Data extraction system, data extraction method and data extraction program

Abstract

To provide a technology capable of improving quality of data used in analysis.SOLUTION: A data extraction system 1 comprises: a setting unit 7 that sets input reference indicating performance required for data analysis; an input unit 2 that acquires an input signal indicating a measurement result of an object to be measured in biological production; a pre-processing unit 3 that performs a pre-processing operation on the input signal to compare it with the input reference, and generates a pre-processed signal; a post-processing unit 4 that performs a post-processing operation on the pre-processed signal using another signal having measured the object to be measured when the pre-processed signal does not satisfy the input reference, and generates a post-processed signal; and an extraction unit 5 that extracts the pre-processed signal as extraction data when the pre-processed signal satisfies the input reference, and extracts the post-processed signal as extraction data when the post-processed signal satisfies the input reference.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a data extraction system, a data extraction method, and a data extraction program. [Background technology]

[0002] There are known techniques for measuring changes in substances over time using sensors. Patent Document 1 discloses a technique for detecting the presence or absence of abnormalities in a culture medium. The technique described in Patent Document 1 obtains the distribution state of characteristic values ​​of the culture medium and sets a threshold value for determining abnormalities based on the distribution state. Patent Document 2 discloses a technique for analyzing the deterioration of plastic materials and the like. The technique described in Patent Document 2 obtains corrected spectral data by performing second-order differentiation processing on near-infrared spectral data of plastic materials and the like obtained by near-infrared spectroscopy. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6977977 [Patent Document 2] Japanese Patent Application Publication No. 2022-7236 Summary of the Invention [Problem to be solved by the invention]

[0004] There is a desire to improve the quality of the data used in the analysis. [Means for solving the problem]

[0005] A data extraction system according to one aspect of the present disclosure includes a setting unit that sets input standards that indicate the performance required for data analysis; an input unit that acquires an input signal that indicates the measurement results of an object to be measured in biological production; a pre-processing unit that performs pre-processing operations on the input signal to compare it with the input standards and generate a pre-processed signal; a post-processing unit that, if the pre-processed signal does not satisfy the input standards, performs post-processing operations on the pre-processed signal using another signal that indicates the measurement results of the object to be measured and generates a post-processed signal; and an extraction unit that extracts the pre-processed signal as extracted data if the pre-processed signal satisfies the input standards, and extracts the post-processed signal as extracted data if the post-processed signal satisfies the input standards.

[0006] In a data extraction system according to one aspect of the present disclosure, a pre-processing operation is performed on an input signal to generate a pre-processed signal. The pre-processed signal is then compared with an input standard indicating performance required for data analysis. If the pre-processed signal satisfies the input standard, the pre-processed signal is extracted as extracted data. On the other hand, if the pre-processed signal does not satisfy the input standard, a post-processing operation is performed on the pre-processed signal to generate a post-processed signal. The post-processed signal is then compared with the input standard. If the post-processed signal satisfies the input standard, the post-processed signal is extracted as extracted data. This makes it possible to extract extracted data with guaranteed quality. As a result, it is possible to improve the quality of data used for analysis. [Effects of the Invention]

[0007] According to the present disclosure, it is possible to provide a technique that can improve the quality of data used for analysis. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram schematically showing the configuration of a data extraction system according to the first embodiment. [Figure 2] FIG. 2 is a schematic diagram showing an example of the operation of the data extraction system. [Figure 3] FIG. 3 shows an example of spectrophotometric measurement in biological production. [Figure 4] FIGS. 4(a) to 4(c) are graphs illustrating the extraction of changes in organic molecule concentration in near-infrared spectroscopy. [Figure 5] FIG. 5 is a schematic diagram illustrating another example of the operation of the data extraction system. [Figure 6] FIG. 6 is a schematic diagram illustrating yet another example of the operation of the data extraction system. [Figure 7] (a) to (d) of FIG. 7 are graphs showing an example for explaining the effect of the pre-processing calculation and the setting of the input standard. [Figure 8] (a) to (d) of FIG. 8 are graphs showing another example for explaining the effect of pre-processing calculations and setting of input criteria. [Figure 9] FIG. 9 is an overall view showing an example of application of a data extraction system according to an embodiment. [Figure 10] FIG. 10 is a diagram showing an example of the overall operation of a biological production system including a data extraction system. [Figure 11] FIG. 11 is a diagram illustrating an example of the operation of the data extraction system. [Figure 12] FIG. 12 is a diagram illustrating an example of a hardware configuration related to the data extraction system. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the description of the drawings, the same elements are designated by the same reference numerals, and duplicated description will be omitted.

