the well-known least squares method based on Im11 of 480. The coefficient of determination R 2 teeth, R 2 =0.1153 (8) It becomes.

[0028] Furthermore, FIG. 11 shows a graph plotting data of the change amount (Re22) of the real part of the reflected wave B measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention as described above and stored in the memory 31 of the analysis unit 30. In the graph of FIG. 11, the horizontal axis is the change amount (Re22) of the real part, and the vertical axis is the moisture content [%]. The graph of FIG. 11 is a graph in the case where 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared. Referring to FIG. 11, FIG. 11 shows 40×12=480 Re22s measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with a moisture content in the range of 2% to 110%. The Re22s of 40 for each piece of lumber 100 are plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the Re22s of 40 vary within a predetermined range on the line. One of the reasons is as described above. Here, when the objective variable y is the moisture content and the explanatory variable x is Re22, the regression equation (9) shown below is obtained by performing regression analysis in the analysis unit 30. y=76.867x+29.937 (9) The partial regression coefficient of the regression equation (9) is calculated by the analysis unit 30 using the well-known least squares method based on Re22 of 480. The coefficient of determination R 2 teeth, R 2 =0.1938 (10) It becomes.

[0029] Furthermore, FIG. 12 shows a graph plotting data of the change amount (Im22) of the imaginary part of the reflected wave B measured in the electromagnetic wave sensing device 1 of the first embodiment of the present invention as described above and stored in the memory 31 of the analysis unit 30. In the graph of FIG. 12, the horizontal axis is the change amount (Im22) of the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 12 is a graph in the case where 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared. Referring to FIG. 12, FIG. 12 shows 40×12=480 Im22 values ​​measured twice at 20 predetermined intervals for each of the 12 pieces of lumber 100 with a moisture content in the range of 2% to 110%. The 40 Im22 values ​​for each piece of lumber 100 are plotted on a moisture content line corresponding to the moisture content of the measured lumber 100, and it can be seen that the 40 Im22 values ​​vary within a predetermined range on the line. One of the reasons is as described above. Here, when the objective variable y is the moisture content and the explanatory variable x is Im22, the regression equation (11) shown below is obtained by performing regression analysis in the analysis unit 30. y=68.073x+59.221 (11) The partial regression coefficient of the regression equation (11) is calculated by the analysis unit 30 using the well-known least squares method based on Im22 of 480. The coefficient of determination R 2 teeth, R 2 =0.2707 (12) It becomes.

[0030] Specifically, the memory 31 of the electromagnetic wave sensing device 1 of the first embodiment has a folder for each piece of lumber 100 with a predetermined moisture content. If lumber 100 with 12 different moisture contents is prepared, and each piece of lumber is called lumber 100a, lumber 100b, ..., lumber 100l, 12 folders for lumber 100a to lumber 100l are provided. In each folder for lumber 100a to lumber 100l, folders for intervals equal in number to the number of predetermined intervals are provided. If the number of predetermined intervals is 20 as described above, 20 folders for intervals 1 to 20 are provided. In each of the 20 folders for intervals 1 to 20, first and second folders are provided. The moisture contents of the lumber 100a to lumber 100l are, for example, about 2.1%, about 12.9%, about 23.0%, about 32.3%, about 41.0%, about 50.1%, about 59.9%, about 69.3%, about 80.1%, about 90.2%, about 102.3%, and about 110.5%. Furthermore, the memory 31 is provided with a measurement data folder, which will be described later, and the measurement data folder is provided with 20 interval folders, interval 1 to interval 20, and each of the 20 folders, interval 1 to interval 20, is provided with a first and second folder. The first and second folders are provided because, as described above, the same measurement is performed twice without changing the lumber 100, and the first folder stores data measured at the first measurement, and the second folder stores data measured at the second measurement. The memory 31 also has an initial settings folder, and the initial settings folder stores data measured in the above-mentioned initial settings.

[0031] Therefore, for example, data of Re12, Im12, Re11, Im11, Re22, and Im22 are measured in sequence every time the lumber 100a with a predetermined moisture content moves a predetermined distance, and are stored in the folder of the interval corresponding to the measurement position in the folder of the lumber 100a in the memory 31. As a result, the first folder of the ... Therefore, since the memory 31 has a first folder for each of the folders at intervals 1 to 20, a set of data Re12, Im12, Re11, Im11, Re22, and Im22 is stored in each of the 20 first folders, and a set of data Re12, Im12, Re11, Im11, Re22, and Im22 is stored in each of the 20 second folders. In this way, a total of 40 sets of data Re12, Im12, Re11, Im11, Re22, and Im22 are stored in the folder of lumber 100a in the memory 31 in the first and second folders.

[0032] Similar to the above-mentioned lumber 100a, similar measurements are performed on the lumber 100b to lumber 100l, and so 40 sets of data Re12, Im12, Re11, Im11, Re22, and Im22 are stored in the folders for lumber 100b to lumber 100l in memory 31, just like the folder for lumber 100a. In this way, when 12 pieces of lumber 100 with different moisture contents are prepared, 40 sets are stored in each of the folders for lumber 100a to lumber 100l, so a total of 40×12=480 sets of data are stored in memory 31.

[0033] <Estimation of Water Content of Sample in Electromagnetic Wave Sensing Device of First Embodiment of the Present Invention> As described above, in the electromagnetic wave sensing device 1 of the first embodiment, when twelve pieces of lumber 100 with moisture content in the range of 2% to 110% are prepared, 40 sets of data of Re12, Im12, Re11, Im11, Re22, and Im22 at predetermined intervals are stored in each folder of lumber 100a to lumber 100l as described above. That is, when lumber 100a to lumber 100l are prepared as lumber 100 with a predetermined moisture content, the data stored in the memory 31 is 40×12=480 sets. Here, Re12, Im12, Re11, Im11, Re22, and Im22 are data of the amount of change, which is the amount of change from the initial setting data measured in a state where no sample is inserted in the electromagnetic wave sensing unit 10. Then, by performing multiple regression analysis in the analysis unit 30 using the moisture content as the objective variable and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is shown in equation (13). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6(13) In equation (13), b0, b1, b2, b3, b4, b5, and b6 are partial regression coefficients, x1 is the value of Re12, x2 is the value of Im12, x3 is the value of Re11, x4 is the value of Im11, x5 is the value of Re22, and x6 is the value of Im22. When the partial regression coefficients of the multiple regression equation in equation (13) are calculated by analysis unit 30 using the well-known least squares method based on the above 480 sets of data, the multiple regression equation shown in equation (14) is obtained. y=51.35+2.67x1+26.20x2-225.87x3+23.67x4+216.40x5+79.91x6(14) The y in equation (14) is the multiple regression equation for the calibration curve calculated from the change data.

[0034] When estimating the moisture content of lumber 100, the prismatic lumber 100, the moisture content of which is to be determined, is fed through the feed section 12, and an electromagnetic wave of, for example, 4200 MHz is supplied from the measurement section 20 to the first antenna 11b via the first coaxial terminal 14a each time the lumber 100 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave of the lumber 100, the moisture content of which is to be determined, are stored in the memory 31. Furthermore, the first antenna 11b receives the reflected wave of, for example, a 4200 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the lumber 100 whose moisture content is to be found. Furthermore, the second antenna 11c receives the reflected wave of, for example, a 4200 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the lumber 100 whose moisture content is to be found. Re12', Im12', Re11', Im11', Re22', and Im22' are data measured every time the sensor moves a predetermined distance. In this case, the number of predetermined distances is 20, as above, and 20 sets of data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined distances. In other words, the 20 sets of data are data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' obtained every 20 predetermined distances. Since the same measurement is repeated twice, 40 sets of data for the lumber 100 for which the moisture content is to be calculated are stored in the memory 31. Here, Re12', Im12', Re11', Im11', Re22', and Im22' are data on the amount of change, which is the amount of change from the initial setting data measured in a state where no sample is inserted in the electromagnetic wave sensing unit 10.

[0035] Specifically, in the electromagnetic wave sensing device 1 of the first embodiment, as described above, a measurement data folder is provided in the memory 31, and folders for intervals 1 to 20 are provided in the measurement data folder. Also, since the same measurement is performed twice without changing the lumber 100 as described above, a first and second folder is provided in each folder for intervals 1 to 20. Then, when data Re12', Im12', Re11', Im11', Re22', and Im22' are measured in sequence every time the lumber 100, whose moisture content is to be determined, moves a predetermined interval, the data is stored in a first folder in the folder for the interval corresponding to the measurement position of the lumber 100 in the measurement data folder in the memory 31. As a result, the first folder of the measurement data folder interval 1 stores the set of data Re12', Im12', Re11', Im11', Re22', and Im22' measured the first time, and the second folder of the measurement data folder interval 1 stores the set of data measured the second time. Similarly, the first and second folders of the measurement data folder interval 2 to interval 20 store the set of data Re12', Im12', Re11', Im11', Re22', and Im22' measured the respective times. In the memory 31, the first folders in the measurement data folder are provided for each of the folders of interval 1 to interval 20, so that the set of data Re12', Im12', Re11', Im11', Re22', and Im22' is stored in each of the 20 first folders, and the second folders are provided for each of the folders of interval 1 to interval 20, so that the set of data Re12', Im12', Re11', Im11', Re22', and Im22' is stored in each of the 20 second folders. In this way, a total of 40 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' measured on the lumber 100 for which the moisture content is to be calculated are stored in the measurement data folders of the memory 31 in the first and second folders of the measurement data folders.

[0036] The moisture content of the lumber 100 whose moisture content is to be determined was estimated by substituting the above-mentioned 40 sets of data, Re12', Im12', Re11', Im11', Re22', and Im22', obtained each time the lumber 100 whose moisture content is to be determined moved a predetermined distance, into the above-mentioned formula (14), which is the formula of the calibration curve. The moisture content estimated by substituting the 40 sets of data is 40. The average value of the 40 moisture contents estimated as the estimated moisture content of the lumber 100 whose moisture content is to be determined can be calculated and presented by the analysis unit 30. Alternatively, the range of the average value of the 40 moisture contents ± the standard deviation may be presented. Here, a graph shown in FIG. 19 is created so that the variation in the moisture content of 40 estimated by the calibration curve of the multiple regression equation shown in Equation (14) can be visualized. The graph shown in FIG. 19 plots the moisture content of 40 with the moisture content [%] of the calibration curve calculated on the horizontal axis and the moisture content [%] of the actual measured value of the lumber 100 for which the moisture content is to be calculated on the vertical axis. In FIG. 19, the estimated moisture content of 40 is shown as a gray rectangle that is raw data. Referring to FIG. 19, the moisture content of 40 calculated by the calibration curve is shown for each of 12 points of the actual measured moisture content of about 2.1%, about 12.9%, about 23.0%, about 32.3%, about 41.0%, about 50.1%, about 59.9%, about 69.3%, about 80.1%, about 90.2%, about 102.3%, and about 110.5%. This is because the 12 pieces of lumber 100 with the moisture contents at the above 12 points are the pieces of lumber 100 whose moisture contents are to be calculated, and the moisture contents for the calibration curve are calculated from the measurement data. Here, when the actual moisture content is set as the objective variable ya and the moisture content x for the calibration curve calculation is used as the explanatory variable to perform regression analysis in the analysis unit 30, the following regression equation (15) is obtained. ya=x-4×10 -13 (15) ya is shown in FIG. 19 as ya (raw data) by a solid gray line, and the partial regression coefficient of the regression equation (15) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 40×12=480. The coefficient of determination R 2 teeth, R 2 =0.8788 (16) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the moisture content of the lumber 100 for which the moisture content is to be calculated. In addition, a graph in which the error between the moisture content of 40 calculated from the calibration curve and the measured moisture content is plotted on the vertical axis and the measured moisture content is plotted on the horizontal axis is shown in Figure 20. In Figure 20, the error is shown as a gray rectangle that is treated as raw data. 19 and 20, it is apparent that the calibration curve shown in the above formula (14) is highly reliable.

[0037] <Measurement of a sample using an electromagnetic wave sensing device according to a second embodiment of the present invention> Although not shown, the electromagnetic wave sensing device 2 of the second embodiment of the present invention has a similar configuration to the electromagnetic wave sensing device 1 of the first embodiment of the present invention. The following describes the configuration in which the electromagnetic wave sensing device 2 of the second embodiment of the present invention differs from the electromagnetic wave sensing device 1 of the first embodiment of the present invention. In the electromagnetic wave sensing device 1 of the first embodiment of the present invention, when 12 pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared, data on the amount of change at predetermined intervals for the 12 pieces of lumber 100, Re12, Im12, Re11, Im11, Re22, and Im22, are stored in memory 31. In contrast, in the electromagnetic wave sensing device 2 of the second embodiment of the present invention, when twelve pieces of wood 100 with a moisture content in the range of 2% to 110% are prepared, the average value of the change data for each specified interval of the twelve pieces of wood 100 is calculated, and the average values ​​of the calculated change data, Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, are stored in memory 31.

[0038] Specifically, in the electromagnetic wave sensing device 2 of the second embodiment, a folder for average values ​​is further provided in each folder for the above-mentioned pieces of wood 100a to 100l in the memory 31, and the average value folder is further provided with folders for the first and second measurements. Also, a folder for average values ​​is further provided in the measurement data folder, and the average value folder is further provided with folders for the first and second measurements. Then, for example, data on the amount of change in Re12, Im12, Re11, Im11, Re22, and Im22 is measured in sequence each time the lumber 100a with a predetermined moisture content moves a predetermined distance, and stored in the folder of the measurement position in the folder of the lumber 100a in the memory 31 in the same manner as in the electromagnetic wave sensing device 1 of the first embodiment. As a result, the first folder in the folder of the folder of the interval 1 stores data on the amount of change in the set of Re12, Im12, Re11, Im11, Re22, and Im22 measured in the first time, and the second folder in the folder of the folder of the interval 1 stores data on the amount of change in the set of Re12, Im12, Re11, Im11, Re22, and Im22 measured in the second time. Similarly, the first and second folders in the folders of the folders of the intervals 2 to 20 store data on the amount of change in the set of Re12, Im12, Re11, Im11, Re22, and Im22 measured in each time. In this way, a total of 40 sets of change amount data of Re12, Im12, Re11, Im11, Re22, and Im22 are stored in the folder of the lumber 100a in the memory 31 in the first folder and the second folder.

[0039] Similar to the above-described lumber 100a, similar measurements are performed on the lumber 100b to lumber 100l, and so, like the folder for lumber 100a, 40 sets of change data, Re12, Im12, Re11, Im11, Re22, and Im22, are stored in the memory 31 for the folders for lumber 100b to lumber 100l. In this way, when 12 pieces of lumber 100 with mutually different moisture contents are prepared, 40 sets are stored in each of the folders for lumber 100a to lumber 100l, and so a total of 40×12=480 sets of data are stored in the memory 31. Here, in the electromagnetic wave sensing device 2 of the second embodiment, unlike the electromagnetic wave sensing device 1 of the first embodiment, the analysis unit 30 calculates the average value for each type of 20 sets of change amount data, Re12, Im12, Re11, Im11, Re22, and Im22, stored in the first folder of the folders of interval 1 to interval 20. For example, data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are average values ​​of the first change amount data of the change amount at each predetermined interval of the lumber 100a calculated by the analysis unit 30, are stored in the first folder of the average value folder in the folder of the lumber 100a, and data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are average values ​​of the second change amount data of the change amount at each predetermined interval of the lumber 100a, are stored in the second folder of the average value folder in the folder of the lumber 100a.

[0040] Then, as in the case of the above-mentioned lumber 100a, the average values ​​of the lumber 100b to lumber 100l are calculated by the analysis unit 30 in the same manner, so that in the average value folders of the lumber 100b to lumber 100l, the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in the first and second folders in the average value folders in the same manner as in the average value folder of the lumber 100a. Therefore, in the average value folders of the lumber 100a to lumber 100l, two sets of average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored, so that 2×12=24 sets of average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in the memory 31. Except for this configuration, the configuration of the electromagnetic wave sensing device 2 of the second embodiment of the present invention is similar to the configuration of the electromagnetic wave sensing device 1 of the first embodiment of the present invention. Therefore, in the description of the electromagnetic wave sensing device 2 of the second embodiment of the present invention, only the different configurations described above will be described, and other descriptions will be omitted.

[0041] In the electromagnetic wave sensing device 2 of the second embodiment of the present invention, the average value data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A of Re12, Im12, Re11, Im11, Re22, and Im22, which are the data of the change amount at each predetermined interval measured in the electromagnetic wave sensing device 1 of the first embodiment, are calculated and stored in the first and second folders of the average value folder in the memory 31. FIG. 13 shows a graph plotting data of the average value (Re12A) of the change in the real part of the transmitted wave stored in the average value folder of the memory 31 of the analysis unit 30. In the graph of FIG. 13, the horizontal axis is the average value (Re12A) of the change in the real part, and the vertical axis is the moisture content [%]. The graph of FIG. 13 is a graph in the case where 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 13, the Re12A calculated from the 20 Re12s measured at 20 predetermined intervals in the first measurement and the Re12A calculated from the 20 Re12s measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Re12A) of the change in the real part of the transmitted wave are calculated for each piece of lumber 100, the average value (Re12A) of the change in the real part of the transmitted wave calculated for the 12 pieces of lumber 100 is 24. The 24 Re12As are plotted on a line of moisture content according to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The Re12As displayed in pairs are the first and second Re12As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when the objective variable y is the moisture content and the explanatory variable x is the Re12A, the regression equation (17) shown below is obtained by performing regression analysis in the analysis unit 30. y=17.643x+60.429 (17) The partial regression coefficient of the regression equation (17) is calculated by the analysis unit 30 using the well-known least squares method based on the Re12A data of the above 24. The coefficient of determination R 2 teeth, R 2 =0.018 (18) It becomes.

[0042] FIG. 14 shows a graph of the average value (Im12A) of the change in the imaginary part of the transmitted wave stored in the memory 31 of the analysis unit 30. In the graph of FIG. 14, the horizontal axis is the average value (Im12A) of the change in the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 14 is a graph in the case where 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 14, the Im12A calculated from the Im12 of 20 pieces measured at 20 predetermined intervals in the first measurement and the Im12A calculated from the Im12 of 20 pieces measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Im12A) of the change in the imaginary part of the transmitted wave are calculated for each piece of lumber 100, the average value (Im12A) of the change in the imaginary part of the transmitted wave calculated for the 12 pieces of lumber 100 is 24. These 24 Im12As are plotted on a line of moisture content according to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The Im12As displayed in pairs are the first and second Im12As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when a regression analysis is performed in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being Im12A, the following regression equation (19) is obtained. y=35.981x+52.054 (19) The partial regression coefficient of the regression equation (19) is calculated by the analysis unit 30 using the known least squares method based on the Im12A data of the above 24. The coefficient of determination R 2 teeth, R 2 =0.1863 (20) It becomes.

[0043] Furthermore, data on the average value (Re11A) of the change in the real part of the reflected wave A stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 15. In the graph of FIG. 15, the horizontal axis is the average value (Re11A) of the change in the real part, and the vertical axis is the moisture content [%]. The graph of FIG. 15 is a graph in the case where 12 pieces of lumber 100 having different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 15, the Re11A calculated from the 20 Re11s measured at 20 predetermined intervals in the first measurement and the Re11A calculated from the 20 Re11s measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 having different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Re11A) of the change in the real part of the reflected wave A are calculated for each piece of lumber 100, the average value (Re11A) of the change in the real part of the reflected wave A calculated for the 12 pieces of lumber 100 is 24. These 24 Re11As are plotted on a line of moisture content according to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The Re11As displayed in pairs are the first and second Re11As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when a regression analysis is performed in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being Re11A, the following regression equation (21) is obtained. y=39.468x+44.549 (21) The partial regression coefficient of the regression equation (21) is calculated by the analysis unit 30 using the well-known least squares method based on the Re11A data of 24. The coefficient of determination R 2 teeth, R 2 =0.0307 (22) It becomes.

[0044] Furthermore, data on the average value (Im11A) of the change in the imaginary part of the reflected wave A stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 16. In the graph of FIG. 16, the horizontal axis is the average value (Im11A) of the change in the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 16 is a graph in the case where 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 16, Im11A calculated from 20 Im11s measured at 20 predetermined intervals in the first measurement and Im11A calculated from 20 Im11s measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Im11A) of the change in the imaginary part of the reflected wave A are calculated for each piece of lumber 100, the average value (Im11A) of the change in the imaginary part of the reflected wave A calculated for the 12 pieces of lumber 100 is 24. These 24 Im11As are plotted on a line of moisture content corresponding to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The Im11As displayed in pairs are the first and second Im11As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when a regression analysis is performed in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being Im11A, the following regression equation (23) is obtained. y=47.942x+61.399 (23) The partial regression coefficient of the regression equation (23) is calculated by the analysis unit 30 using the known least squares method based on the data of Im11A in 24. The coefficient of determination R 2 teeth, R 2 =0.1173 (24) It becomes.

[0045] Furthermore, data of the average value (Re22A) of the change in the real part of the reflected wave B stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 17. In the graph of FIG. 17, the horizontal axis is the average value (Re22A) of the change in the real part, and the vertical axis is the moisture content [%]. The graph of FIG. 17 is a graph in the case where 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 17, the Re22A calculated from the Re22 of 20 pieces measured at 20 predetermined intervals in the first measurement and the Re22A calculated from the Re22 of 20 pieces measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 with different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Re22A) of the change in the real part of the reflected wave B are calculated for each piece of lumber 100, the average value (Re22A) of the change in the real part of the reflected wave B calculated for the 12 pieces of lumber 100 is 24. The 24 Re22As are plotted on a line of moisture content corresponding to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The two Re22As displayed are the first and second Re22As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when the objective variable y is the moisture content and the explanatory variable x is Re22A, the regression equation (25) shown below is obtained by performing regression analysis in the analysis unit 30. y=78.766x+29.289 (25) The partial regression coefficient of the regression equation (25) is calculated by the analysis unit 30 using the well-known least squares method based on the data of Re22A in 24. The coefficient of determination R 2 teeth, R 2 =0.1986 (26) It becomes.

[0046] Furthermore, data of the average value (Im22A) of the change in the imaginary part of the reflected wave B stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 18. In the graph of FIG. 18, the horizontal axis is the average value (Im22A) of the change in the imaginary part, and the vertical axis is the moisture content [%]. The graph of FIG. 18 is a graph in the case where 12 pieces of lumber 100 having different moisture contents in the range of 2% to 110% are prepared. Referring to FIG. 18, Im22A calculated from 20 Im22s measured at 20 predetermined intervals in the first measurement and Im22A calculated from 20 Im22s measured in the second measurement are shown by black circles for each of the 12 pieces of lumber 100 having different moisture contents in the range of 2% to 110%. In this case, since two average values ​​(Im22A) of the change in the imaginary part of the reflected wave B are calculated for each piece of lumber 100, the average value (Im22A) of the change in the imaginary part of the reflected wave B calculated for the 12 pieces of lumber 100 is 24. The 24 Im22As are plotted on a line of moisture content corresponding to the moisture content of the measured lumber 100, and are displayed in pairs at approximately the same positions on the line. The Im22As displayed in pairs are the first and second Im22As measured on the same piece of lumber 100, and it can be seen that they are uniform without any variation. Here, when a regression analysis is performed in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being Im22A, the following regression equation (27) is obtained. y=68.96x+59.261 (27) The partial regression coefficient of the regression equation (27) is calculated by the analysis unit 30 using the well-known least squares method based on the data of ReIm22A in 24. The coefficient of determination R 2 teeth, R 2 =0.2742 (28) It becomes.

[0047] <Estimation of Water Content of Sample in Electromagnetic Wave Sensing Device of Second Embodiment of the Present Invention> As described above, in the electromagnetic wave sensing device 2 of the second embodiment, when, for example, twelve pieces of wood 100 with moisture content in the range of 2% to 110% are prepared, data sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are average values ​​of the amount of change at predetermined intervals for the twelve pieces of wood 100, are stored in the first folder and the second folder in the average value folder of the folders for wood 100a to wood 100k in the memory 31. That is, a total of 24 sets of data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored. Here, the data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are the average values ​​of the data of the amount of change Re12, Im12, Re11, Im11, Re22, and Im22. Therefore, by performing multiple regression analysis in the analysis unit 30 with the moisture content as the objective variable and the six parameters of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A as the explanatory variables, the multiple regression equation shown in equation (29) of the calibration curve can be obtained. This multiple regression equation is the same as equation (13) described above. y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6(29) where b0, b1, b2, b3, b4, b5, and b6 are partial regression coefficients, x1 is the numerical value of Re12A, x2 is the numerical value of Im12A, x3 is the numerical value of Re11A, x4 is the numerical value of Im11A, x5 is the numerical value of Re22A, and x6 is the numerical value of Im22A. When the partial regression coefficients of the above multiple regression equation are calculated by analysis unit 30 using the well-known least squares method based on the above 24 sets of data, the multiple regression equation shown in equation (30) is obtained. y=56.42-2.53x1+26.26x2-256.26x3+2.11x4+229.85x5+105.30x6(30) The y in equation (30) is the multiple regression equation for the calibration curve calculated from the average value of the change amount data.

[0048] When it is desired to estimate the moisture content of lumber 100, similarly to the electromagnetic wave sensing device 1 of the first embodiment, the prismatic lumber 100, the moisture content of which is to be determined, is fed through the feeding section 12, and an electromagnetic wave of, for example, 4200 MHz is supplied from the measuring section 20 to the first antenna 11b via the first coaxial terminal 14a each time the lumber 100 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave of the lumber 100, the moisture content of which is to be determined, are stored in the memory 31. Furthermore, the first antenna 11b receives the reflected wave of, for example, a 4200 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the lumber 100 whose moisture content is to be found. Furthermore, the second antenna 11c receives the reflected wave of, for example, a 4200 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the lumber 100 whose moisture content is to be found. Re12', Im12', Re11', Im11', Re22', and Im22' are data measured every time the sensor moves a predetermined distance. In this case, the number of predetermined distances is 20, as above, and 20 sets of data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined distances. In other words, the 20 sets of data are data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' obtained every 20 predetermined distances. Since the same measurement is repeated twice, 40 sets of data for the lumber 100 for which the moisture content is to be calculated are stored in the memory 31. Here, Re12', Im12', Re11', Im11', Re22', and Im22' are data on the amount of change, which is the amount of change from the initial setting data measured in a state where no sample is inserted in the electromagnetic wave sensing unit 10.

[0049] Next, the analysis unit 30 calculates the average value of each of the 20 pieces of data Re12', Im12', Re11', Im11', Re22', and Im22' stored in the first folder of the folders for intervals 1 to 20 of the measurement data folder. The data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average values ​​of the data of the amount of change at each predetermined interval of the lumber 100 calculated by the analysis unit 30, are stored in the first folder of the folder for average values ​​in the measurement data folder. Next, the analysis unit 30 calculates the average value of each of the 20 pieces of data Re12', Im12', Re11', Im11', Re22', and Im22' stored in the second folder of the folders for intervals 1 to 20. The data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average values ​​of the data on the amount of change in lumber 100 at specified intervals calculated by analysis unit 30, are stored in the second folder of the average value folder in the measurement data folder.

[0050] The data of two sets of average values ​​of the average amounts of change, Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', obtained each time the wood 100, whose moisture content is to be determined, moves a predetermined distance, are substituted into the above formula (30), which is the formula for the calibration curve, to estimate the two moisture contents of the wood 100, whose moisture content is to be determined. Then, the analysis unit 30 can calculate and present the average value of the two estimated moisture contents as the estimated moisture contents of the wood 100, whose moisture content is to be determined. Here, a graph shown in FIG. 19 is created so that the variation of the two moisture contents estimated by the calibration curve of the multiple regression equation shown in Equation (30) can be visualized. The graph shown in FIG. 19 plots the two moisture contents with the moisture content [%] of the calibration curve calculated on the horizontal axis and the moisture content [%] of the actual measured value of the lumber 100 for which the moisture content is to be calculated on the vertical axis. In FIG. 19, the estimated moisture content is shown as a white circle that is the two averaged data. Referring to FIG. 19, the moisture contents of the two calibration curves are shown for each of 12 points of the actual measured moisture content of about 2.1%, about 12.9%, about 23.0%, about 32.3%, about 41.0%, about 50.1%, about 59.9%, about 69.3%, about 80.1%, about 90.2%, about 102.3%, and about 110.5%. This is because the case is shown where the 12 pieces of lumber 100 with the moisture contents at the above 12 points are the pieces of lumber 100 for which moisture contents are to be calculated, and the moisture contents for the calibration curve are calculated from the measurement data. Here, when the moisture content of the actual measurement value is used as the objective variable yb and the moisture content x for the calibration curve calculation is used as the explanatory variable to perform regression analysis in the analysis unit 30, the following regression equation (31) is obtained. yb=0.9825x+0.5847 (31) yb is shown in FIG. 19 as a straight dashed line, yb (averaged data), and the partial regression coefficient of the regression equation (31) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 2×12=24. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.9233 (32) This shows that the degree to which the objective variable can be explained by the explanatory variables is higher. In other words, it is possible to quantitatively estimate the moisture content of the lumber 100 for which the moisture content is to be calculated. In addition, a graph in which the error between the moisture content calculated by the calibration curve and the moisture content of the actual measured value is plotted on the vertical axis and the moisture content of the actual measured value on the horizontal axis is shown in Fig. 20. In Fig. 20, the error is shown as a white circle that is the average of two data. 19 and 20, it is apparent that the calibration curve shown in the above formula (30) is highly reliable.

[0051] Referring to the graph in FIG. 19, it can be seen that regression equation yb (31), which is created from the relationship between the moisture content and the average amount of change in the real and imaginary parts of the transmitted wave and reflected wave at regular intervals, has a higher coefficient of determination and is closer to the actual measured values ​​than regression equation ya (15), which is created from the relationship between the moisture content and the amount of change in the real and imaginary parts of the transmitted wave and reflected wave at regular intervals. Also, as can be seen by referring to the graph in Fig. 20, the calculated values ​​of the calibration curve of the multiple regression equation (30) created from the relationship between the average amount of change in the real and imaginary parts of the transmitted and reflected waves at regular intervals and the moisture content have less error than the calculated values ​​of the calibration curve of the multiple regression equation (14) created from the relationship between the amount of change in the real and imaginary parts of the transmitted and reflected waves at regular intervals and the moisture content. In other words, referring to Figs. 19 and 20, it can be seen that the calibration curve shown in the above formula (30) is more reliable than the calibration curve shown in the above formula (14). From the above, it can be said that the reliability of the calibration curve created from the relationship between the average changes in the real and imaginary parts of the transmitted and reflected waves at regular intervals and the moisture content is very high.

[0052] <Measurement of a sample using an electromagnetic wave sensing device according to a third embodiment of the present invention> Although not shown, the electromagnetic wave sensing device 3 of the third embodiment of the present invention has the same configuration as the electromagnetic wave sensing device 2 of the second embodiment of the present invention, and the measuring unit 20 has a function capable of measuring VSWR. The configuration of the electromagnetic wave sensing device 3 of the third embodiment of the present invention that differs from the electromagnetic wave sensing device 2 of the second embodiment of the present invention will be described. In the electromagnetic wave sensing device 2 of the second embodiment of the present invention, when eleven pieces of lumber 100 with a moisture content in the range of 2% to 110% are prepared, data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A that are average values ​​of data of the amount of change at predetermined intervals for the eleven pieces of lumber 100a to lumber 100k are stored in an average value folder in the folders of lumber 100a to lumber 100k in memory 31. In addition to this configuration, when measuring the data of the amount of change at each predetermined interval of the lumber 100, the measurement unit 20 measures the VSWR at each predetermined interval of the lumber 100 and stores the measured values ​​in folders of lumber 100a to lumber 100k in the memory 31. Next, the average value of the measured VSWR at each predetermined interval is calculated by the analysis unit 30. The calculated average value data of the VSWR is stored in the memory 31 together with the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A.

[0053] Specifically, in the electromagnetic wave sensing device 3 of the third embodiment, VSWR data is stored in addition to the data on the amount of change in each of the folders for intervals 1 to 20 of the folders for lumber 100a to lumber 100k in the memory 31, and average value VSWR data is stored in addition to the data on the average amount of change in the first and second folders of the average value folders of each of the folders for lumber 100a to lumber 100k. Also, VSWR data is stored in addition to the data on the amount of change in each of the folders for intervals 1 to 20 of the measurement data folder, and average value VSWR data is stored in addition to the data on the average amount of change in the first and second folders of the average value folders of each of the measurement data folders. The other configurations are the same as those of the electromagnetic wave sensing device 2 of the second embodiment of the present invention. Therefore, in the description of the electromagnetic wave sensing device 3 of the third embodiment of the present invention, only the different configurations will be described, and other descriptions will be omitted. Note that the measurement unit 20 in the electromagnetic wave sensing device 3 of the third embodiment of the present invention can be configured, for example, by a vector network analyzer capable of measuring VSWR. The frequency of the electromagnetic wave irradiated to the rectangular columnar lumber 100 that is the sample in the electromagnetic wave sensing device 3 is set to, for example, 4200 MHz.

[0054] In the electromagnetic wave sensing device 3 of the third embodiment of the present invention, data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are the average values ​​of the data of the amount of change at predetermined intervals of multiple pieces of wood 100 with a moisture content in the range of 2% to 110%, and the average value of the VSWR at predetermined intervals of the pieces of wood 100, are measured by the measurement unit 20 and analysis unit 30, and stored in the first and second folders of each average value folder in the measurement data folder in memory 31. When the data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A stored in the memory 31 of the analysis unit 30 are graphed, they are as shown in the above-mentioned Figures 13 to 18. In addition, in the configuration shown in Figure 6(b), when the distance between the first antenna 11b and the sample lumber 100 is set to, for example, 253.8 mm, the average VSWR data measured the first and second times from the first antenna 11b side is graphed, as shown in Figure 21. In the graph of Figure 21, the horizontal axis is the average VSWR (VSWR1), and the vertical axis is the moisture content [%]. The graph of Figure 21 is a graph in the case where 12 pieces of lumber 100 with moisture contents in the range of 2% to 110% are prepared. 21, VSWR1 is shown as data on the average value of VSWR measured for 12 pieces of lumber 100 with moisture content in the range of 2% to 110%, and two VSWR1s measured the first and second time are displayed in approximately the same position. Since 12 pieces of lumber 100 are prepared, the total number of VSWR1s displayed in FIG. 21 is 24. Then, by performing regression analysis in analysis unit 30 with moisture content as objective variable y and VSWR1 as explanatory variable x, the following regression equation (33) is obtained. y=-76149x+79557 (33) The partial regression coefficient of the regression equation (33) is calculated by the analysis unit 30 using the well-known least squares method based on the VSWR1 data of the above 24. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.9832 (34) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is high. One reason for this is that the distance (253.8 mm) at which the correlation between the distance between the first antenna 11b and the lumber 100 and the moisture content of the lumber 100 is high is selected. In this case, the distance at which the correlation is high is measured in advance.

[0055] The VSWR can also be measured from the second antenna 11c side. In this case, the distance at which the correlation between the distance between the second antenna 11c and the lumber 100 and the moisture content of the lumber 100 becomes high is measured to be 747 mm. In the configuration shown in FIG. 6(b), the distance between the second antenna 11c and the sample lumber 100 is set to, for example, 747 mm, and the average VSWR data measured the first and second times from the second antenna 11c side is graphed to obtain the graph shown in FIG. 22. In the graph of FIG. 22, the horizontal axis is the average VSWR (VSWR2), and the vertical axis is the moisture content [%]. The graph of FIG. 22 is a graph in the case where 12 pieces of lumber 100 with moisture contents ranging from 2% to 110% are prepared. Referring to Fig. 22, the VSWR2 of 12 pieces of lumber 100 with different moisture contents ranging from 2% to 110% is shown, and the two VSWR2s measured the first and second time are displayed in approximately the same position. Since 12 pieces of lumber 100 are prepared, the total number of VSWR2s displayed in Fig. 22 is 24. Then, by performing a regression analysis in the analysis unit 30 with the objective variable y being the moisture content and the explanatory variable x being the VSWR2, the following regression equation (35) is obtained. y=-699.96x+813.97 (35) The partial regression coefficient of the regression equation (35) is calculated by the analysis unit 30 using the well-known least squares method based on the VSWR2 data of the above 24. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.9601 (36) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is high. One reason for this is that the distance (747 mm) between the second antenna 11c and the lumber 100 is selected to provide a high correlation between the moisture content of the lumber 100 and the distance between the second antenna 11c and the lumber 100.

[0056] <Estimation of Water Content of Sample in Electromagnetic Wave Sensing Device of Third Embodiment of the Present Invention> As described above, in the electromagnetic wave sensing device 3 of the third embodiment, when a predetermined number of pieces of lumber 100 with moisture content in the range of 2% to 110% are prepared, data sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are average values ​​of the amount of change at each predetermined interval for each of the predetermined number of pieces of lumber 100, plus data sets of average values ​​VSWR1 and VSWR2 of average values ​​VSWR measured at each predetermined interval from the first antenna 11b and second antenna 11c are stored in the first and second folders of the average value folder in the memory 31. Here, when the number of pieces of lumber 100 with different types of moisture content prepared is, for example, 12 pieces, 12×2=24 sets of data are stored in the memory 31. In this case, the data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are the same as the data in the electromagnetic wave sensing device 2 of the second embodiment, and are as shown in the graphs in Figs. 13 to 18. The data of VSWR1 and VSWR2 are as described above, and are as shown in Figs. 21 and 22. Therefore, by performing multiple regression analysis in the analysis unit 30 with the moisture content as the response variable and the eight parameters of Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1, and VSWR2 as the explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is the following equation (37). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6 +b7x7+b8x8(37) where b0, b1, b2, b3, b4, b5, b6, b7, and b8 are partial regression coefficients, x1 is the numerical value of Re12A, x2 is the numerical value of Im12A, x3 is the numerical value of Re11A, x4 is the numerical value of Im11A, x5 is the numerical value of Re22A, x6 is the numerical value of Im22A, x7 is the numerical value of VSWR1, and x8 is the numerical value of VSWR2. When the partial regression coefficients of the above multiple regression equation are found by analysis unit 30 using the well-known least squares method based on the above 24 sets of data, equation (38) is obtained. y= 58666.25-4.98x1-25.65x2+9.82x3+6.88x4-10.56x5+14.11x6-55984.9x7-145.24x8(38) The y in equation (38) is the multiple regression equation for the calibration curve calculated from the average value of the data on the amount of change, including VSWR.

[0057] When it is desired to estimate the moisture content of lumber 100, as in the electromagnetic wave sensing device 2 of the second embodiment, the rectangular prismatic lumber 100, the moisture content of which is to be determined, is fed through the feed section 12, and each time the lumber 100 moves a predetermined distance, an electromagnetic wave of, for example, 4200 MHz is supplied from the measurement section 20 to the first antenna 11b via the first coaxial terminal 14a. Then, the electromagnetic waves irradiated from the first antenna 11b are received by the second antenna 11c, and the change in the real and imaginary parts of the wave transmitted through the piece of wood 100 for which the moisture content is to be found (referred to as Re12', Im12'"), the change in the real and imaginary parts of the reflected wave A (Re11', Im11'), the change in the real and imaginary parts of the reflected wave B (Re22', Im22'), the VSWR' from the first antenna 11b side, and the VSWR" from the second antenna 11c side are measured. This measurement is repeated twice. Then, each time the piece of wood 100 for which the moisture content is to be found moves a predetermined distance, the data Re12', Im12', Re11', Im11', Re22', Im22', VSWR', and VSWR" are measured, and the data are stored in the first folder in the folder for the interval corresponding to the measurement position of the piece of wood 100 in the measurement data folder in memory 31. As a result, the first folder of the measurement data folder interval 1 stores the set of data Re12', Im12', Re11', Im11', Re22', Im22', VSWR', VSWR" measured the first time, and the second folder of the measurement data folder of the measurement data folder interval 1 stores the data measured the second time.Similarly, the first and second folders of the measurement data folder of the measurement data folders of the interval 2 to interval 20 store the set of data Re12', Im12', Re11', Im11', Re22', Im22', VSWR', VSWR" measured the respective times.

[0058] Next, the analysis unit 30 calculates the average value of each of the 20 pieces of data Re12', Im12', Re11', Im11', Re22', Im22', VSWR', and VSWR" stored in the first folder of the folders for intervals 1 to 20 of the measurement data folder. The data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average values ​​of the amount of change at each predetermined interval of the lumber 100 calculated by the analysis unit 30, and the data VSWR1' and VSWR2', which are the average values ​​of VSWR' and VSWR" are stored in the first folder of the average value folder in the measurement data folder. Next, the analysis unit 30 calculates the average value of each of the 20 pieces of data Re12', Im12', Re11', Im11', Re22', and Im22' stored in the second folder of the folders for intervals 1 to 20. The data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' which are the average values ​​of the amount of change in the lumber 100 at specified intervals calculated by the analysis unit 30, and the data VSWR1' and VSWR2' which are the average values ​​of VSWR' and VSWR" are stored in a second folder under the average values ​​folder in the measurement data folder.