[0010] [overview] The data extraction system of the present disclosure extracts data measured by a sensor for data analysis. Hereinafter, data used for data analysis will be referred to as "extracted data." In data analysis, minimum performance may be required for the extracted data. Hereinafter, the minimum performance will be referred to as "evaluation index." The data extraction system may be applied to data analysis of, for example, bioproduction or pharmaceutical manufacturing, but is not limited thereto. This disclosure will describe an example in which the data extraction system is applied to data analysis in bioproduction.

[0011] The data extraction system acquires a signal measured by a sensor as an input signal. The data extraction system sets an index of the input signal corresponding to the evaluation index as an input criterion. The data extraction system performs a preprocessing operation on the input signal to compare with the input criterion and generate a preprocessed signal. The data extraction system compares the preprocessed signal with the input criterion and determines whether the input criterion is satisfied.

[0012] The data extraction system performs post-processing calculations on pre-processed signals that do not satisfy the input criteria using another signal measured by a sensor, thereby generating a post-processed signal. The another signal may be, for example, a signal before or after the input signal in a time series. The data extraction system compares the post-processed signal with the input criteria to determine whether the input criteria are satisfied. The data extraction system then extracts, as extracted data, the pre-processed signals or post-processed signals that satisfy the input criteria. The data extraction system may also remove post-processed signals that do not satisfy the input criteria. The data extraction system may also extract, as extracted data, data that has been subjected to calculations for output or the like on the pre-processed signals or post-processed signals. In this way, the data extraction system extracts extracted data that satisfies the evaluation index.

[0013] [Embodiment] 1 is a diagram schematically illustrating a configuration of a data extraction system 1 according to an embodiment. The data extraction system 1 is communicably connected to a sensor 30 and a database 40. The data extraction system 1 acquires an input signal from, for example, the sensor 30, and extracts a plurality of extracted data. The data extraction system 1 stores the plurality of extracted data in the database 40.

[0014] The sensor 30 is a measuring device that outputs a signal indicating the measurement result of an object to be measured in biological production. The number of sensors 30 is not limited. Examples of the sensor 30 include, but are not limited to, a hydrogen ion concentration (pH: Potential Hydrogen) sensor, a dissolved oxygen (DO: Dissolved Oxygen) sensor, a temperature sensor, a near-infrared sensor, an RGB sensor, a potential sensor, and a torque sensor. The sensor 30 is installed in the culture tank 20. The culture tank 20 is a container used for culturing cells, etc. The culture tank 20 contains a culture medium. The sensor 30 outputs a signal indicating the measurement result of the state of the culture tank 20 to the data extraction system 1.

[0015] The database 40 is a non-transitory storage medium or storage device that stores information. The database 40 stores, for example, evaluation indexes and extracted data. The database 40 may store signals acquired by the sensors 30 as data. The database 40 may be constructed as a single database or may be a collection of multiple databases. The location where the database 40 is installed is not limited. For example, the database 40 may be installed in a computer system separate from the data extraction system 1.

[0016] The data extraction system 1 includes, as functional elements, an input unit 2, a pre-processing unit 3, a post-processing unit 4, an extraction unit 5, and a control block 10. The control block 10 includes, as functional elements, a control unit 6, a setting unit 7, a comparison unit 8, and an abnormality determination unit 9.