[0059] The two sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1', and VSWR2', of the average values ​​of the amount of change obtained each time the wood 100 whose moisture content is to be determined moves a predetermined distance, were substituted into the above formula (38), which is the formula for the calibration curve, to estimate the two moisture contents of the wood 100 whose moisture content is to be determined. Figure 23 shows a graph in which the two estimated moisture contents are plotted with the actual measured moisture content [%] on the vertical axis and the moisture content [%] calculated by the calibration curve of the wood 100 whose moisture content is to be determined on the horizontal axis. In Figure 23, the two estimated moisture contents are shown as white circles with VSWR. Referring to Fig. 23, two moisture contents calculated from the calibration curve are shown for each of 12 points of actual moisture content, which are about 2.1%, about 12.9%, about 23.0%, about 32.3%, about 41.0%, about 50.1%, about 59.9%, about 69.3%, about 80.1%, about 90.2%, about 102.3%, and about 110.5%. This is because the moisture content of the 12 pieces of lumber 100 with the moisture content of the above 12 points is the lumber 100 whose moisture content is to be obtained, and the moisture content of the calibration curve is obtained from the measurement data. Here, the regression equation yd of the calibration curve calculation is obtained by performing regression analysis in the analysis unit 30 with the moisture content yd of the calibration curve calculation as the response variable and the actual moisture content x as the explanatory variable. yd=0.9979x+0.2223 (39) yd is shown in FIG. 23 as a straight dashed line yd (with VSWR), and the partial regression coefficient of the regression equation (39) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 2×12=24. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.9959 (40) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is very high. In other words, it can be seen that the moisture content of the lumber 100, the moisture content of which is to be calculated, can be quantitatively estimated sufficiently.

[0060] In addition, in Fig. 23, the two moisture contents estimated in the electromagnetic wave sensing device 2 of the second embodiment of the present invention are plotted as gray squares with no VSWR. This gray square plot is the same as the plot shown as an open circle in Fig. 19. In this case, regression equation yc is obtained by performing regression analysis in analysis unit 30 with the moisture content calculated by the calibration curve as the objective variable and the actual moisture content as the explanatory variable, and yc is shown as a gray solid line in Fig. 23 as yc (without VSWR), but yc (without VSWR) and yb (averaged data) are the same. In other words, regression equation yc is considered to be the same as regression equation yb in equation (31) above, and the coefficient of determination R of this regression equation yc is 2 is 0.9233 as shown in (32) above. Comparing the moisture content estimated by the electromagnetic wave sensing device 3 of the third embodiment with the regression equation yd plotted as an open circle in the same figure with VSWR, it can be seen that the degree to which the objective variable can be explained by the explanatory variables is higher in the regression equation yd. In other words, it can be seen that the electromagnetic wave sensing device 3 of the third embodiment of the present invention can quantitatively estimate the moisture content of the lumber 100, the moisture content of which is to be determined, more satisfactorily than the electromagnetic wave sensing device 2 of the second embodiment. Moreover, a graph in which the error between the moisture content calculated by the calibration curve and the moisture content of the actual measurement is plotted on the vertical axis and the moisture content of the actual measurement is plotted on the horizontal axis is shown in Fig. 24. In Fig. 24, the error is shown as an open circle representing averaged data, and the two moisture contents estimated by the electromagnetic wave sensing device 2 of the second embodiment of the present invention and plotted as an open circle in Fig. 20 are plotted as gray rectangles with no VSWR. 23 and 24, it is apparent that the calibration curve shown in the above formula (38) is more reliable.

[0061] Referring to the graph in FIG. 23, it can be seen that regression equation yd (38), which was created when VSWR1 and VSWR2 are also added as explanatory variables, has a higher coefficient of determination and is closer to the actual measured values ​​than regression equation yb (31), which was created from the relationship between the moisture content and the average amount of change in the real and imaginary parts of the transmitted and reflected waves at regular intervals. In addition, as can be seen from the graph in FIG. 24, compared to the calculated values ​​of the calibration curve of Equation (31) created from the relationship between the average value of the change in the real and imaginary parts of the transmitted wave and reflected wave at regular intervals and the moisture content, the calculated values ​​of the calibration curve of Equation (38) created when VSWR1 and VSWR2 are also added as explanatory variables has less error. From the above, it can be said that the reliability of the calibration curve created with moisture content as the objective variable and eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1, and VSWR2 as explanatory variables is very high.

[0062] <Radio wave absorber provided in electromagnetic wave sensing device> In the electromagnetic wave sensing devices according to the embodiments of the present invention, which are the electromagnetic wave sensing device 1 according to the first embodiment to the electromagnetic wave sensing device 3 according to the third embodiment of the present invention, as shown in Fig. 3 and Fig. 4(a)(b), a radio wave absorber 11a for suppressing the reflection of electromagnetic waves is attached to the wall surface of the sensing housing 11 except for the antenna, a radio wave absorber 12a having a rectangular cylindrical shape is attached to the entire inner wall surface of the input section 12, and a radio wave absorber 13a having a rectangular cylindrical shape is attached to the entire inner wall surface of the output section 13. In this way, by attaching the radio wave absorbers 11a, 12a, and 13a to the electromagnetic wave sensing section 10 to cover the metal surface, the electromagnetic wave sensing device according to the embodiments of the present invention can suppress the occurrence of reflection and scattering of the electromagnetic waves irradiated from the first antenna 11b and the second antenna 11c. This can suppress the variation in the measurement data of the amplitude and phase of the transmitted wave in the electromagnetic wave sensing section 10, and can obtain highly reliable measurement data. Therefore, the reliability of the calibration curve created based on the measurement data can be improved.

[0063] 25 and 26 are diagrams showing the effect of attaching the radio wave absorbers 11a, 12a, and 13a to the electromagnetic wave sensing unit 10. Fig. 25 is a graph showing the change ΔReal in the measurement data of the real part of the transmitted wave received by the second antenna 11c when electromagnetic waves in the frequency range of 700 M to 4500 MHz are irradiated from the first antenna 11b for a certain period of time in a state in which the wood 100 is not inserted into the electromagnetic wave sensing unit 10. In the graph shown in Fig. 25, the horizontal axis is frequency and the vertical axis is the change ΔReal in the real part, and the graph shows a comparison between the case where the radio wave absorbers 11a, 12a, and 13a are attached to the electromagnetic wave sensing unit 10 and the case where they are not attached. That is, in the graph shown in Figure 25, the plots shown as white circles are measurement data taken at regular intervals when radio wave absorbers 11a, 12a, and 13a are attached to the electromagnetic wave sensing unit 10, and the plots shown as gray rectangles are measurement data taken at regular intervals when radio wave absorbers 11a, 12a, and 13a are not attached to the electromagnetic wave sensing unit 10.

[0064] Fig. 26 is a graph showing the change amount ΔImaginary of the measurement data of the imaginary part of the transmitted wave received by the second antenna 11c when the first antenna 11b irradiates electromagnetic waves in the frequency range of 700 M to 4500 MHz for a certain time in a state where the wood 100 is not inserted in the electromagnetic wave sensing unit 10. In the graph shown in Fig. 26, the horizontal axis is frequency and the vertical axis is the change amount ΔImaginary of the imaginary part, and the graph shows a comparison between the case where the radio wave absorbers 11a, 12a, and 13a are attached to the electromagnetic wave sensing unit 10 and the case where they are not attached. That is, in the graph shown in Fig. 26, the plots shown by the white circles are the measurement data at fixed intervals when the radio wave absorbers 11a, 12a, and 13a are attached to the electromagnetic wave sensing unit 10, and the plots shown by the gray rectangles are the measurement data at fixed intervals when the radio wave absorbers 11a, 12a, and 13a are not attached to the electromagnetic wave sensing unit 10. 25 and 26, it can be seen that the variation in the amount of change ΔReal of the real part and the amount of change ΔImaginary of the imaginary part is suppressed when the electromagnetic wave sensing unit 10 is fitted with the radio wave absorbers 11a, 12a, and 13a. It should be noted that highly reliable measurement data can be obtained at a frequency at which the amount of change ΔReal of the real part and the amount of change ΔImaginary of the imaginary part are small. For this reason, in the electromagnetic wave sensing device according to the embodiment of the present invention, the frequency at which measurement data is measured is set to, for example, 4200 MHz, at which the variation in the amount of change ΔReal of the real part and the amount of change ΔImaginary of the imaginary part is suppressed.

[0065] <Electromagnetic wave sensing device according to the fourth embodiment of the present invention> In the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention, instead of the sample of the prismatic lumber 100, wood chips 210 are used as samples, and the moisture content of the wood chips 210 is estimated. A configuration for measuring the wood chips 210 used as samples in the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention is shown in Fig. 27 and Fig. 28(a) and (b). Fig. 27 is a perspective view of the electromagnetic wave sensing unit 10 in the configuration for measuring the wood chips 210 with the electromagnetic wave sensing device 4 of the fourth embodiment, Fig. 28(a) is a side view of the electromagnetic wave sensing unit 10 in the configuration for measuring the wood chips 210 with the electromagnetic wave sensing device 4 of the fourth embodiment, and Fig. 28(b) is a cross-sectional view of the electromagnetic wave sensing unit 10 cut along line BB in the configuration for measuring the wood chips 210 with the electromagnetic wave sensing device 4 of the fourth embodiment. The electromagnetic wave sensing device 4 of the fourth embodiment differs from the electromagnetic wave sensing device 1 of the first embodiment in that a non-metallic container, for example an acrylic container 200 containing wood chips 210, is fed into the electromagnetic wave sensing unit 10 instead of the lumber 100, and the other configurations are the same as those of the electromagnetic wave sensing device 1 of the first embodiment. Therefore, a description of the configurations of the electromagnetic wave sensing unit 10, the measurement unit 20, and the analysis unit 30 will be omitted.

[0066] <Measurement of a sample (wood chips) using an electromagnetic wave sensing device according to a fourth embodiment of the present invention> In the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention, in order to quantitatively estimate the moisture content of wood chips 210, which are samples whose moisture content varies depending on the position, a container 200 containing wood chips 210, which are samples, is fed from a feed section 12 in the electromagnetic wave sensing section 10 as shown in Fig. 27, and the container 200 containing the wood chips 210 is positioned in the sensing housing 11 as shown in Fig. 28(b), and the tip side of the container 200 containing the wood chips 210 is sent out from a send section 13. Then, in the electromagnetic wave sensing device section 10, an electromagnetic wave of an arbitrary frequency within a frequency range of 100k to 50GHz, for example, 770MHz, is irradiated to the container 200 containing the wood chips 210, and the amplitude and phase of the transmitted wave and the reflected wave from the container 200 containing the wood chips 210 are measured. In this case, the amplitude and phase of the reflected wave and the transmitted wave can also be expressed as the real and imaginary parts of the complex representation of the reflected wave and the transmitted wave, and here, data on the real and imaginary parts of the reflected wave and the transmitted wave are measured. However, before measuring the container 200 containing the wood chips 210, the same initial setting as in the electromagnetic wave sensing device 1 of the first embodiment is performed. The initial setting is as described above and will not be described in detail, but in the initial setting, when nothing is fed into the electromagnetic wave sensing unit 10, the Re12 base, Im12 base, Re11 base, Im11 base, Re22 base, and Im22 base are measured and stored in the memory 31 as described above.

[0067] Next, a plurality of wood chips 210, for example, nine types, each having a known moisture content in the range of 0% to 36% are prepared, and for each container 200 storing the wood chips 210 with the nine types of moisture content, data on the amount of change at a predetermined interval, Re12, Im12, Re11, Im11, Re22, and Im22, are measured in the same manner as in the electromagnetic wave sensing device 1 of the first embodiment, and the measured data on the amount of change is stored in the memory 31. Note that the data on the amount of change is the amount of change from the initial setting data. By carrying out the above-mentioned operation, data on the amount of change of Re12, Im12, Re11, Im11, Re22, and Im22 at each predetermined interval of the container 200 storing wood chips 210 with a predetermined moisture content is obtained. In this case, the number of predetermined intervals is set to, for example, 20, and 20 sets of data on the amount of change of Re12, Im12, Re11, Im11, Re22, and Im22, the same number as the number of predetermined intervals, are obtained. Next, by carrying out the above-mentioned operation for other wood chips 210 with different moisture contents, 20 sets of data on the amount of change of Re12, Im12, Re11, Im11, Re22, and Im22 for the other wood chips 210 with different moisture contents are obtained. By repeating this process, 20 sets of change data consisting of Re12, Im12, Re11, Im11, Re22, and Im22 for the prepared wood chips 210 with multiple types of moisture content are stored in the memory 31. Note that the above-mentioned measurement is repeated four times for the same wood chip 210, so 20×4=80 sets of change data are measured for one wood chip 210. If the number of prepared wood chips 210 with multiple types of moisture content is, for example, nine types, 80 sets of change data are measured for each type, and 80×9=720 sets of change data are measured.

[0068] <Sample in the electromagnetic wave sensing device according to the fourth embodiment of the present invention (Estimation of moisture content of wood chips 1> As described above, in the electromagnetic wave sensing device 4 of the fourth embodiment, when wood chips 210 having, for example, nine different moisture contents in the range of 0% to 36% are prepared, 80 sets of change amount data are measured for each type as described above, and 80×9=720 sets of change amount data are stored in the memory 31. Then, by performing multiple regression analysis in the analysis unit 30 with the moisture content as the objective variable y and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 as explanatory variables, the multiple regression equation of the calibration curve can be obtained. This multiple regression equation is similar to the above-mentioned equation (13). When the partial regression coefficients of the multiple regression equation of equation (13) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 720 sets of change amount data, the multiple regression equation shown in the following equation (41) is obtained. y=-289.54-17.75x1-138.99x2+135.99x3-169.18x4+120.00x5-254.40x6(41) The y in equation (41) is the multiple regression equation for the calibration curve calculated from the change data.

[0069] When it is desired to estimate the moisture content of wood chips 210 for which the moisture content is to be determined, a container 200 containing the wood chips 210 for which the moisture content is to be determined is fed from the feed section 12, and an electromagnetic wave of, for example, 770 MHz is supplied from the measurement section 20 to the first antenna 11b via the first coaxial terminal 14a every time the container 200 containing the wood chips 210 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave of the wood chips 210 for which the moisture content is to be determined are stored in the memory 31. Furthermore, the first antenna 11b receives the reflected wave of, for example, a 770 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the wood chip 210, whose moisture content is to be found. Furthermore, the second antenna 11c receives the reflected wave of, for example, a 770 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the wood chip 210, whose moisture content is to be found. Re12', Im12', Re11', Im11', Re22', and Im22' are change amount data measured every time the wood chip 210 moves a predetermined interval, and in this case, the number of predetermined intervals is 20 as above, and 20 sets of change amount data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. And, since the same measurement is repeated four times, 80 sets of change amount data for the wood chip 210 for which the moisture content is to be calculated are stored in the memory 31.

[0070] The moisture content of the wood chip 210 for which the moisture content is to be determined can be estimated by substituting the above-mentioned 80 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 200 storing the wood chip 210 for which the moisture content is to be determined moves a predetermined distance into the above-mentioned formula (41), which is the formula of the calibration curve. When the above-mentioned 80 sets of data are respectively substituted into the formula (41), the number of estimated moisture contents becomes 80. Then, the average value of the 80 estimated moisture contents can be calculated by the analysis unit 30 as the estimated moisture content of the wood chip 210 for which the moisture content is to be determined, and presented. Alternatively, the range of the average value of the 80 moisture contents ± the standard deviation may be presented. Here, the graph shown in FIG. 29 is created so that the variation in the moisture content of 80 estimated by the calibration curve of the multiple regression equation shown in Equation (41) can be visualized. The graph shown in FIG. 29 plots the moisture content of 80 with the moisture content [%] of the calibration curve calculated on the horizontal axis and the moisture content [%] of the actual measured value of the wood chip 210 for which the moisture content is to be calculated on the vertical axis. The numerical values ​​of the above 80 sets of measurement data are omitted, but in FIG. 29, the estimated moisture content of 80 is shown as a gray rectangle that is raw data. Referring to FIG. 29, the moisture content of 80 calculated by the calibration curve is shown for each of nine points of the moisture content of the actual measured value, namely 1.16%, about 5.48%, about 9.85%, about 14.35%, about 18.99%, about 24.10%, about 28.37%, about 32.06%, and about 36.30. This is because the nine types of wood chips 210 with the moisture contents at the nine points are the wood chips 210 for which moisture contents are to be calculated, and the moisture contents for the calibration curve are calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in Fig. 29 with the actual moisture content ye as the response variable and the moisture content x for the calibration curve calculation as the explanatory variable, the following regression equation (42) is obtained. ye=x+2×10 -13 (42) ye is shown in FIG. 29 as ye (raw data) by a solid gray line, and the partial regression coefficient of the regression equation (42) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 80×9=720. The coefficient of determination R 2 teeth, R 2 =0.994 (43) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the moisture content of the wood chips 210 for which the moisture content is to be calculated. Fig. 30 shows a graph in which the error between the moisture content at 40 calculated from the calibration curve and the measured moisture content in the case shown in Fig. 29 is plotted on the vertical axis and the measured moisture content on the horizontal axis, plotting the error in the moisture content at 80 calculated from the calibration curve. Referring to Fig. 30, the error is shown as a gray square representing raw data, and it can be seen that the error is within approximately ±3.0%. Therefore, referring to Figs. 29 and 30, it is evident that the calibration curve shown in the above formula (41) is highly reliable.

[0071] <Estimation of moisture content of a sample (wood chips) using an electromagnetic wave sensing device according to a fourth embodiment of the present invention, part 2> In step 2 of estimating the moisture content of a sample in the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention, the analysis unit 30 calculates the average values ​​of the data of the amount of change at a predetermined interval for the nine types of moisture content of the wood chips 210 obtained in step 1 of estimating the moisture content of a sample in the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention, and stores the calculated average values ​​in the memory 31. If the average values ​​of the data of the amount of change of Re12, Im12, Re11, Im11, Re22, and Im22 are Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, the data of the average values ​​of the sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A is stored in the memory 31. Here, the average value of four sets of change data is calculated by measuring the moisture content of wood chips 210 four times, so if measurements of wood chips 210 with nine different moisture contents are repeated four times, a total of 4×9=36 sets of change data average data are stored in memory 31. Then, by performing multiple regression analysis in analysis unit 30 using moisture content y as the objective variable and six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is the same as the above-mentioned equation (29), and when the partial regression coefficient of the multiple regression equation of equation (29) is calculated by analysis unit 30 using the well-known least squares method based on the above 36 sets of data, the multiple regression equation shown in equation (44) is obtained. y=56.42-2.53x1+26.26x2-256.26x3+2.11x4+229.85x5+105.30x6(44) However, x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, and x6 is the value of Im22A. The y in equation (44) is the multiple regression equation for the calibration curve calculated from the average value of the change amount data.

[0072] When it is desired to estimate the moisture content of the wood chips 210 for which the moisture content is to be determined, a measurement similar to that of the electromagnetic wave sensing device 2 of the second embodiment is performed, and the average data of a set of change amount data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which is the average value of the change amount data at predetermined intervals of the wood chips 210 calculated by the analysis unit 30 for each measurement, is stored in the memory 31. Here, since the measurement is performed repeatedly four times, the average data of four sets of change amount data is stored in the memory 31. The moisture content of the wood chip 210 for which the moisture content is to be determined can be estimated by substituting the above-mentioned four sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average values ​​of the data of the amount of change at predetermined intervals of the wood chip 210 for which the moisture content is to be determined, into the above-mentioned formula (44), which is the formula of the calibration curve. When the above-mentioned four sets of data are respectively substituted into formula (44), the number of estimated moisture contents becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated moisture contents as the estimated moisture content of the wood chip 210 for which the moisture content is to be determined.

[0073] Here, the graph shown in Fig. 29 is created so that the variation in the four moisture contents estimated by the calibration curve of the multiple regression equation shown in Equation (44) can be visualized. The graph shown in Fig. 29 plots the above four moisture contents with the moisture content [%] calculated by the calibration curve on the horizontal axis and the moisture content [%] of the actual measured value of the wood chip 210 for which the moisture content is to be calculated on the vertical axis. Here, the numerical values ​​of the four sets of measurement data are omitted, but referring to the graph in FIG. 29, the estimated moisture content is shown as four white circles that are averaged data. In FIG. 29, four moisture contents calculated from the calibration curve are shown for nine points of moisture contents of actual measurement values ​​of 1.16%, about 5.48%, about 9.85%, about 14.35%, about 18.99%, about 24.10%, about 28.37%, about 32.06%, and about 36.30. This is because the nine types of wood chips 210 with the moisture contents of the nine points are wood chips 210 whose moisture contents are to be obtained, and the moisture contents of the calibration curve are obtained from the measurement data. In other words, since four estimated moisture contents are shown for each point, the number of estimated moisture contents shown in FIG. 29 is 4×9=36. Here, when the actual moisture content is set as the objective variable yf and the moisture content x of the calibration curve calculation 36 is set as the explanatory variable and regression analysis is performed in the analysis unit 30 for the case shown in FIG. 29, the following regression equation (45) is obtained. yf=x+4×10 -14 (45) yf is shown in FIG. 29 as a straight dashed line as yf (averaged data), and the partial regression coefficient of the regression equation (45) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 4×9=36. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.998 (46) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is higher. In other words, it can be seen that calculating the average value of the data of the amount of change in the wood chips 210 at predetermined intervals allows a more quantitative estimation of the moisture content of the wood chips 210 for which the moisture content is to be calculated. Fig. 30 shows a graph in which the error between the four moisture contents calculated from the calibration curve and the measured moisture contents is plotted on the vertical axis, and the measured moisture contents are plotted on the horizontal axis, in the case shown by the dashed line in Fig. 29. In Fig. 30, the error is shown as four open circles representing averaged data, and it can be seen that the error is between about -1.5% and about 1.0%. 29 and 30, it can be seen that the calibration curve shown in the above formula (44) is more reliable than the calibration curve shown in the above formula (41).

[0074] <Estimation of moisture content of sample (wood chips) using electromagnetic wave sensing device according to the fourth embodiment of the present invention, part 3> In the estimation of the moisture content of a sample in the electromagnetic wave sensing device 4 of the fourth embodiment of the present invention, similar to the estimation of the sample in the electromagnetic wave sensing device of the third embodiment of the present invention, the VSWR is measured and added to the explanatory variables, and multiple regression analysis is performed in the analysis unit 30. In this case, in the electromagnetic wave sensing device 4 of the fourth embodiment, the memory 31 stores VSWR data in addition to the data on the amount of change, and stores average VSWR data in addition to the data on the average amount of change data. Specifically, the distance between the first antenna 11b and the wood chips 210 stored in the sample container 200 is set to, for example, 253.8 mm, and an electromagnetic wave of 770 MHz is irradiated from the first antenna 11b side, and VSWR data VSWR1 is also measured when data (Re12, Im12, Re11, Im11) of the change in the transmitted wave and the reflected wave at each predetermined interval of the container 200 storing the wood chips 210 is measured. In addition, an electromagnetic wave of 770 MHz is irradiated from the second antenna 11c side, and data (Re22, Im22) of the change in the reflected wave at each predetermined interval of the container 200 storing the wood chips 210 is measured, and VSWR data VSWR2 is also measured. Next, the average value of the change amount data measured at predetermined intervals is calculated by the analysis unit 30, and the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A mentioned above, the average value data VSWR1A of VSWR1, and the average value data VSWR2A of VSWR2 are stored in the memory 31. In this case, measurements of wood chips 210 with nine types of known moisture content are repeated four times, and the average data of a total of 36 sets of change amount data are stored in memory 31. Then, by performing multiple regression analysis in analysis unit 30 using moisture content y as the objective variable and eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is the same as equation (37) described above, and is shown below as equation (47). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6 +b7x7+b8x8(47) where b0, b1, b2, b3, b4, b5, b6, b7, and b8 are partial regression coefficients, x1 is the numerical value of Re12A, x2 is the numerical value of Im12A, x3 is the numerical value of Re11A, x4 is the numerical value of Im11A, x5 is the numerical value of Re22A, x6 is the numerical value of Im22A, x7 is the numerical value of VSWR1A, and x8 is the numerical value of VSWR2A. Although the numerical values ​​of the 36 sets of measurement data are omitted here, when the partial regression coefficients of the multiple regression equation are calculated by analysis unit 30 based on the 36 sets of data using the well-known least squares method, equation (48) is obtained. y=-106.64-12.59x1-208.64x2+190.63x3-165.23x4+75.97x5-189.95x6-337.06x7+69.17x8(48) The y in equation (48) is the multiple regression equation for the calibration curve calculated from the average value of the data on the amount of change, including VSWR.

[0075] When it is desired to estimate the moisture content of the wood chip 210 for which the moisture content is to be obtained, a measurement similar to that of the electromagnetic wave sensing device 3 of the third embodiment is performed, and the average data of one set of change amount data Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which is the average value of the change amount data at each predetermined interval of the wood chip 210 calculated by the analysis unit 30 for each measurement, is stored in the memory 31. Here, since the measurement is performed repeatedly four times, the average data of four sets of change amount data is stored in the memory 31. The moisture content of the wood chip 210 for which the moisture content is to be determined can be estimated by substituting the above-mentioned four sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data of the amount of change at predetermined intervals of the wood chip 210 for which the moisture content is to be determined, into the above-mentioned formula (48), which is the formula of the calibration curve. When the above-mentioned four sets of data are respectively substituted into the formula (48), the number of estimated moisture contents becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated moisture contents as the estimated moisture content of the wood chip 210 for which the moisture content is to be determined.

[0076] Here, in order to visualize the variation in the four moisture contents estimated by the calibration curve of the multiple regression equation shown in Equation (48), the graph shown in Fig. 31 is created. The graph shown in Fig. 31 plots the above four moisture contents with the moisture content [%] calculated by the calibration curve on the horizontal axis and the moisture content [%] of the actual measured value of the wood chip 210 for which the moisture content is to be calculated on the vertical axis. Here, the numerical values ​​of the four sets of measurement data are omitted, but referring to the graph in FIG. 31, the estimated moisture content is shown as four white circles that are averaged data. In FIG. 31, four moisture contents calculated from the calibration curve are shown for nine points of actual moisture content values ​​of about 1.16%, about 5.48%, about 9.85%, about 14.35%, about 18.99%, about 24.10%, about 28.37%, about 32.06%, and about 36.30%. This is because the nine types of wood chips 210 with the moisture contents at the nine points are wood chips 210 whose moisture content is to be calculated, and the moisture content calculated from the calibration curve is shown from the measurement data. In other words, since four estimated moisture contents are shown for each point, the number of estimated moisture contents shown in FIG. 31 is 4×9=36. Here, when the actual moisture content is set as the objective variable yg and the moisture content x of the calibration curve calculation 36 is set as the explanatory variable and regression analysis is performed in the analysis unit 30 for the case shown in FIG. 31, the following regression equation (49) is obtained. yg=x+4×10 -14 (49) yg is shown in FIG. 31 as a straight dashed line yg (with VSWR), and the partial regression coefficient of the regression equation (49) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated moisture content of 4×9=36. The coefficient of determination R 2 teeth, R 2 =0.9986 (50) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is further improved. In other words, it can be seen that the moisture content of the wood chip 210, the moisture content of which is to be calculated, can be more quantitatively estimated by calculating the average value of the data of the amount of change including the VSWR data of the wood chip 210 at predetermined intervals. Note that the gray rectangular plot shown as yf (without VSWR) in Fig. 31 is the same as the plot shown by the white circle in Fig. 29. FIG. 32 shows a graph in which the error of the 36 moisture contents calculated by the calibration curve is plotted on a graph with the error between the four moisture contents calculated by the calibration curve and the measured moisture contents on the vertical axis and the measured moisture contents on the horizontal axis in the case shown by the dashed line in FIG. 31. In FIG. 32, the error is shown by four open circles with VSWR, and it can be seen that the error is within about ±1.0%. Note that the gray square plot shown as "without VSWR" in FIG. 32 is the same as the plot shown by the open circles in FIG. 30. 31 and 32, it can be seen that the calibration curve shown in the above formula (48), which is created when VSWR1A and VSWRA2 are also added as explanatory variables, is more reliable than the calibration curve shown in the above formula (44). From the above, it can be said that the reliability of the calibration curve created with moisture content as the objective variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables is very high.

[0077] <Measurement of concentration of solution in electromagnetic wave sensing device according to the fifth embodiment of the present invention> In the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, a solution 310 containing a solute is used as a sample instead of a sample of a rectangular prismatic lumber 100, and the concentration of the solute in the solution 310 is estimated. A configuration for measuring the solution 310 used as a sample in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention is shown in Fig. 33 and Fig. 34(a) and (b). Fig. 33 is a perspective view of the electromagnetic wave sensing unit 10 in the configuration for measuring the solution 310 with the electromagnetic wave sensing device 5 of the fifth embodiment, Fig. 34(a) is a side view of the electromagnetic wave sensing unit 10 in the configuration for measuring the solution 310 with the electromagnetic wave sensing device 5 of the fifth embodiment, and Fig. 34(b) is a cross-sectional view of the electromagnetic wave sensing unit 10 cut along line BB in the configuration for measuring the solution 310 with the electromagnetic wave sensing device 5 of the fifth embodiment. The electromagnetic wave sensing device 5 of the fifth embodiment differs from the electromagnetic wave sensing device 1 of the first embodiment in that a non-metallic container, for example an acrylic container 300, containing a solution 310 is fed into the electromagnetic wave sensing unit 10 instead of the lumber 100, and the other configurations are the same as those of the electromagnetic wave sensing device 1 of the first embodiment. Therefore, a description of the configurations of the electromagnetic wave sensing unit 10, the measurement unit 20, and the analysis unit 30 will be omitted.

[0078] In the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, in order to quantitatively estimate the concentration of a solute in a solution 310, a container 300 containing a sample solution 310 is fed from an inlet section 12 to an electromagnetic wave sensing section 10 as shown in Fig. 33, and the container 300 containing the solution 310 is positioned in a sensing housing 11 as shown in Fig. 34(b), and the tip side of the container 300 containing the solution 310 is fed from a outlet section 13. Then, in the electromagnetic wave sensing device section 10, an electromagnetic wave of an arbitrary frequency within a frequency range of 100k to 50GHz is irradiated to the container 300 containing the solution 310, and the amplitude and phase of the transmitted wave and the reflected wave from the container 300 containing the solution 310 are measured. In this case, the frequency can be changed depending on the solute, for example, 3578 MHz, 1176 MHz, or 748 MHz, and the amplitude and phase of the reflected wave and transmitted wave can also be expressed as the real and imaginary parts of the complex representation of the reflected wave and transmitted wave, and here, data on the real and imaginary parts of the reflected wave and transmitted wave are measured. However, before measuring the container 300 containing the solution 310, an initial setting is performed. In the initial setting, in a state in which nothing is fed into the electromagnetic wave sensing unit 10, the Re12 base, Im12 base, Re11 base, Im11 base, Re22 base, and Im22 base are measured and stored in the memory 31 as described above.

[0079] Here, when the solution 310 is prepared with a plurality of different solute concentrations, for example, in the range of 0% to 30%, by carrying out the above-mentioned measurement, data on the amount of change at a predetermined interval in the container 300 containing the solutions 310 of the plurality of concentrations, Re12, Im12, Re11, Im11, Re22, and Im22, are measured for each type and stored in the memory 31. The data on the amount of change is the amount of change from the initial setting data. By carrying out the above-mentioned operation, data on the change amount of Re12, Im12, Re11, Im11, Re22, and Im22 at each predetermined interval of the container 300 containing the solution 310 of a solute with a predetermined concentration is obtained. In this case, the number of predetermined intervals is, for example, 20, and the data on the change amount of Re12, Im12, Re11, Im11, Re22, and Im22 is obtained in the same number as the number of predetermined intervals, that is, 20 sets. In this case, the above-mentioned measurement is repeated four times, so that 20×4=80 sets of change amount data are measured. And, when seven types of solutions with different concentrations are prepared, the above-mentioned operation is carried out for each solution 310 with different concentrations, so that a total of 80×7=560 sets of change amount data of Re12, Im12, Re11, Im11, Re22, and Im22 are obtained and stored in the memory 31. The number of solutions 310 with different concentrations prepared is not limited to seven, and may be six or another number of types. If six types are used, a total of 80×6=480 sets of change amount data are acquired and stored in the memory 31. The solution 310 can be an aqueous solution 320. The concentration of the solute in the aqueous solution 320 can be a sugar content, a salt concentration, an alcohol content, or the like, and it is preferable to prepare the solution 310 with a plurality of different concentrations within a predetermined range.

[0080] <Estimation of concentration of solute in solution in electromagnetic wave sensing device according to the fifth embodiment of the present invention, Part 1> As described above, in the electromagnetic wave sensing device 5 of the fifth embodiment, when aqueous solutions 320 with seven different concentrations in the range of 0% to 30% are prepared, 80 sets of data on the changes in Re12, Im12, Re11, Im11, Re22, and Im22 are measured for each type as described above, and a total of 80×7=560 sets of change data are measured and stored in the memory 31. Then, by performing multiple regression analysis in the analysis unit 30 with the concentration as the response variable y and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is the same as the above-mentioned equation (13), and is shown below as equation (51). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6(51) In equation (51), b0, b1, b2, b3, b4, b5, and b6 are partial regression coefficients, x1 is the value of Re12, x2 is the value of Im12, x3 is the value of Re11, x4 is the value of Im11, x5 is the value of Re22, and x6 is the value of Im22. The partial regression coefficients in equation (51) can be found by the analysis unit 30 using the well-known least squares method based on the above 80 x 7 = 560 sets of data. y in equation (51) is the multiple regression equation for the calibration curve calculated from the change amount data.

[0081] When it is desired to estimate the concentration of a solution 310, the container 300 containing the solution 310 is fed from the feed unit 12, and Re12', Im12', Re11', Im11', Re22', and Im22' are measured every time the solution moves at a predetermined interval. In this case, the number of predetermined intervals is 20, as above, and 20 sets of change amount data, each set consisting of Re12', Im12', Re11', Im11', Re22', and Im22', are measured, the same number as the number of predetermined intervals. Then, the same measurement is repeated four times, so that 80 sets of data for the solution 310 are stored in the memory 31. The above-mentioned 80 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the solution 310 whose concentration is to be determined moves a predetermined distance are substituted into the above formula (51) which is the formula of the calibration curve, to estimate the concentration of the solution 310 whose concentration is to be determined. The concentration estimated by substituting the 80 sets of data is 80. Then, the analysis unit 30 can calculate and present the average value of the 80 concentrations estimated as the estimated concentration of the solution 310 whose concentration is to be determined. Alternatively, the range of the average water content of 80 ± the standard deviation may be presented.

[0082] <Estimation of concentration of solute in solution in the electromagnetic wave sensing device according to the fifth embodiment of the present invention, part 2> In step 2 of estimating the concentration of a solute in a solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average value of the data on the amount of change at each predetermined interval acquired in step 1 of estimating the concentration of a solute in a solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention is calculated by the analysis unit 30. Then, the analysis unit 30 performs multiple regression analysis using the calculated average data on the amount of change as an explanatory variable. Specifically, data on the amount of change at each predetermined interval of the container 300 containing the solution 310 is measured, and the average value of the data on the amount of change at each predetermined interval is calculated in the analysis unit 30. Here, if the average data of the amount of change data Re12, Im12, Re11, Im11, Re22, and Im22 is represented as Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, a set of data Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A is stored in the memory 31. In this case, if the same measurement is performed four times, four sets of average data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in the memory 31. When solutions 310 of seven different concentrations are prepared, four sets of average value data are obtained for each type, and a total of 4×7=28 sets of average value data are stored in memory 31. Next, the multiple regression equation of the calibration curve can be obtained by performing multiple regression analysis in the analysis unit 30 with the concentration as the response variable y and the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A as explanatory variables. This multiple regression equation is the same as the above-mentioned equation (29), and is shown below as equation (52). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6(52) In equation (52), b0, b1, b2, b3, b4, b5, and b6 are partial regression coefficients, and x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, and x6 is the value of Im22A. The partial regression coefficients of this multiple regression equation can be determined by the analysis unit 30 using the well-known least squares method based on the above 28 sets of data. y in equation (52) is the multiple regression equation of the calibration curve calculated from the average value of the change amount data.

[0083] When it is desired to estimate the concentration of a solution 310, the container 300 containing the solution 310 is fed from the feed unit 12 in the same manner as in the estimation of the concentration of a solute in a solution in the electromagnetic wave sensing device of the fifth embodiment of the present invention, and Re12', Im12', Re11', Im11', Re22', and Im22' are measured every time the solution moves at a predetermined interval. In this case, the number of predetermined intervals is set to 20 as above, and 20 sets of data on the amount of change each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. Then, the same measurement is repeated four times, so that 80 sets of data on the solution 310 for which the concentration is desired are stored in the memory 31.

[0084] Next, the analysis unit 30 calculates the average value of the change amount data for each predetermined interval stored in the memory 31, and stores the data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average values ​​of the calculated change amount data, in the memory 31. In this case, the same measurement is performed four times for each solution 310 whose concentration is to be determined, so that the data of the average values ​​of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' is stored in the memory 31. The concentration of the solution 310 whose concentration is to be determined can be estimated by substituting the four sets of data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average data of Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the solution 310 whose concentration is to be determined moves a predetermined distance, into the above formula (52) which is the formula of the calibration curve. When the four sets of data are respectively substituted into the formula (52), the number of estimated concentrations becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated concentrations as the estimated concentration of the solution 310 whose concentration is to be determined.

[0085] <Estimation of concentration of solute in solution in the electromagnetic wave sensing device according to the fifth embodiment of the present invention, part 3> In step 3 of estimating the concentration of a solute in a solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, in addition to the average data of the change amount data acquired in step 2 of estimating the concentration of a solute in a solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average data of the VSWR is added to the explanatory variables and multiple regression analysis is performed in the analysis unit 30. In this case, in the electromagnetic wave sensing device 5 of the fifth embodiment, in addition to the change amount data, VSWR data is measured and stored in the memory 31. Then, the analysis unit 30 calculates average data of the change amount data and the VSWR data stored in the memory 31, and the average data of the change amount data and the average data of the VSWR are stored in the memory 31. Specifically, the distance between the first antenna 11b and the container 300 containing the solution 310 is set as a predetermined distance, and electromagnetic waves of 3578 MHz, 1176 MHz, or 748 MHz are irradiated from the first antenna 11b side to measure data (Re12, Im12, Re11, Im11) of the change in the transmitted wave and the reflected wave at every predetermined interval of the container 300 containing the solution 310, and VSWR data VSWR1 is also measured. In addition, the distance between the second antenna 11c and the container 300 containing the solution 310 is set as a predetermined distance, and electromagnetic waves of 3578 MHz, 1176 MHz, or 748 MHz are irradiated from the second antenna 11c side to measure data (Re22, Im22) of the change in the reflected wave at every predetermined interval of the container 300 containing the solution 310, and VSWR data VSWR2 is also measured. Next, the average value of the change amount data measured at predetermined intervals is calculated by the analysis unit 30, and the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A mentioned above, the average value data VSWR1A of VSWR1, and the average value data VSWR2A of VSWR2 are stored in the memory 31.

[0086] In this case, since the measurement of one solution 310 is repeated four times, the average data of four sets of change amount data is calculated for each type, and if solutions 310 of seven concentrations are prepared, the average data of a total of 4×7=28 sets of change amount data is stored in memory 31. Then, by performing multiple regression analysis in the analysis unit 30 with the concentration as the objective variable y and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is the same as the above-mentioned equation (37), and is shown below as equation (53). y=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6 +b7x7+b8x8(53) Here, b0, b1, b2, b3, b4, b5, b6, b7, and b8 are partial regression coefficients, and x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, x6 is the value of Im22A, x7 is the value of VSWR1A, and x8 is the value of VSWR2A. The partial regression coefficients of the above multiple regression equation can be calculated by the analysis unit 30 using the well-known least squares method based on the above 28 sets of data. y in equation (53) is the multiple regression equation of the calibration curve calculated from the average value of the data of the amount of change including VSWR.

[0087] When it is desired to estimate the concentration of the solution 310, similarly to the estimation 2 of the concentration of the solute of the solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, four sets of data Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are data on the average values ​​of Re12', Im12', Re11', Im11', Re22', and Im22', are stored in the memory 31. Furthermore, in the solution 310, the concentration of which is desired to be determined, the operation of measuring VSWR1' and VSWR2', which are data on the amount of change in VSWR described above, is performed, and further, the analysis unit 30 calculates data VSWR1A' and VSWR2A', which are average values ​​of the data on the amount of change in VSWR1' and VSWR2', and stores the data in the memory 31. The concentration of the solution 310 whose concentration is to be determined can be estimated by substituting the four sets of data Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average data of Re12', Im12', Re11', Im11', Re22', Im22', VSWR1', and VSWR2' obtained each time the container 300 containing the solution 310 whose concentration is to be determined moves a predetermined distance, into the above formula (53) which is the formula of the calibration curve. When the four sets of data are respectively substituted into the formula (53), the number of estimated concentrations becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated concentrations as the estimated concentration of the solution 310 whose concentration is to be determined.

[0088] <Estimation of sugar content of aqueous solution using electromagnetic wave sensing device according to the fifth embodiment of the present invention, 1> Next, a case will be described in which the concentration of the solute in the solution 310 is taken as the sugar content of the aqueous solution 320 in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, and the sugar content of the aqueous solution 320 is estimated. In this case, the aqueous solution 320 is prepared with a plurality of different sugar contents ranging from 0% to 30%, for example, seven types of sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. The above-mentioned measurement is performed with the frequency of the electromagnetic waves emitted from the first antenna 11b or the second antenna 11c set to 3578 MHz. By performing the above-mentioned measurement, data of the change amount at each predetermined interval of the container 300 containing the aqueous solution 320 with the seven types of sugar contents, Re12, Im12, Re11, Im11, Re22, and Im22, is measured for each type of sugar content and stored in the memory 31. The number of predetermined intervals of the container 300 where the change amount data is measured is, for example, 20, and the change amount data with the set of Re12, Im12, Re11, Im11, Re22, and Im22, which is the same number as the number of predetermined intervals, is obtained as 20 sets. Next, the above-mentioned operation is performed on other aqueous solutions 320 with different sugar contents, and 20 sets of change data consisting of Re12, Im12, Re11, Im11, Re22, and Im22 for the other aqueous solutions 320 with different sugar contents are obtained. By repeating this process, 20 sets of change data consisting of Re12, Im12, Re11, Im11, Re22, and Im22 for each of the seven other aqueous solutions 320 with different sugar contents prepared are stored in the memory 31. Since the above-mentioned measurement is repeated four times for the same aqueous solution 320, 20×4=80 sets of change data are measured for each aqueous solution 320. Here, the number of aqueous solutions 320 with different sugar contents prepared is seven, but it may be seven or more, and if seven types are used, 80 sets of change data are measured for each type, and 80×7=560 sets of change data are measured and stored in the memory 31.

[0089] Next, the sugar content is set as the objective variable y, and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 are set as explanatory variables to perform multiple regression analysis in the analysis unit 30, thereby obtaining the multiple regression equation of the calibration curve. This multiple regression equation is similar to the above-mentioned equation (51). Here, the numerical values ​​of the above-mentioned 560 sets of measurement data are omitted, but if the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (51) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 80 x 7 = 560 sets of data, the multiple regression equation shown in equation (54) is obtained. y=20.04+12.13x1+6.67x2-21.41x3+36.53x4+76.65x5+81.03x6(54) In equation (54), b0, b1, b2, b3, b4, b5, and b6 are partial regression coefficients, x1 is the value of Re12, x2 is the value of Im12, x3 is the value of Re11, x4 is the value of Im11, x5 is the value of Re22, and x6 is the value of Im22. y in equation (54) is the multiple regression equation for the calibration curve calculated from the change amount data.

[0090] When it is desired to estimate the sugar content of an aqueous solution 320 for which the sugar content is to be determined, the container 300 containing the aqueous solution 320 for which the sugar content is to be determined is fed in from the feed unit 12, and an electromagnetic wave of, for example, 3578 MHz is supplied from the measurement unit 20 to the first antenna 11b via the first coaxial terminal 14a each time the container 300 containing the aqueous solution 320 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave through the container 300 containing the aqueous solution 320 for which the sugar content is to be determined are stored in the memory 31. Furthermore, the first antenna 11b receives a reflected wave of, for example, 3578 MHz electromagnetic waves, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the container 300 containing the aqueous solution 320 whose sugar content is to be determined. Furthermore, the second antenna 11c receives a reflected wave of, for example, 3578 MHz electromagnetic waves, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the container 300 containing the aqueous solution 320 whose sugar content is to be determined. Re12', Im12', Re11', Im11', Re22', and Im22' are data measured every time the sensor moves a predetermined interval. In this case, the number of predetermined intervals is 20, as above, and 20 sets of data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. In other words, the 20 sets of data are data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' obtained every 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 whose sugar content is to be calculated are stored in the memory 31. Here, Re12', Im12', Re11', Im11', Re22', and Im22' are data on the amount of change, which is the amount of change from the initial setting data measured in a state where the container 300 is not inserted in the electromagnetic wave sensing unit 10.

[0091] The 80 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose sugar content is to be determined moves a predetermined distance were substituted into the above formula (54), which is the formula for the calibration curve, to estimate the sugar content of the aqueous solution 320 whose sugar content is to be determined. The sugar content estimated by substituting the 80 sets of data is 80. The analysis unit 30 can then calculate and display the average sugar content of the 80 estimated sugar contents as the estimated sugar content of the aqueous solution 320 whose sugar content is to be determined. Alternatively, the range of the average sugar content of 80 ± the standard deviation may be displayed. Here, the graph shown in Fig. 41 is created to visualize the variation in sugar content of 80 estimated by the calibration curve of the multiple regression equation shown in Equation (54). The graph shown in Fig. 41 plots the sugar content of 80 with the sugar content [%] calculated by the calibration curve on the horizontal axis and the sugar content [%] of the actual measured value of aqueous solution 320 whose sugar content is to be calculated on the vertical axis. Here, the numerical values ​​of the 80 sets of measurement data are omitted, but referring to the graph in FIG. 41, the estimated 80 sugar contents are shown as gray rectangles that are raw data. In FIG. 41, the sugar contents of the 80 calibration curve calculations are shown for each of the seven points of actual measurement sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. This is because the seven types of aqueous solutions 320 with sugar contents at the above seven points are aqueous solutions 320 for which sugar contents are to be calculated, and the calibration curve calculation sugar contents are calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 41 with the actual sugar contents as the objective variable yh and the calibration curve calculation sugar contents as the explanatory variable x, the following regression equation (55) is obtained. yh=x-5×10 -14 (55) yh is shown in FIG. 41 as yh (raw data) by a solid gray line, and the partial regression coefficient of the regression equation (55) is calculated by the analysis unit 30 using the well-known least squares method based on the sugar content of 80×7=560 estimated above. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.9834 (56) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the sugar content of the aqueous solution 320 whose sugar content is to be determined. Additionally, Figure 42 shows a graph in which the error in sugar content at 560 calculated from the standard curve for the case shown in Figure 41 is plotted on a graph with the error between the sugar content at 560 calculated from the standard curve and the sugar content of the actual measured value on the vertical axis and the sugar content of the actual measured value on the horizontal axis. Referring to Figure 42, the error is shown as a grey square representing the raw data, and it can be seen that the error is within approximately ±4.0%. Therefore, referring to Figs. 41 and 42, it is evident that the calibration curve shown in the above formula (54) is highly reliable.

[0092] <Estimation of sugar content of aqueous solution using the electromagnetic wave sensing device according to the fifth embodiment of the present invention, Part 2> In step 2 of estimating the sugar content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22, which are the change amount data of the seven types of sugar content at predetermined intervals obtained in step 1 of estimating the sugar content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, are calculated by the analysis unit 30 and stored in the memory 31. Here, if the average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22 are Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, the average value data of the set of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A is stored in the memory 31. In this case, the same measurement is performed four times for each container 300 containing aqueous solutions 320 of seven different sugar contents, and 4×7=28 sets of average value data for Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in memory 31. Here, when aqueous solutions 320 having seven different sugar contents, for example, approximately 0%, approximately 5%, approximately 10%, approximately 15%, approximately 20%, approximately 25%, and approximately 30%, are prepared, a graph is shown in Figure 35 that plots data on the average value (Re12A) of the change in the real part of the transmitted wave stored in the average value folder in memory 31 of analysis unit 30.

[0093] In the graph of FIG. 35, the horizontal axis is the average value of the change in the real part of the transmitted wave (Re12A), and the vertical axis is the sugar content [%]. The graph of FIG. 35 is a graph in the case where seven types of aqueous solutions with different sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30% are prepared. Referring to FIG. 35, Re12A calculated from 20 Re12s measured at a predetermined interval for each container 300 containing seven types of aqueous solutions 320 with different sugar contents is shown by black circles. In this case, four Re12A are calculated for each aqueous solution 320, so the Re12A calculated for the aqueous solutions 320 with seven types of sugar content is 28. The 28 Re12A are plotted on a line of sugar content corresponding to the sugar content of the measured aqueous solution 320, and four are displayed at almost the same position on the line. The four Re12A values ​​displayed are the first through fourth Re12A values ​​measured using the same aqueous solution 320, and it can be seen that they are grouped together without any variation. If a regression analysis is performed in the analysis unit 30 with the objective variable y being sugar content and the explanatory variable x being Re12A, the following regression equation (57) is obtained. y=87.983x+13.613 (57) The partial regression coefficient of the regression equation (57) is calculated by the analysis unit 30 using the well-known least squares method based on the Re12A data of 28. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.6964 (58) It becomes.

[0094] In addition, in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, when seven different types of aqueous solutions 320 with sugar contents ranging from 0% to 30%, for example, sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, are prepared, data of the average value (Im12A) of the change in the imaginary part of the transmitted wave stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 36. In the graph of FIG. 36, the horizontal axis is the average value (Im12A) of the change in the imaginary part of the transmitted wave, and the vertical axis is the sugar content [%]. The graph of FIG. 36 is a graph when seven types of aqueous solutions with different sugar contents ranging from 0% to 30% are prepared. Referring to FIG. 36, Im12A calculated from 20 Im12s measured in the first to fourth measurements at a predetermined interval for each container 300 containing seven types of aqueous solutions 320 with different sugar contents ranging from 0% to 30% are shown as black circles. In this case, Im12A is calculated four times for each aqueous solution 320, so Im12A calculated for the aqueous solutions 320 with seven types of sugar contents is 28. These 28 Im12A are plotted on a line of sugar contents corresponding to the sugar contents of the measured aqueous solutions 320, and three Im12A are displayed at almost the same positions on the line. The four Im12A displayed are the first to fourth Im12A measured for the same aqueous solution 320, and it can be seen that they are uniform without variation. Here, when regression analysis is performed in the analysis unit 30 with the objective variable y as sugar content and the explanatory variable x as Im12A, the regression equation (59) shown below is obtained. y=-72.354x+53.322 (59) The partial regression coefficient of the regression equation (59) is calculated by the analysis unit 30 using the well-known least squares method based on the data of Im12A in 28 above. The coefficient of determination R 2 teeth, R 2 =0.7799 (60) It becomes.

[0095] Next, in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, when seven different types of aqueous solutions 320 with sugar contents ranging from 0% to 30%, for example, sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, are prepared, a graph is shown in FIG. 37 plotting data of the average value (Re11A) of the change in the real part of the reflected wave A stored in the average value folder of the memory 31 of the analysis unit 30. The reflected wave A is a reflected wave that is an electromagnetic wave irradiated from the first antenna 11b and reflected by the container 300 containing the aqueous solution 320 in the electromagnetic wave sensing unit 10. In the graph of FIG. 37, the horizontal axis is the average value (Re11A) of the change in the real part of the reflected wave A, and the vertical axis is the sugar content [%]. The graph of FIG. 37 is a graph when seven types of aqueous solutions with different sugar contents ranging from 0% to 30% are prepared. Referring to FIG. 37, Re11A calculated from 20 Re11s measured at a predetermined interval for each container 300 containing seven types of aqueous solutions 320 with different sugar contents ranging from 0% to 30% is shown by black circles. In this case, Re11A is calculated four times for each aqueous solution 320, so Re11A calculated for the aqueous solutions 320 with seven types of sugar contents is 28. These 28 Re11A are plotted on a line of sugar content corresponding to the sugar content of the measured aqueous solution 320, and are displayed in groups of four at approximately the same position on the line. The Re11A displayed in groups of four are the Re11A from the first to fourth measurements of the same aqueous solution 320, and it can be seen that they are uniform without variation. Here, when regression analysis is performed in the analysis unit 30 with the objective variable y as sugar content and the explanatory variable x as Re11A, the following regression equation (61) is obtained. y=-61.907x-1.2089 (61) The partial regression coefficient of the regression equation (61) is calculated by the analysis unit 30 using the well-known least squares method based on the Re11A data of 28. The coefficient of determination R 2 teeth, R 2 =0.592 (62) It becomes.

[0096] In addition, in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, when seven different types of aqueous solutions 320 with sugar contents in the range of 0% to 30%, for example, about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, are prepared, the data of the average value (Im11A) of the change in the imaginary part of the reflected wave A stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 38. In the graph of FIG. 38, the horizontal axis is the average value (Im11A) of the change in the imaginary part of the reflected wave A, and the vertical axis is the sugar content [%]. The graph of FIG. 38 is a graph in the case where seven types of aqueous solutions with sugar contents in the range of 0% to 30% are prepared. Referring to FIG. 38, Im11A calculated from 20 Im11s measured in the first to fourth measurements at predetermined intervals for each container 300 containing seven types of aqueous solutions 320 with sugar contents in the range of 0% to 30% are shown by black circles. In this case, Im11A is calculated four times for each aqueous solution 320, so the Im11A calculated for the aqueous solutions 320 with seven types of sugar content is 28. These 28 Im11A are plotted on a line of sugar content corresponding to the sugar content of the measured aqueous solution 320, and are displayed in groups of four at approximately the same position on the line. The Im11A displayed in groups of four are the Im11A from the first to fourth measurements taken with the same aqueous solution 320, and it can be seen that they are grouped together without any variation. Here, if a regression analysis is performed in the analysis unit 30 with the objective variable y as sugar content and the explanatory variable x as Im11A, the following regression equation (63) is obtained. y=32.05x+32.923 (63) The partial regression coefficient of the regression equation (63) is calculated by the analysis unit 30 using the known least squares method based on the data of Im11A in 28 above. The coefficient of determination R 2 teeth, R 2 =0.6953 (64) It becomes.

[0097] Next, in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, when seven different types of aqueous solutions 320 with sugar contents ranging from 0% to 30%, for example, about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, are prepared, a graph is shown in FIG. 39, which plots data on the average value (Re22A) of the change in the real part of the reflected wave B stored in the average value folder of the memory 31 of the analysis unit 30. The reflected wave B is a reflected wave that is an electromagnetic wave irradiated from the second antenna 11c and reflected by the container 300 containing the aqueous solution 320 in the electromagnetic wave sensing unit 10. In the graph of FIG. 39, the horizontal axis is the average value (Re22A) of the change in the real part of the reflected wave B, and the vertical axis is the sugar content [%]. The graph of FIG. 39 is a graph when seven types of aqueous solutions with different sugar contents ranging from 0% to 30% are prepared. Referring to FIG. 39, Re22A calculated from 20 Re22s measured at a predetermined interval for each container 300 containing seven types of aqueous solutions 320 with different sugar contents ranging from 0% to 30% is shown by black circles. In this case, Re22A is calculated four times for each aqueous solution 320, so Re22A calculated for the aqueous solutions 320 with seven types of sugar contents is 28. These 28 Re22A are plotted on a line of sugar contents corresponding to the sugar contents of the measured aqueous solutions 320, and four are displayed at almost the same positions on the line. The four displayed Re22A are the first to fourth Re22A measured for the same aqueous solution 320, and it can be seen that they are uniform without variation. Here, when regression analysis is performed in the analysis unit 30 with the objective variable y as sugar content and the explanatory variable x as Re22A, the following regression equation (65) is obtained. y=-136.77x+83.075 (65) The partial regression coefficient of the regression equation (65) is calculated by the analysis unit 30 using the well-known least squares method based on the data of Re22A in 21 above. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.2102 (66) It becomes.

[0098] In addition, in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, when seven different aqueous solutions 320 with sugar contents ranging from 0% to 30%, for example, sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, are prepared, data of the average value (Im22A) of the change in the imaginary part of the reflected wave B stored in the memory 31 of the analysis unit 30 is graphed and shown in FIG. 40. In the graph of FIG. 40, the horizontal axis is the average value (Im22A) of the change in the imaginary part of the reflected wave B, and the vertical axis is the sugar content [%]. The graph of FIG. 40 is a graph when seven different aqueous solutions with sugar contents ranging from 0% to 30% are prepared. Referring to FIG. 40, Im22A calculated from 20 Im22s measured in the first to fourth measurements at a predetermined interval for each container 300 containing seven types of aqueous solutions 320 with different sugar contents ranging from 0% to 30% is shown as black circles. In this case, Im22A is calculated four times for each aqueous solution 320, so Im22A calculated for the aqueous solutions 320 with seven types of sugar contents is 28. These 28 Im22A are plotted on a line of sugar content corresponding to the sugar content of the measured aqueous solution 320, and four Im22A are displayed at almost the same position on the line. The four Im22A displayed are the first to fourth Im22A measured for the same aqueous solution 320, and it can be seen that they are uniform without variation. Here, if regression analysis is performed in the analysis unit 30 with the objective variable y as sugar content and the explanatory variable x as Im22A, the following regression equation (67) is obtained. y=452.7x+189.72 (67) The partial regression coefficient of the regression equation (67) is calculated by the analysis unit 30 using the well-known least squares method based on the data of Im22A in 21 above. The coefficient of determination of this regression equation R 2 teeth, R 2 =0.4345 (68) It becomes.

[0099] As described above, when the aqueous solutions 320 of the fifth embodiment with sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30% are prepared, data sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, which are average values ​​of the amount of change at a predetermined interval in the container 300 containing seven types of aqueous solutions 320 with different sugar contents, are stored in the memory 31. Note that the above-mentioned measurement is repeated four times with the same aqueous solution 320, so a total of 28 sets of data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored. Here, the data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are the average values ​​of the data of the amount of change Re12, Im12, Re11, Im11, Re22, and Im22. Therefore, if the sugar content of the aqueous solution 320 is set as the objective variable y and the six parameters of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are used as explanatory variables to perform multiple regression analysis in the analysis unit 30, a multiple regression equation of the calibration curve can be obtained. The multiple regression equation of the calibration curve is the same as the above-mentioned equation (52). If the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (52) are calculated by the analysis unit 30 using the well-known least squares method based on the above 28 sets of data, the multiple regression equation shown in equation (69) can be obtained. y=26.03+35.32x1-11.62x2-2.62x3+31.52x4+106.82x5+104.46x6(69) In equation (69), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, and x6 is the value of Im22A. y in equation (69) is the multiple regression equation for the calibration curve calculated from the average value of the change amount data.

[0100] When it is desired to estimate the sugar content of an aqueous solution 320 for which the sugar content is to be determined, similarly to step 1 of estimating the sugar content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the container 300 containing the aqueous solution 320 for which the sugar content is to be determined is fed from the feed unit 12, and measurements are taken every time the container 300 containing the aqueous solution 320 moves a predetermined distance, and the measured data of the amount of change in Re12', Im12', Re11', Im11', Re22', and Im22' is stored in the memory 31. Re12', Im12', Re11', Im11', Re22', and Im22' are regarded as data measured every time the container 300 moves a predetermined distance, and in this case, the number of predetermined intervals is 20 as above, and 20 sets of data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. That is, the 20 sets of data are data sets of Re12', Im12', Re11', Im11', Re22', and Im22' acquired at 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 whose sugar content is to be determined are stored in the memory 31.

[0101] Next, the analysis unit 30 calculates the average value of the data of the amount of change at a predetermined interval for each container 300 containing the aqueous solution 320 whose sugar content is to be determined, and stores the calculated average data of the amount of change, Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', is stored in the memory 31. In this case, the same measurement is performed four times for the container 300 containing the aqueous solution 320 whose sugar content is to be determined, so that the data of the average values ​​of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' is stored in the memory 31. The sugar content of the solution 310 whose sugar content is to be determined can be estimated by substituting the four sets of average data of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average data of Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose sugar content is to be determined moves a predetermined distance, into the above formula (69) which is the formula of the calibration curve. When the above four sets of data are respectively substituted into formula (69), the number of estimated sugar contents becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated sugar contents as the estimated sugar content of the aqueous solution 320 whose sugar content is to be determined.

[0102] Here, the graph shown in Fig. 41 is created so that the variation in the four sugar contents estimated by the calibration curve of the multiple regression equation shown in Equation (69) can be visualized. The graph shown in Fig. 41 plots the above four sugar contents with the sugar content [%] calculated by the calibration curve on the horizontal axis and the sugar content [%] of the actual measured value of the aqueous solution 320 whose sugar content is to be calculated on the vertical axis. Here, the values ​​of the four sets of measurement data are omitted, but the four estimated sugar contents are shown as white circles that are averaged data. Referring to FIG. 41, the sugar contents of the four calibration curve calculations are shown for seven points of actual measurement sugar contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. This is because the seven types of aqueous solutions 320 with sugar contents at the above seven points are aqueous solutions 320 for which sugar contents are to be calculated, and the case shown is one in which sugar contents of the calibration curve calculations are calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 41 with the actual sugar contents as the objective variable yi and the sugar contents of the calibration curve calculation x as the explanatory variable, the following regression equation (70) is obtained. yi=0.9968x+0.0858 (70) yi is shown in FIG. 41 as a straight dashed line, yi (averaged data), and the partial regression coefficient of the regression equation (70) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated 4×7=28 sugar content values. The coefficient of determination of this regression equation, R 2 teeth, R 2=0.9965 (71) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the sugar content of the aqueous solution 320 whose sugar content is to be determined. Additionally, Figure 42 shows a graph in which the error between the four sugar contents calculated from the standard curve and the actual sugar contents is plotted on the vertical axis, and the actual sugar contents are plotted on the horizontal axis, plotting the 4 x 7 = 28 sugar contents error calculated from the standard curve. Referring to Figure 42, the error is shown as an open circle representing averaged data, and it can be seen that the error is between approximately -1.2% and approximately 1.0%. Therefore, referring to FIGS. 41 and 42, it is clear that the calibration curve calculated from the average value of the change amount data shown in the above formula (69) is more reliable.

[0103] <Estimation of sugar content of aqueous solution using the electromagnetic wave sensing device according to the fifth embodiment of the present invention, 3> In step 3 of estimating the sugar content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, in addition to the average data of the amount of change acquired in step 2 of estimating the sugar content in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average data of the VSWR is added to the explanatory variables and multiple regression analysis is performed in the analysis unit 30. In this case, in the electromagnetic wave sensing device 5 of the fifth embodiment, in addition to the amount of change data, VSWR data is measured and stored in the memory 31. Then, the analysis unit 30 calculates the average data of the amount of change data and the VSWR data stored in the memory 31, and the average data of the amount of change data and the average data of the VSWR are stored in the memory 31. Specifically, the distance between the first antenna 11b and the container 300 containing the aqueous solution 320 is set to, for example, about 546.1 mm, and an electromagnetic wave of 3578 MHz is irradiated from the first antenna 11b side to measure data (Re12, Im12, Re11, Im11) of the change in the transmitted wave and the reflected wave at every predetermined interval of the container 300 containing the solution 310, while VSWR data VSWR1 is also measured. In addition, the distance between the second antenna 11c and the container 300 containing the aqueous solution 320 is set to, for example, about 645.5 mm, and an electromagnetic wave of 3578 MHz is irradiated from the second antenna 11c side to measure data (Re22, Im22) of the change in the reflected wave at every predetermined interval of the container 300 containing the aqueous solution 320, while VSWR data VSWR2 is also measured. Next, the average value of the change amount data measured at predetermined intervals is calculated by the analysis unit 30, and the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A mentioned above, the average value data VSWR1A of VSWR1, and the average value data VSWR2A of VSWR2 are stored in the memory 31. In this case, seven types of aqueous solutions 320 of known sugar content are measured four times, so a total of 4×7=28 sets of average value data for the groups Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are measured and stored in memory 31.

[0104] Here, in the configuration shown in FIG. 34(b), the distance between the first antenna 11b and the container 300 containing the aqueous solution 320 is set to, for example, 546.1 mm, and data VSWR1A of the average value of VSWR1 measured from the first antenna 11b side is shown in the graph of FIG. 43. In the graph of FIG. 43, the horizontal axis is the average value of VSWR1 (VSWR1A), and the vertical axis is the sugar content [%]. The graph of FIG. 43 is a graph in the case where aqueous solutions 320 with seven types of sugar content, approximately 0%, approximately 5%, approximately 10%, approximately 15%, approximately 20%, approximately 25%, and approximately 30%, are prepared. Referring to FIG. 43, VSWR1A is shown as data of the average value of VSWR1 measured for aqueous solutions 320 with seven types of sugar content, approximately 0%, approximately 5%, approximately 10%, approximately 15%, approximately 20%, approximately 25%, and approximately 30%, and four VSWR1A measured repeatedly four times are displayed at approximately the same position. In this example, since aqueous solutions 320 with seven different sugar contents are prepared, the total number of VSWR1A displayed in Fig. 43 is 28. Then, by performing regression analysis in the analysis unit 30 with the sugar content as the objective variable y and the VSWR1A as the explanatory variable x, the following regression equation (72) is obtained. y=-595.28x+656.96 (72) The partial regression coefficient of the regression equation (72) is calculated by the analysis unit 30 using the well-known least squares method based on the VSWR1A data of 28. The coefficient of determination R 2 teeth, R 2 =0.9985 (73) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is high. One reason for this is that the distance (546.1 mm) was selected at which the correlation between the distance between the first antenna 11b and the container 300 containing the aqueous solution 320 and the sugar content of the aqueous solution 320 is high. In this case, the distance at which the correlation is high is measured in advance.

[0105] Next, in the configuration shown in FIG. 34(b), the distance between the second antenna 11c and the container 300 containing the aqueous solution 320 is set to, for example, 645.5 mm, and data VSWR2A of the average value of VSWR2 measured from the second antenna 11c side is shown in the graph of FIG. 44. In the graph of FIG. 44, the horizontal axis is the average value of VSWR2 (VSWR2A), and the vertical axis is the sugar content [%]. The graph of FIG. 44 is a graph in the case where aqueous solutions 320 with seven types of sugar content, approximately 0%, approximately 5%, approximately 10%, approximately 15%, approximately 20%, approximately 25%, and approximately 30%, are prepared. Referring to FIG. 44, VSWR2A is shown as data of the average value of VSWR2 measured for aqueous solutions 320 with seven types of sugar content, approximately 0%, approximately 5%, approximately 10%, approximately 15%, approximately 20%, approximately 25%, and approximately 30%, and four VSWR2A measured repeatedly four times are displayed at approximately the same position. In this example, since aqueous solutions 320 with seven different sugar contents are prepared, the total number of VSWR2A values ​​displayed in Fig. 44 is 28. Then, by performing regression analysis in the analysis unit 30 with the sugar content as the objective variable y and the VSWR2A as the explanatory variable x, the following regression equation (74) is obtained. y=-504.4x+578.28 (74) The partial regression coefficient of the regression equation (72) is calculated by the analysis unit 30 using the well-known least squares method based on the VSWR1A data of 28. The coefficient of determination R 2 teeth, R 2 =0.9987 (75) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is high. One reason for this is that the distance (645.5 mm) was selected at which the correlation between the distance between the second antenna 11c and the container 300 containing the aqueous solution 320 and the sugar content of the aqueous solution 320 is high. In this case, the distance at which the correlation is high is measured in advance.

[0106] As described above, eight parameters, namely, average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, average value data VSWR1A of VSWR1, and average value data VSWR2A of VSWR2A, are stored in memory 31. Since measurements of seven types of aqueous solutions 320 with known sugar contents are repeated four times, a total of 4×7=28 sets of average value data of the amount of change made up of the above eight parameters are stored in memory 31. Therefore, by performing multiple regression analysis in the analysis unit 30 with the sugar content y as the response variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables, it is possible to obtain a multiple regression equation for the calibration curve. This multiple regression equation is similar to the above-mentioned equation (53), and when the partial regression coefficients of the multiple regression equation for equation (53) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 28 sets of data, equation (76) is obtained. y=574.22+10.00x1-10.27x2+6.49x3-4.56x4+2.90x5+12.20x6-59.47x7-436.44x8(76) In equation (76), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, x6 is the value of Im22A, x7 is the value of VSWR1A, and x8 is the value of VSWR2A. y in equation (76) is the multiple regression equation for the calibration curve calculated from the average value of the change data including VSWR.

[0107] When it is desired to estimate the sugar content of the aqueous solution 320 for which the sugar content is to be determined, the same measurement is performed as for the seven types of aqueous solutions 320 with known sugar contents described above, and data on the amount of change at a predetermined interval is measured for the container 300 containing the aqueous solution 320 for which the sugar content is to be determined. In the above measurement, the frequency of the electromagnetic waves emitted from the first antenna 11b or the second antenna 11c is set to 3578 MHz. Next, the data Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data on the amount of change at a predetermined interval for the aqueous solution 320 for which the sugar content is to be determined, are calculated by the analysis unit 30, and the average data of the calculated amount of change is stored in the memory 31. Here, since the measurement is repeated four times, data on the average values ​​of the four sets of amount of change is stored in the memory 31. The sugar content of the aqueous solution 320 whose sugar content is to be determined can be estimated by substituting the above-mentioned four sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data of the amount of change at a predetermined interval of the aqueous solution 320 whose sugar content is to be determined, into the above-mentioned formula (76), which is the formula of the calibration curve. When the above-mentioned four sets of data are respectively substituted into formula (76), the number of estimated sugar contents becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated sugar contents as the estimated sugar content of the aqueous solution 320 whose sugar content is to be determined.

[0108] Here, the graph shown in Fig. 45 is created so that the variation in the four sugar contents estimated by the calibration curve of the multiple regression equation shown in Equation (76) can be visualized. The graph shown in Fig. 45 plots the above four sugar contents with the sugar content [%] calculated by the calibration curve on the horizontal axis and the sugar content [%] of the actual measured value of the aqueous solution 320 whose sugar content is to be calculated on the vertical axis. Here, the values ​​of the four sets of measurement data are omitted, but the estimated sugar content is shown as four white circles that are averaged data. In Fig. 45, the sugar content of the actual measurement value is shown for seven points of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%, and four sugar contents of the calibration curve are shown for each of these seven points. This is because the seven types of aqueous solutions 320 with the sugar content of the seven points are the aqueous solutions 320 for which the sugar content is to be calculated, and the sugar content of the calibration curve is calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in Fig. 45 with the actual sugar content as the objective variable yj and the sugar content of the calibration curve as the explanatory variable x, the following regression equation (77) is obtained. yj=x-0.00000 (77) yj is shown in FIG. 45 as a straight dashed line yj (with VSWR), and the partial regression coefficient of the regression equation yj in Eq. (77) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated 4×7=28 sugar contents. The coefficient of determination R 2 teeth, R 2 =0.99998 (78) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is higher. In other words, it can be seen that the sugar content of the aqueous solution 320 whose sugar content is to be determined can be estimated more quantitatively. The gray rectangular plot shown as yi (without VSWR) in FIG. 45 is the same as the plot shown by the white circle in FIG. 41. Here, when the regression equation yj is compared with the regression equation yi, it can be seen that the reliability of the calibration curve is higher when the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are used as explanatory variables than when the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are used as explanatory variables. FIG. 46 shows a graph in which the error between the four sugar contents calculated from the calibration curve and the actual sugar contents shown by the dashed lines in FIG. 45 is plotted on a graph with the vertical axis representing the error between the four sugar contents calculated from the calibration curve and the actual sugar contents shown by the horizontal axis representing the sugar contents of the actual measurements. In FIG. 46, the error is shown by four open circles with VSWR, and it can be seen that the error is within about ±1.0%. Note that the gray rectangular plot shown as "no VSWR" in FIG. 46 is the same as the open circle plot in FIG. 42. 45 and 46, it can be seen that the calibration curve shown in the above formula (76), which is created when VSWR1A and VSWRA2 are also added as explanatory variables, is more reliable than the calibration curve shown in the above formula (69). From the above, it can be said that the reliability of the calibration curve created using sugar content as the response variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables is very high.

[0109] <Estimation of Salt Concentration of an Aqueous Solution Using an Electromagnetic Wave Sensing Device According to a Fifth Embodiment of the Present Invention, Part 1> Next, a case will be described in which the concentration of the solute in the solution 310 is taken as the salinity of the aqueous solution 320 in the electromagnetic wave sensing device 5 according to the fifth embodiment of the present invention, and the salinity of the aqueous solution 320 is estimated. In this case, aqueous solutions 320 having different salinities in the range of 0% to 25%, for example, six salinities of about 0%, about 5%, about 10%, about 15%, about 20%, and about 25%, are prepared. The above-mentioned measurement is performed with the frequency of the electromagnetic wave emitted from the first antenna 11b or the second antenna 11c set to 1176 MHz. By performing the above-mentioned measurement, data of the change amount at each predetermined interval of the container 300 containing the aqueous solutions 320 of six salinities, that is, Re12, Im12, Re11, Im11, Re22, and Im22, is measured for each type of salinity and stored in the memory 31. The number of predetermined intervals of the container 300 where the change amount data is measured is, for example, 20, and the change amount data having a set of Re12, Im12, Re11, Im11, Re22, and Im22, which is the same number as the number of predetermined intervals, is obtained as 20 sets. Next, the above-mentioned operation is performed on other aqueous solutions 320 with different salt concentrations to obtain 20 sets of change amount data, each set consisting of Re12, Im12, Re11, Im11, Re22, and Im22, for the other aqueous solutions 320 with different salt concentrations. By repeating this process, 20 sets of change amount data, each set consisting of Re12, Im12, Re11, Im11, Re22, and Im22, for each of the six other aqueous solutions 320 with different salt concentrations prepared are stored in the memory 31. Since the above-mentioned measurement is repeated four times for the same aqueous solution 320, 20×4=80 sets of change amount data are measured for each aqueous solution 320. Here, the number of aqueous solutions 320 with different salt concentrations prepared is six, but it may be six or more, and if six types are used, 80 sets of change amount data are measured for each type, so that 80×6=480 sets of change amount data are measured and stored in the memory 31.

[0110] Next, by performing multiple regression analysis in the analysis unit 30 with the salinity as the objective variable y and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 as explanatory variables, the multiple regression equation of the calibration curve can be obtained. This multiple regression equation is similar to the above-mentioned equation (51). Here, the numerical values ​​of the above-mentioned 480 sets of measurement data are omitted, but if the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (51) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 80 x 6 = 480 sets of data, the multiple regression equation shown in equation (79) can be obtained. y=-387.37-476.68x1-282.12x2+14.84x3+11.87x4-5.81x5-22.96x6(79) In equation (79), x1 is the value of Re12, x2 is the value of Im12, x3 is the value of Re11, x4 is the value of Im11, x5 is the value of Re22, and x6 is the value of Im22. y in equation (79) is the multiple regression equation for the calibration curve calculated from the change amount data.

[0111] When it is desired to estimate the salinity of an aqueous solution 320, the container 300 containing the aqueous solution 320 whose salinity is to be determined is fed from the feed unit 12, and an electromagnetic wave of, for example, 1176 MHz is supplied from the measurement unit 20 to the first antenna 11b via the first coaxial terminal 14a every time the container 300 containing the aqueous solution 320 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave of the container 300 containing the aqueous solution 320 whose salinity is to be determined are stored in the memory 31. Furthermore, the first antenna 11b receives a reflected wave of, for example, an 1176 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the container 300 containing the aqueous solution 320 whose salinity is to be determined. Furthermore, the second antenna 11c receives a reflected wave of, for example, an 1176 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the container 300 containing the aqueous solution 320 whose salinity is to be determined. Re12', Im12', Re11', Im11', Re22', and Im22' are data measured every time the sensor moves at a predetermined interval. In this case, the number of predetermined intervals is 20, as described above, and 20 sets of data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. That is, the 20 sets of data are data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' obtained at 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 whose salinity is to be calculated are stored in the memory 31. Here, Re12', Im12', Re11', Im11', Re22', and Im22' are data on the amount of change, which is the amount of change from the initial setting data measured in a state where the container 300 is not inserted in the electromagnetic wave sensing unit 10.

[0112] The above-mentioned 80 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose salinity is to be determined moves a predetermined distance were substituted into the above formula (79) which is the formula of the calibration curve, to estimate the salinity of the aqueous solution 320 whose salinity is to be determined. The salinity estimated by substituting the 80 sets of data is 80. Then, the analysis unit 30 can calculate and present the average value of the 80 salinities estimated as the estimated salinity of the aqueous solution 320 whose salinity is to be determined. Alternatively, the range of the average value of the 80 salinities ± the standard deviation may be presented. Here, the graph shown in Fig. 47 is created so that the variation in the salinity of 80 estimated by the calibration curve of the multiple regression equation shown in Equation (79) can be visualized. The graph shown in Fig. 47 plots the salinity of 80 with the salinity [%] calculated by the calibration curve on the horizontal axis and the salinity [%] of the actual measured value of aqueous solution 320 whose salinity is to be determined on the vertical axis. Here, the numerical values ​​of the 80 sets of measurement data are omitted, but in FIG. 47, the estimated salinity is shown as a gray rectangle that is treated as raw data. Referring to FIG. 47, 80 salinities of the calibration curve calculation are shown for each of six points of actual measurement salinity values ​​of about 0%, about 5%, about 10%, about 15%, about 20%, and about 25%. This is because the six types of aqueous solutions 320 with the salinities of the above six points are aqueous solutions 320 whose salinities are to be obtained, and the salinity of the calibration curve calculation is obtained from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 47 with the actual salinity as the response variable yk and the salinity of the calibration curve calculation x as the explanatory variable, the following regression equation (80) is obtained. yk=x-2×10 -13 (80) yk is shown in FIG. 47 as yk (raw data) by a solid gray line, and the partial regression coefficient of the regression equation (80) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated salinity of 80×6=480. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.973 (81) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the salinity of the aqueous solution 320, the salinity of which is to be determined. Fig. 48 shows a graph in which the error in the salinity at 480 calculated from the calibration curve for the case shown in Fig. 47 is plotted on a graph with the error between the salinity at 480 calculated from the calibration curve and the actual measured salinity on the y-axis and the actual measured salinity on the x-axis. Referring to Fig. 48, the error is shown as a grey rectangle representing raw data, and it can be seen that the error is between about -6.0% and about 7.0%. Therefore, referring to Figs. 47 and 48, it is evident that the calibration curve shown in the above formula (79) is highly reliable.

[0113] <Estimation of Salt Concentration of an Aqueous Solution Using an Electromagnetic Wave Sensing Device of a Fifth Example of the Present Invention, Part 2> In step 2 of estimating the salinity of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the analysis unit 30 calculates average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22, which are data of the change amount at predetermined intervals of six types of salinity obtained in step 1 of estimating the salinity of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, and stores the average values ​​in the memory 31. Here, if the average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22 are Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, the average value data of the sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A is stored in the memory 31. In this case, the same measurement is performed four times for each container 300 containing aqueous solutions 320 of six different salt concentrations, so that 4 x 6 = 24 sets of average value data for Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in memory 31.

[0114] Here, the salinity of the aqueous solution 320 is set as the objective variable y, and the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are set as explanatory variables, and the multiple regression equation of the calibration curve is obtained by performing multiple regression analysis in the analysis unit 30. The multiple regression equation of the calibration curve is the same as the above-mentioned equation (52). Here, the numerical values ​​of the above-mentioned 24 sets of measurement data are omitted, but when the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (52) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 24 sets of data, the multiple regression equation shown in equation (82) is obtained. y=-172.41+5.24x1+162.50x2+16.16x3+10.94x4-9.83x5-46.98x6(82) In equation (82), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, and x6 is the value of Im22A. y in equation (82) is the multiple regression equation for the calibration curve calculated from the average value of the change amount data.

[0115] When it is desired to estimate the salinity of an aqueous solution 320, the salinity of which is to be determined is estimated in the same manner as in the estimation 1 of the salinity of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the container 300 containing the aqueous solution 320 containing the salinity is fed from the feed unit 12, and data on the amount of change in Re12', Im12', Re11', Im11', Re22', and Im22' is measured every time the container 300 containing the aqueous solution 320 moves a predetermined distance, and the measured data on the amount of change is stored in the memory 31. Re12', Im12', Re11', Im11', Re22', and Im22' are regarded as data measured every time the container 300 moves a predetermined distance, and in this case, the number of predetermined intervals is 20 as above, and 20 sets of data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. That is, the 20 sets of data are data sets of Re12', Im12', Re11', Im11', Re22', and Im22' acquired at 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 whose salinity is to be determined are stored in the memory 31.

[0116] Next, the analysis unit 30 calculates the average value of the data of the amount of change at a predetermined interval for each container 300 containing the aqueous solution 320 whose salinity is to be determined, and stores the calculated average values ​​of the data of the amount of change at a predetermined interval for the container 300 containing the aqueous solution 320 whose salinity is to be determined, and stores the data of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' in the memory 31. In this case, the same measurement is performed four times for the container 300 containing the aqueous solution 320 whose salinity is to be determined, and thus the data of the average values ​​of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' is stored in the memory 31. The salinity of the solution 310 whose salinity is to be determined can be estimated by substituting the average data of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average data of Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose salinity is to be determined moves at a predetermined distance, into the above formula (82) which is the formula of the calibration curve. When the above four sets of data are respectively substituted into the formula (82), the number of estimated salinities becomes four. Then, the average value of the four estimated salinities can be calculated and presented by the analysis unit 30 as the estimated salinity of the aqueous solution 320 whose salinity is to be determined.

[0117] Here, the graph shown in Fig. 47 is created so that the variation in the four salinity concentrations estimated by the calibration curve of the multiple regression equation shown in Equation (82) can be visualized. The graph shown in Fig. 47 plots the above 80 salinity concentrations with the salinity [%] calculated by the calibration curve on the horizontal axis and the salinity [%] of the actual measured value of aqueous solution 320 whose salinity is to be determined on the vertical axis. Here, the values ​​of the four sets of measurement data are omitted, but the four estimated salinities are shown as white circles that are averaged data. Referring to FIG. 47, the salinities of the four calibration curve calculations are shown for six points of actual measurement salinity of about 0%, about 5%, about 10%, about 15%, about 20%, and about 25%. This is because the six types of aqueous solutions 320 with the salinities of the above six points are aqueous solutions 320 whose salinities are to be obtained, and the salinities of the calibration curve calculations are obtained from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 47 with the actual salinity as the response variable ym and the salinity of the calibration curve calculation x as the explanatory variable, the following regression equation (83) is obtained. ym=1x+1×10 -7 (83) ym is shown in FIG. 47 as a straight dashed line, ym (averaged data), and the partial regression coefficient of the regression equation (83) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated 4×6=24 salinity concentrations. The coefficient of determination R 2 teeth, R 2 =0.9972 (84) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the salinity of the aqueous solution 320, the salinity of which is to be determined. Moreover, Figure 48 shows a graph in which the error in the salinity of 24 calculated from the calibration curve for the case shown in Figure 47 is plotted on a graph with the error between the salinity of 24 calculated from the calibration curve and the actual measured salinity on the y-axis and the actual measured salinity on the x-axis. Referring to Figure 48, the error is shown as an outlined circle representing averaged data, and it can be seen that the error is within approximately ±1.0%. Therefore, with reference to FIGS. 47 and 48, it is clear that the calibration curve calculated from the average value of the change amount data shown in the above formula (82) is more reliable.

[0118] <Estimation of Salt Concentration of Aqueous Solution Using Electromagnetic Wave Sensing Device of Fifth Example of the Present Invention, Part 3> In step 3 of estimating the salinity of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, in addition to the average data of the amount of change acquired in step 2 of estimating the salinity of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average data of the VSWR is added to the explanatory variables and multiple regression analysis is performed in the analysis unit 30. In this case, in the electromagnetic wave sensing device 5 of the fifth embodiment, in addition to the amount of change data, VSWR data is measured and stored in the memory 31. Then, the analysis unit 30 calculates average data of the amount of change data and the VSWR data stored in the memory 31, and the average data of the amount of change data and the average data of the VSWR are stored in the memory 31. Specifically, the distance between the first antenna 11b and the aqueous solution 320 in the container 300 is set to, for example, about 546.1 mm, and an electromagnetic wave of 1176 MHz is irradiated from the first antenna 11b side to measure data (Re12, Im12, Re11, Im11) of the change in the transmitted wave and the reflected wave at every predetermined interval of the container 300 containing the solution 310, and VSWR data VSWR1 is also measured. In addition, the distance between the second antenna 11c and the aqueous solution 320 in the container 300 is set to, for example, about 645.5 mm, and an electromagnetic wave of 1176 MHz is irradiated from the second antenna 11c side to measure data (Re22, Im22) of the change in the reflected wave at every predetermined interval of the container 300 containing the aqueous solution 320, and VSWR data VSWR2 is also measured. Next, the average value of the change amount data measured at predetermined intervals is calculated by the analysis unit 30, and the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A mentioned above, the average value data VSWR1A of VSWR1, and the average value data VSWR2A of VSWR2 are stored in the memory 31. In this case, the same measurement is performed four times for each container 300 containing aqueous solutions 320 of six different salt concentrations, so that 4 x 6 = 24 sets of data, namely Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A, are stored in memory 31.

[0119] As described above, eight parameter sets of Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are stored in memory 31. Then, measurements of six types of aqueous solutions 320 having known salinity concentrations are repeated four times, and thus a total of 4×6=24 sets of average data of the amount of change made up of the above eight parameters are stored in memory 31. Therefore, by performing multiple regression analysis in the analysis unit 30 with the salinity concentration y as the response variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables, it is possible to obtain a multiple regression equation for the calibration curve. This multiple regression equation is similar to the above-mentioned equation (53), and although the numerical values ​​of the above-mentioned 24 sets of measurement data are omitted here, when the partial regression coefficients of the multiple regression equation of equation (53) are found by the analysis unit 30 based on the above-mentioned 24 sets of data using the well-known least squares method, equation (85) is obtained. y=68725.18+36.42x1-187.56x2+8.69x3+0.21x4-1.13x5-19.86x6-66379.38x7+394.19x8(85) In equation (85), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, x6 is the value of Im22A, x7 is the value of VSWR1A, and x8 is the value of VSWR2A. y in equation (85) is the multiple regression equation for the calibration curve calculated from the average value of the change data including VSWR.

[0120] When it is desired to estimate the salinity of the aqueous solution 320, the same measurement is performed as for the aqueous solutions 320 having the six known salinities described above, and data on the amount of change at each predetermined interval is measured for the container 300 containing the aqueous solution 320 whose salinity is desired to be determined. In the above measurement, the frequency of the electromagnetic wave emitted from the first antenna 11b or the second antenna 11c is set to 1176 MHz. Next, the analysis unit 30 calculates data on Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data on the amount of change at each predetermined interval for the aqueous solution 320 whose salinity is desired to be determined, and the data on the average values ​​of the calculated data on the amount of change is stored in the memory 31. Here, since the measurement is repeated four times, data on the average values ​​of four sets of data on the amount of change is stored in the memory 31. The salinity of the aqueous solution 320 whose salinity is to be determined can be estimated by substituting the above-mentioned four sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data of the amount of change at predetermined intervals of the aqueous solution 320 whose salinity is to be determined, into the above-mentioned formula (85), which is the formula of the calibration curve. When the above-mentioned four sets of data are respectively substituted into the formula (85), the number of estimated salinities becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated salinities as the estimated salinity of the aqueous solution 320 whose salinity is to be determined.

[0121] Here, the graph shown in Fig. 49 is created so that the variation in the four salinity concentrations estimated by the calibration curve of the multiple regression equation shown in Equation (85) can be visualized. The graph shown in Fig. 49 plots the above four salinity concentrations with the salinity [%] calculated by the calibration curve on the horizontal axis and the salinity [%] of the actual measured value of aqueous solution 320 whose salinity is to be determined on the vertical axis. Here, the values ​​of the four sets of measurement data are omitted, but the estimated salinity is shown as four white circles that are averaged data. In FIG. 49, the salinity of the actual measurement value is shown at six points of about 0%, about 5%, about 10%, about 15%, about 20%, and about 25%, and four salinities of the calibration curve calculation are shown for each of the six points. This is because the six types of aqueous solutions 320 with the salinities of the above six points are the aqueous solutions 320 whose salinities are to be obtained, and the salinity of the calibration curve calculation is obtained from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 49 with the actual salinity as the response variable yn and the salinity of the calibration curve calculation as the explanatory variable x, the following regression equation (86) is obtained. yn=1x+1×10 -11 (86) yn is shown in FIG. 49 as a straight dashed line yn (with VSWR), and the partial regression coefficient of the regression equation yn in Eq. (86) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated salinity of 4×6=24. The coefficient of determination R 2 teeth, R 2 =0.99994 (87) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is higher. In other words, it can be seen that the salinity of the aqueous solution 320 whose salinity is to be obtained can be estimated more quantitatively. The gray rectangular plot shown as ym (without VSWR) in FIG. 49 is the same as the plot shown by the white circle in FIG. 47. Here, when the regression equation yn is compared with the regression equation ym, it can be seen that the reliability of the calibration curve is higher when the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are used as explanatory variables than when the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are used as explanatory variables. FIG. 50 shows a graph in which the error of the 28 salinity concentrations calculated from the calibration curve is plotted on a graph with the error between the four salinity concentrations calculated from the calibration curve and the actual measured salinity on the vertical axis and the actual measured salinity on the horizontal axis, in the case shown by the dashed line in FIG. 49. In FIG. 50, the error is shown as four open circles with VSWR, and it can be seen that the error is within about ±0.5%. Note that the gray rectangular plot shown as "without VSWR" in FIG. 50 is the same as the plot shown as the open circle in FIG. 48. 49 and 50, it can be seen that the calibration curve shown in the above formula (85), which is created when VSWR1A and VSWRA2 are also added as explanatory variables, is more reliable than the calibration curve shown in the above formula (82). From the above, it can be said that the reliability of the calibration curve created with salinity as the response variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables is very high.

[0122] <Estimation of alcohol content of aqueous solution using electromagnetic wave sensing device according to the fifth embodiment of the present invention, 1> Next, a case will be described in which the concentration of the solute in the solution 310 is taken as the alcohol content of the aqueous solution 320 in the electromagnetic wave sensing device 5 according to the fifth embodiment of the present invention, and the alcohol content of the aqueous solution 320 is estimated. In this case, the aqueous solution 320 is prepared with a plurality of different alcohol contents ranging from 0% to 30%, for example, seven types of alcohol contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. The above-mentioned measurement is performed with the frequency of the electromagnetic wave emitted from the first antenna 11b or the second antenna 11c set to 748MHz. By performing the above-mentioned measurement, data of the change amount at each predetermined interval of the container 300 containing the aqueous solution 320 with the seven types of alcohol contents, Re12, Im12, Re11, Im11, Re22, and Im22, is measured for each type of alcohol content and stored in the memory 31. The number of predetermined intervals of the container 300 where the change amount data is measured is, for example, 20, and the change amount data with the set of Re12, Im12, Re11, Im11, Re22, and Im22 is obtained in the number of 20 sets, which is the same number as the number of predetermined intervals. Next, the above-mentioned operation is performed on other aqueous solutions 320 with different alcohol contents, and 20 sets of change amount data, each consisting of Re12, Im12, Re11, Im11, Re22, and Im22, are obtained for the other aqueous solutions 320 with different alcohol contents. By repeating this process, 20 sets of change amount data, each consisting of Re12, Im12, Re11, Im11, Re22, and Im22, for each of the seven types of other aqueous solutions 320 with different alcohol contents are stored in the memory 31. Since the above-mentioned measurement is repeated four times for the same aqueous solution 320, 20×4=80 sets of change amount data are measured for each aqueous solution 320. Here, the number of aqueous solutions 320 with different alcohol contents prepared is seven, but the number may be seven or more, and if seven types are used, 80 sets of change amount data are measured for each type, and 80×7=560 sets of change amount data are measured and stored in the memory 31.

[0123] Next, by performing multiple regression analysis in the analysis unit 30 with the alcohol content as the objective variable y and the six parameters Re12, Im12, Re11, Im11, Re22, and Im22 as explanatory variables, the multiple regression equation of the calibration curve can be obtained. This multiple regression equation is similar to the above-mentioned equation (51). Here, the numerical values ​​of the above-mentioned 560 sets of measurement data are omitted, but when the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (51) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 80 x 7 = 560 sets of data, the multiple regression equation shown in equation (88) can be obtained. y=953.37-867.29x1-654.74x2+784.82x3+277.33x4-935.74x5-282.46x6(88) In equation (88), x1 is the value of Re12, x2 is the value of Im12, x3 is the value of Re11, x4 is the value of Im11, x5 is the value of Re22, and x6 is the value of Im22. y in equation (88) is the multiple regression equation for the calibration curve calculated from the change amount data.

[0124] When it is desired to estimate the alcohol content of an aqueous solution 320, the container 300 containing the aqueous solution 320 is fed from the feed unit 12, and an electromagnetic wave of, for example, 748 MHz is supplied from the measurement unit 20 to the first antenna 11b via the first coaxial terminal 14a every time the container 300 containing the aqueous solution 320 moves a predetermined distance. Then, the transmitted wave of the electromagnetic wave irradiated from the first antenna 11b is received by the second antenna 11c, and the change in the real part (hereinafter referred to as "Re12'") and the change in the imaginary part (hereinafter referred to as "Im12'") of the transmitted wave of the container 300 containing the aqueous solution 320 is stored in the memory 31. Furthermore, the first antenna 11b receives a reflected wave of, for example, a 748 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re11'") and the amount of change in the imaginary part (hereinafter referred to as "Im11'") of the reflected wave A from the container 300 containing the aqueous solution 320 whose alcohol content is to be determined. Furthermore, the second antenna 11c receives a reflected wave of, for example, a 748 MHz electromagnetic wave, and stores in the memory 31 the amount of change in the real part (hereinafter referred to as "Re22'") and the amount of change in the imaginary part (hereinafter referred to as "Im22'") of the reflected wave B from the container 300 containing the aqueous solution 320 whose alcohol content is to be determined. Re12', Im12', Re11', Im11', Re22', and Im22' are data measured every time the predetermined interval is moved. In this case, the number of predetermined intervals is 20 as above, and 20 sets of data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. In other words, the 20 sets of data are data consisting of Re12', Im12', Re11', Im11', Re22', and Im22' obtained every 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 for which the alcohol content is to be calculated are stored in the memory 31. Here, Re12', Im12', Re11', Im11', Re22', and Im22' are data on the amount of change, which is the amount of change from the initial setting data measured in a state where the container 300 is not inserted in the electromagnetic wave sensing unit 10.

[0125] The 80 sets of data Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose alcohol content is to be determined moves a predetermined distance are substituted into the above formula (88) which is the formula for the calibration curve, to estimate the alcohol content of the aqueous solution 320 whose alcohol content is to be determined. The estimated alcohol content is 80 as a result of substituting the 80 sets of data. The analysis unit 30 can then calculate and present the average value of the 80 alcohol content values ​​estimated as the estimated alcohol content of the aqueous solution 320 whose alcohol content is to be determined. Alternatively, the range of the average value of the 80 alcohol content values ​​± the standard deviation may be presented. Here, the graph shown in Fig. 51 is created so that the variation in the alcohol content of 80 estimated by the calibration curve of the multiple regression equation shown in Equation (88) can be visualized. The graph shown in Fig. 51 plots the alcohol content of 80 with the horizontal axis representing the alcohol content [%] calculated by the calibration curve and the vertical axis representing the alcohol content [%] of the actual measured value of aqueous solution 320 whose alcohol content is to be calculated. Here, the numerical values ​​of the 80 sets of measurement data are omitted, but in FIG. 51, the estimated alcohol content is shown as a gray rectangle that is raw data. Referring to FIG. 51, 80 alcohol content values ​​calculated using the calibration curve are shown for each of the seven points of actual measurement alcohol content of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. This is because the six types of aqueous solutions 320 with the above seven alcohol content points are the aqueous solutions 320 for which the alcohol content is to be calculated, and the alcohol content value calculated using the calibration curve is calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 51 with the actual alcohol content as the objective variable yp and the alcohol content value calculated using the calibration curve as the explanatory variable x, the following regression equation (89) is obtained. yp=x-3×10 -13 (89) yp is shown in FIG. 51 as yp (raw data) by a solid gray line, and the partial regression coefficient of the regression equation (89) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated alcohol content of 80×7=560. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.9795 (90) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the alcohol content of the aqueous solution 320, the alcohol content of which is to be determined. Also, Fig. 52 shows a graph in which the error in the alcohol content of 560 calculated from the calibration curve for the case shown in Fig. 51 is plotted on a graph with the error between the alcohol content of 560 calculated from the calibration curve and the alcohol content of the actual measured value on the vertical axis and the alcohol content of the actual measured value on the horizontal axis. Referring to Fig. 52, the error is shown as a gray square that represents the raw data, and it can be seen that the error is within approximately ±5.0%. Therefore, referring to Figs. 51 and 52, it is evident that the calibration curve shown in the above formula (88) is highly reliable.

[0126] <Estimation of alcohol content of aqueous solution using electromagnetic wave sensing device according to the fifth embodiment of the present invention, part 2> In step 2 of estimating the alcohol content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22, which are the change amount data of the seven types of alcohol content at predetermined intervals obtained in step 1 of estimating the alcohol content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, are calculated by the analysis unit 30 and stored in the memory 31. Here, if the average values ​​of the change amount data of Re12, Im12, Re11, Im11, Re22, and Im22 are Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A, the average value data of the sets of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A is stored in the memory 31. In this case, the same measurement is performed four times for each container 300 containing aqueous solutions 320 of seven different alcohol contents, and 4×7=28 sets of average value data for Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are stored in memory 31.

[0127] Here, the alcohol content of the aqueous solution 320 is set as the objective variable y, and the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are set as explanatory variables, and the multiple regression equation of the calibration curve is obtained by performing multiple regression analysis in the analysis unit 30. The multiple regression equation of the calibration curve is the same as the above-mentioned equation (52). Here, the numerical values ​​of the above-mentioned 28 sets of measurement data are omitted, but when the partial regression coefficients b0, b1, b2, b3, b4, b5, and b6 of the multiple regression equation of equation (52) are calculated by the analysis unit 30 using the well-known least squares method based on the above-mentioned 28 sets of data, the multiple regression equation shown in equation (91) is obtained. y=838.47-1328.47x1-968.47x2+1442.28x3+658.58x4-694.81x5-223.51x6(91) In equation (91), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, and x6 is the value of Im22A. y in equation (91) is the multiple regression equation for the calibration curve calculated from the average value of the change amount data.

[0128] When it is desired to estimate the alcohol content of the aqueous solution 320, the container 300 containing the aqueous solution 320 is fed from the feed unit 12 in the same manner as in the estimation of the alcohol content of the aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, and data on the amount of change in Re12', Im12', Re11', Im11', Re22', and Im22' is measured every time the container 300 containing the aqueous solution 320 moves a predetermined distance, and the measured data on the amount of change is stored in the memory 31. Re12', Im12', Re11', Im11', Re22', and Im22' are regarded as data measured every time the container 300 moves a predetermined distance, and in this case, the number of predetermined intervals is 20 as above, and 20 sets of data each consisting of Re12', Im12', Re11', Im11', Re22', and Im22' are measured, the same number as the number of predetermined intervals. That is, the 20 sets of data are data sets of Re12', Im12', Re11', Im11', Re22', and Im22' acquired at 20 predetermined intervals. Then, by repeating the same measurement four times, 80 sets of data for the container 300 containing the aqueous solution 320 whose alcohol content is to be calculated are stored in the memory 31.

[0129] Next, the analysis unit 30 calculates the average value of the data of the amount of change at a predetermined interval for each container 300 containing the aqueous solution 320 whose alcohol content is to be determined, and stores the calculated average data of the amount of change, Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', is stored in the memory 31. In this case, the same measurement is performed four times for the container 300 containing the aqueous solution 320 whose alcohol content is to be determined, so that the data of the average values ​​of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A' is stored in the memory 31. The data of four sets of Re12A', Im12A', Re11A', Im11A', Re22A', and Im22A', which are the average data of Re12', Im12', Re11', Im11', Re22', and Im22' obtained each time the container 300 containing the aqueous solution 320 whose alcohol content is to be determined moves a predetermined distance, can be substituted into the above formula (91), which is the formula of the calibration curve, to estimate the alcohol content of the solution 310 whose alcohol content is to be determined. When the above four sets of data are substituted into the formula (91), the number of estimated alcohol content values ​​becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated alcohol content values ​​as the estimated alcohol content of the aqueous solution 320 whose alcohol content is to be determined.

[0130] Here, in order to visualize the variation in the four alcohol contents estimated by the calibration curve of the multiple regression equation shown in Equation (82), the graph shown in Fig. 51 is created. The graph shown in Fig. 51 plots the above four alcohol contents with the horizontal axis representing the alcohol content [%] calculated by the calibration curve and the vertical axis representing the alcohol content [%] of the actual measured value of the aqueous solution 320 whose alcohol content is to be calculated. Here, the values ​​of the four sets of measurement data are omitted, but the four estimated alcohol contents are shown as white circles that are averaged data. Referring to FIG. 51, the alcohol contents of the four calibration curve calculations are shown for seven points of actual measurement alcohol contents of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. This is because the seven types of aqueous solutions 320 with the alcohol contents of the seven points are the aqueous solutions 320 whose alcohol contents are to be obtained, and the calibration curve calculation alcohol contents are obtained from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 51 with the actual alcohol contents as the objective variable yq and the calibration curve calculation alcohol contents x as the explanatory variable, the following regression equation (92) is obtained. yq=x-2×10 -13 (92) yq is shown in FIG. 51 as a straight dashed line, yq (averaged data), and the partial regression coefficient of the regression equation (92) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated 4×7=28 alcohol content. The coefficient of determination of this regression equation, R 2 teeth, R 2 =0.9967 (93) This shows that the degree to which the objective variable can be explained by the explanatory variables is high. In other words, it is possible to quantitatively and satisfactorily estimate the alcohol content of the aqueous solution 320, the alcohol content of which is to be determined. In addition, Fig. 52 shows a graph in which the error in the alcohol content of 28 calculated from the calibration curve for the case shown in Fig. 51 is plotted on a graph with the error between the alcohol content of 28 calculated from the calibration curve and the actual measured alcohol content on the vertical axis and the actual measured alcohol content on the horizontal axis. Referring to Fig. 52, the error is shown as an open circle representing averaged data, and it can be seen that the error is within approximately ±1.2%. Therefore, referring to FIGS. 51 and 52, it can be seen that the calibration curve calculated from the average value of the change amount data shown in the above formula (91) is more reliable.

[0131] <Estimation of alcohol content of aqueous solution using electromagnetic wave sensing device according to the fifth embodiment of the present invention, 3> In step 3 of estimating the alcohol content of an aqueous solution in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, in addition to the average data of the amount of change acquired in step 2 of estimating the alcohol content in the electromagnetic wave sensing device 5 of the fifth embodiment of the present invention, the average data of the VSWR is added to the explanatory variables and multiple regression analysis is performed in the analysis unit 30. In this case, in the electromagnetic wave sensing device 5 of the fifth embodiment, in addition to the amount of change data, VSWR data is measured and stored in the memory 31. Then, the analysis unit 30 calculates average data of the amount of change data and the VSWR data stored in the memory 31, and the average data of the amount of change data and the average data of the VSWR are stored in the memory 31. Specifically, the distance between the first antenna 11b and the aqueous solution 320 in the container 300 is set to, for example, about 546.1 mm, and an electromagnetic wave of 748 MHz is irradiated from the first antenna 11b side to measure data (Re12, Im12, Re11, Im11) of the change in the transmitted wave and the reflected wave at every predetermined interval of the container 300 containing the solution 310, while VSWR data VSWR1 is also measured. In addition, the distance between the second antenna 11c and the aqueous solution 320 in the container 300 is set to, for example, about 645.5 mm, and an electromagnetic wave of 748 MHz is irradiated from the second antenna 11c side to measure data (Re22, Im22) of the change in the reflected wave at every predetermined interval of the container 300 containing the aqueous solution 320, while VSWR data VSWR2 is also measured. Next, the average value of the change amount data measured at predetermined intervals is calculated by the analysis unit 30, and the average value data of Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A mentioned above, the average value data VSWR1A of VSWR1, and the average value data VSWR2A of VSWR2 are stored in the memory 31. In this case, the same measurement is performed four times for each of the containers 300 containing aqueous solutions 320 with seven different alcohol contents, so that 4×7=28 sets of data, Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A, are stored in the memory 31.

[0132] As described above, eight parameter sets of Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are stored in the memory 31. Then, since the measurements of the seven types of aqueous solutions 320 with known alcohol contents are repeated four times, a total of 4×7=28 sets of average data of the amount of change made up of the above eight parameters are stored in the memory 31. Therefore, by performing multiple regression analysis in the analysis unit 30 with the alcohol content y as the objective variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables, a multiple regression equation for the calibration curve can be obtained. This multiple regression equation is similar to the above-mentioned equation (53), and although the numerical values ​​of the above-mentioned 28 sets of measurement data are omitted here, when the partial regression coefficients of the multiple regression equation for equation (53) are calculated by the analysis unit 30 based on the above-mentioned 28 sets of data using the well-known least squares method, equation (94) is obtained. y=373.17-19.01x1-95.15x2+40.87x3+86.27x4-85.66x5-78.78x6-220.69x7+59.07x8(94) In equation (94), x1 is the value of Re12A, x2 is the value of Im12A, x3 is the value of Re11A, x4 is the value of Im11A, x5 is the value of Re22A, x6 is the value of Im22A, x7 is the value of VSWR1A, and x8 is the value of VSWR2A. y in equation (94) is the multiple regression equation for the calibration curve calculated from the average value of the change data including VSWR.

[0133] When it is desired to estimate the alcohol content of the aqueous solution 320 for which the alcohol content is to be calculated, the same measurement is performed as for the seven types of aqueous solutions 320 with known alcohol content described above, and data on the amount of change at a predetermined interval in the container 300 containing the aqueous solution 320 for which the alcohol content is to be calculated is measured. In the above measurement, the frequency of the electromagnetic wave emitted from the first antenna 11b or the second antenna 11c is set to 748 MHz. Next, the data Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data on the amount of change at a predetermined interval in the aqueous solution 320 for which the alcohol content is to be calculated, are calculated by the analysis unit 30, and the average data of the calculated amount of change is stored in the memory 31. Here, since the measurement is repeated four times, data on the average values ​​of the four sets of amount of change is stored in the memory 31. The above four sets of data, Re12A', Im12A', Re11A', Im11A', Re22A', Im22A', VSWR1A', and VSWR2A', which are the average values ​​of the data of the amount of change at a predetermined interval of the aqueous solution 320 whose alcohol content is to be determined, are substituted into the above formula (85), which is the formula of the calibration curve, to estimate the alcohol content of the aqueous solution 320 whose alcohol content is to be determined. When the above four sets of data are substituted into the formula (85), the number of estimated alcohol content values ​​becomes four. Then, the analysis unit 30 can calculate and present the average value of the four estimated alcohol content values ​​as the estimated alcohol content of the aqueous solution 320 whose alcohol content is to be determined.

[0134] Here, a graph shown in Fig. 53 is created so that the variation in the four alcohol contents estimated by the calibration curve of the multiple regression equation shown in Equation (85) can be visualized. The graph shown in Fig. 53 plots the above four alcohol contents with the horizontal axis representing the alcohol content [%] calculated by the calibration curve and the vertical axis representing the alcohol content [%] of the actual measured value of the aqueous solution 320 whose alcohol content is to be calculated. Here, the values ​​of the four sets of measurement data are omitted, but the estimated alcohol content is shown as four white circles that are averaged data. In FIG. 53, the alcohol content of the four calibration curve calculations is shown for each of seven points of actual measurement alcohol content of about 0%, about 5%, about 10%, about 15%, about 20%, about 25%, and about 30%. This is because the six types of aqueous solutions 320 with the above seven alcohol content points are the aqueous solutions 320 whose alcohol content is to be calculated, and the calibration curve calculation alcohol content is calculated from the measurement data. Here, when the analysis unit 30 performs regression analysis on the case shown in FIG. 53 with the actual alcohol content as the objective variable yn and the calibration curve calculation alcohol content x as the explanatory variable, the following regression equation (95) is obtained. yr=x-6×10 -15 (95) yr is shown in FIG. 53 as a straight dashed line yr (with VSWR), and the partial regression coefficient of the regression equation yr in equation (95) is calculated by the analysis unit 30 using the well-known least squares method based on the estimated 4×7=28 alcohol content. The coefficient of determination R 2 teeth, R 2 =0.99994 (96) It can be seen that the degree to which the objective variable can be explained by the explanatory variables is higher. In other words, it can be seen that the alcohol content of the aqueous solution 320 whose alcohol content is to be obtained can be estimated more quantitatively. The gray rectangular plot shown as yq (without VSWR) in FIG. 53 is the same as the plot shown by the white circle in FIG. 51. Here, when the regression equation yr is compared with the regression equation yq, it can be seen that the reliability of the calibration curve is higher when the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A are used as explanatory variables than when the six parameters Re12A, Im12A, Re11A, Im11A, Re22A, and Im22A are used as explanatory variables. FIG. 54 shows a graph in which the error between the four alcohol content values ​​calculated from the calibration curve and the actual alcohol content values ​​is plotted on the vertical axis, and the actual alcohol content values ​​are plotted on the horizontal axis, in the case shown by the dashed line in FIG. 49. In FIG. 54, the error is shown as four open circles with VSWR, and it can be seen that the error is within about ±0.3%. Note that the gray square plot shown as "without VSWR" in FIG. 54 is the same as the open circle plot shown in FIG. 52. 53 and 54, it can be seen that the calibration curve shown in the above formula (94), which is created when VSWR1A and VSWRA2 are also added as explanatory variables, is more reliable than the calibration curve shown in the above formula (91). From the above, it can be said that the reliability of the calibration curve created with alcohol content as the dependent variable and the eight parameters Re12A, Im12A, Re11A, Im11A, Re22A, Im22A, VSWR1A, and VSWR2A as explanatory variables is extremely high. [Industrial Applicability]

[0135] In the electromagnetic wave sensing device according to the embodiment of the present invention described above, the parameters of the explanatory variables in the multiple regression equation include the real and imaginary parts of the transmitted wave and the real and imaginary parts of the reflected wave, but only one of the real and imaginary parts of the transmitted wave and the real and imaginary parts of the reflected wave may be included in the parameters. In addition, the electromagnetic wave sensing device according to the embodiment of the present invention is provided with a first antenna and a second antenna, but either the first antenna or the second antenna may be provided. When either the first antenna or the second antenna is provided, the parameters of the explanatory variables include only the real and imaginary parts of the reflected wave.

[0136] As described above, in the electromagnetic wave sensing device according to the embodiment of the present invention, when creating a calibration curve, the content of the ingredients contained in the samples is known in advance, for example, a plurality of samples with the content in the range of 2% to 110% are prepared, and data on the amplitude and phase of the reflected wave and the transmitted wave, or data on the real part and the imaginary part of the reflected wave and the transmitted wave expressed in complex numbers are measured. In addition, the above data is measured at predetermined intervals of the sample, and the measurement to obtain the above data is repeated twice. This is to improve the reliability of the calibration curve created based on the above data. In this case, it is preferable to prepare a plurality of samples in which the content is actually measured and the content is spaced apart from each other by a predetermined distance. In the above description, the number of samples to be prepared is 12, 9, 7, or 6, but the number of samples is not limited to this and may be other numbers. For example, in the case of a sample of lumber, the number of pieces of lumber 100 having a moisture content in the range of 2% to 110% is 12 pieces of lumber 100 having moisture contents of about 2.1%, about 12.9%, about 23.0%, about 32.3%, about 41.0%, about 50.1%, about 59.9%, about 69.3%, about 80.1%, about 90.2%, about 102.3%, and about 110.5%, but the number of samples is not limited to this and may be any other number of pieces of lumber 100 having different moisture contents. Furthermore, when the sample is wood chips or a solution, the range of moisture content of the wood chips and the range of solute concentration of the solution are not limited to the ranges mentioned above, and multiple samples with different moisture contents or concentrations may be prepared. Furthermore, in the electromagnetic wave sensing device of the embodiment of the present invention described above, the same measurement is repeated two or four times, but this is not limited to this and a single measurement may be performed, or the measurement may be repeated three or five or more times.

[0137] Furthermore, in the electromagnetic wave sensing device according to the embodiment of the present invention described above, when data is stored in the memory 31, a folder that allows data to be distinguished is provided and the data is stored in the folder, but this is not limiting, and a flag or the like that allows data to be distinguished may be added to the data and stored in the memory 31. In other words, the data may be stored in the memory 31 in a manner that allows the data to be distinguished. Furthermore, in the electromagnetic wave sensing device of the embodiment of the present invention described above, the number of predetermined intervals over which the sample moves is 20, but this is not limited to this and the number of intervals may be less than 20 or greater than 21. Furthermore, in the electromagnetic wave sensing device according to the embodiment of the present invention described above, the frequency at which data is measured is a specific frequency according to the sample, but it is not limited to the above-mentioned frequency and can be any frequency as long as the variation in the amount of change data is suppressed. Also, the data is data of the real and imaginary parts of the reflected and transmitted waves expressed in complex numbers, but the data may be data of the amplitude and phase of the reflected and transmitted waves. Furthermore, in the electromagnetic wave sensing device according to the embodiment of the present invention described above, the regression equation obtained by regression analysis is a linear function, i.e., a regression line. Furthermore, data on the change in the real part or the average value of the change in the real part of the transmitted wave received by the first antenna 11b, which is a transmitted wave of the electromagnetic wave irradiated from the second antenna 11c, and received by the first antenna 11b, and data on the change in the imaginary part or the average value of the change in the imaginary part of the transmitted wave may be added to the explanatory variables of the regression analysis. Furthermore, it goes without saying that, instead of receiving the transmitted wave of the electromagnetic wave irradiated from the first antenna by the second antenna and using data on the change in the real part or the average value of the change in the real part of the transmitted wave received by the second antenna, and data on the change in the imaginary part or the average value of the change in the imaginary part of the transmitted wave, it is also possible to receive the transmitted wave of the electromagnetic wave irradiated from the second antenna by the first antenna and use data on the change in the real part or the average value of the change in the real part of the transmitted wave received by the first antenna, and data on the change in the imaginary part or the average value of the change in the imaginary part of the transmitted wave.

[0138] In the electromagnetic wave sensing device according to the embodiment of the present invention described above, only the amount of change in the amplitude and phase of the transmitted wave or only the amount of change in the amplitude and phase of the reflected wave may be used as parameters of explanatory variables in the multiple regression equation. In this case, it is preferable to measure the amount of change in the amplitude and phase of the transmitted wave or the amount of change in the amplitude and phase of the reflected wave at a frequency at which the reliability of the multiple regression equation is high. The electromagnetic wave sensing device according to the embodiment of the present invention described above can estimate the amount of a substance contained in a sample. Specifically, the moisture content of lumber and wood chips can be estimated, and the concentration of a solute in a solution can be estimated. Specifically, the sugar content, salt concentration, and alcohol content of an aqueous solution can be estimated. Furthermore, the aqueous solution may be, for example, an aqueous solution of agricultural products (fruits, vegetables), food (pickles, fermented foods), or beverages (alcohol, juice), and the amount of a substance contained therein can be the salt concentration, sugar content, alcohol content, etc. [Explanation of symbols]

[0139] 1 electromagnetic wave sensing device, 10 electromagnetic wave sensing section, 11 sensing housing, 11a radio wave absorber, 11b first antenna, 11c second antenna, 12 input section, 12a radio wave absorber, 13 output section, 13a radio wave absorber, 14a coaxial terminal, 14b coaxial terminal, 20 measuring section, 30 analyzing section, 31 memory, 100 lumber, 200 container, 210 wood chips, 300 container, 310 solution, 320 aqueous solution

Claims

1. an electromagnetic wave sensing unit that irradiates an electromagnetic wave onto a sample that is moved at a predetermined interval and outputs a transmitted wave and / or a reflected wave; a measurement unit which receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing unit each time the sample moves a predetermined distance, and measures either the amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance, or the amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance; an analysis unit that performs regression analysis on a relationship between first transmitted wave measurement data of changes in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance and / or first reflected wave measurement data of changes in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, which are previously measured by the measurement unit, and an amount of a content of an inclusion contained in the sample, to obtain a regression equation in which the content of the sample is a response variable and the first transmitted wave measurement data and the first reflected wave measurement data are explanatory variables; The change in amplitude and phase is a change in amplitude and phase from when the sample is not present, An electromagnetic wave sensing device characterized in that, in the sensing unit and the measurement unit, second transmitted wave measurement data of the amount of change in amplitude and phase of the transmitted wave measured each time the measured sample moves a predetermined distance, and second reflected wave measurement data of the amount of change in amplitude and phase of the reflected wave measured each time the measured sample moves a predetermined distance, are substituted into the regression equation determined by the analysis unit, thereby estimating the content of ingredients contained in the measured sample.

2. an electromagnetic wave sensing unit that irradiates an electromagnetic wave onto a sample that is moved at a predetermined interval and outputs a transmitted wave and / or a reflected wave; a measurement unit which receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing unit each time the sample moves a predetermined distance, and measures either the amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance, or the amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance; an analysis unit that performs regression analysis on a relationship between first transmitted wave average measurement data of an average amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance and / or first reflected wave average measurement data of an average amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, which are previously measured by the measurement unit, and an amount of a content of an inclusion contained in the sample, to obtain a second regression equation in which the content of the sample is used as a response variable and the first transmitted wave average measurement data and the first reflected wave average measurement data are explanatory variables; The change in amplitude and phase is a change in amplitude and phase from when the sample is not present, An electromagnetic wave sensing device characterized in that, in the sensing unit and the measurement unit, second transmitted wave averaged measurement data of the average amount of change in amplitude and phase of the transmitted wave measured each time the measured sample moves a predetermined distance, and second reflected wave averaged measurement data of the average amount of change in amplitude and phase of the reflected wave measured each time the measured sample moves a predetermined distance, are substituted into the second regression equation determined by the analysis unit, thereby estimating the content of ingredients contained in the measured sample.

3. an electromagnetic wave sensing unit that irradiates an electromagnetic wave onto a sample that is moved at a predetermined interval and outputs a transmitted wave and / or a reflected wave; a measurement unit which receives the transmitted wave and / or the reflected wave output from the electromagnetic wave sensing unit each time the sample moves a predetermined distance, and measures either the amount of change in amplitude and phase of the transmitted wave each time the sample moves a predetermined distance, or the amount of change in amplitude and phase of the reflected wave each time the sample moves a predetermined distance, and a VSWR each time the sample moves a predetermined distance; an analysis unit that performs regression analysis on a relationship between first transmitted wave average measurement data of an average value of changes in amplitude and phase of the transmitted wave every time the sample moves a predetermined distance and / or first reflected wave average measurement data of an average value of changes in amplitude and phase of the reflected wave every time the sample moves a predetermined distance, which are previously measured by the measurement unit, first VSWR average measurement data of an average value of the VSWR every time the sample moves a predetermined distance, and an amount of a content contained in the sample, to obtain a third regression equation in which the content of the sample is used as a response variable and the first transmitted wave average measurement data and / or the first reflected wave average measurement data and the first VSWR average measurement data are explanatory variables, The change in amplitude and phase is the amplitude and phase from when the sample is not present, The electromagnetic wave sensing device is characterized in that, in the sensing unit and the measurement unit, the content of ingredients contained in the measured sample is estimated by substituting, into the third regression equation determined by the analysis unit, second transmitted wave averaged measurement data which is the average value of the changes in amplitude and phase of the transmitted wave measured each time the measured sample moves a predetermined distance, second reflected wave averaged measurement data which is the average value of the changes in amplitude and phase of the reflected wave measured each time the measured sample moves a predetermined distance, and second VSWR averaged measurement data which is the average value of the VSWR measured each time the measured sample moves a predetermined distance.

4. 4. An electromagnetic wave sensing device as described in any one of claims 1 to 3, characterized in that in the measurement unit and the analysis unit, the amount of change in the amplitude and phase of the transmitted wave is represented by the amount of change in the real part and the imaginary part of the complex representation of the transmitted wave, and the amount of change in the amplitude and phase of the reflected wave is represented by the amount of change in the real part and the imaginary part of the complex representation of the reflected wave.

5. the electromagnetic wave sensing unit includes a metallic sensing housing in which at least one antenna capable of receiving electromagnetic waves from the sample or the measured sample is provided, a metallic feed unit that feeds the sample or the measured sample into the sensing housing, and a metallic feed unit that feeds the sample or the measured sample from the sensing housing; 4. An electromagnetic wave sensing device as described in any one of claims 1 to 3, characterized in that a radio wave absorber is attached to the inner wall surfaces of the sensing housing except for the surface on which the antenna is provided, and to the inner wall surfaces of the input section and the output section.

6. 4. An electromagnetic wave sensing device according to claim 1, wherein the sample and the measured sample are lumber, and the content of ingredients contained in the sample and the measured sample is regarded as the moisture content of the lumber.

7. An electromagnetic wave sensing device as described in any one of claims 1 to 3, characterized in that the sample and the measured sample are wood chips stored in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is the moisture content of the wood chips stored in the container.

8. 4. An electromagnetic wave sensing device as described in any one of claims 1 to 3, characterized in that the sample and the measured sample are solutions placed in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is the concentration of solutes in the solution placed in the container.

9. An electromagnetic wave sensing device as described in any one of claims 1 to 3, characterized in that the sample and the measured sample are aqueous solutions placed in a non-metallic container, and the content of ingredients contained in the sample and the measured sample is either the sugar content, salt concentration, or alcohol content of the aqueous solution placed in the container.

Images