[0017] The input unit 2 acquires signals output from the sensors 30 as input signals. The input unit 2 acquires M input signals (M>0) from, for example, M sensors 30. Examples of input signals include, but are not limited to, pH, dissolved oxygen content, temperature, optical spectrum, RGB value, electric potential (electrode potential), and agitator rotation speed (Agit). The input unit 2 may acquire signals stored in a database 40 as input signals.

[0018] The control unit 6 determines which sensor 30 measured the input signal. For example, the control unit 6 may distinguish between sensors 30 by reading out the sensor ID attached to the input signal. The control unit 6 sets the sensor 30 throughout the culture period to optimize the input signal. For example, the control unit 6 sets the range of the sensor 30 that outputs the optical spectrum according to the culture stage. Since the transmittance of the culture medium is high in the early stages of culture, the control unit 6 narrows the range (shortens the exposure time). As the culture progresses, the transmittance of the culture medium decreases, so the control unit widens the range (longens the exposure time).

[0019] The setting unit 7 sets input criteria that indicate the performance required for data analysis. The setting unit 7 sets an index of the input signal corresponding to the evaluation index as the input criteria. The setting unit 7 may acquire the evaluation index from, for example, the database 40 or an external device, or may accept input of the evaluation index. In one example, the input criteria may be a signal-to-noise ratio (SNR).

[0020] The pre-processing unit 3 performs pre-processing operations on the input signal to compare it with an input reference and generate a pre-processed signal. That is, the pre-processing unit 3 processes the input signal to make it correspond to the format of the input reference. For example, the pre-processing unit 3 may perform gain adjustment, calibration, etc. as pre-processing operations. The pre-processing unit 3 may perform operations using multiple input signals as pre-processing operations. The pre-processing unit 3 may calculate an SNR as pre-processing operations.

[0021] The comparison unit 8 compares the signal with an input reference corresponding to the signal to generate a comparison result. The comparison result may indicate whether the signal satisfies the input reference. The comparison unit 8 may compare the pre-processed signal with the input reference corresponding to the pre-processed signal to generate a comparison result. The comparison unit 8 may compare the post-processed signal with the input reference corresponding to the post-processed signal to generate a comparison result.

[0022] If the pre-processed signal does not satisfy the input criteria, the post-processing unit 4 performs a post-processing operation on the pre-processed signal using another signal obtained by measuring the object to generate a post-processed signal. For example, the post-processing operation may be an arithmetic average or moving average of the signal. If the pre-processed signal satisfies the input criteria, the post-processing unit 4 does not perform the post-processing operation. In other words, if the pre-processed signal satisfies the input criteria, the post-processing unit 4 does not generate a post-processed signal.

[0023] The extraction unit 5 extracts extracted data based on the comparison result by the comparison unit 8. If the pre-processed signal satisfies the input criteria, the extraction unit 5 extracts the pre-processed signal as extracted data. If the post-processed signal satisfies the input criteria, the extraction unit 5 extracts the post-processed signal as extracted data. The extraction unit 5 may extract data obtained by performing an output operation on the pre-processed signal or the post-processed signal as extracted data. If the post-processed signal does not satisfy the input criteria, the extraction unit 5 may remove the post-processed signal. In other words, the extraction unit 5 converts and selects data.

[0024] The abnormality determination unit 9 determines an abnormality based on the input signal, the pre-processed signal, or the post-processed signal. For example, the abnormality determination unit 9 may determine an abnormality based on a change in the input signal or a correlation between input signals. In one example, the abnormality determination unit 9 may detect an abnormality based on a change in the amount of dissolved oxygen. In another example, the abnormality determination unit 9 may detect an abnormality based on a correlation between the amount of dissolved oxygen or pH and the electrode potential. The abnormality determination unit 9 may invalidate the extracted data as a process in the event of an abnormality. "Invalidating the extracted data" may mean replacing the pre-processed signal, the post-processed signal, or the extracted data with None data.

[0025] How the evaluation index and input criteria are determined may vary depending on the content of the input signal. Three examples of evaluation indexes (first, second, and third cases) will be described with reference to Fig. 2. Fig. 2 is a schematic diagram showing an example of the operation of the data extraction system 1. In the example shown in Fig. 2, three pieces of extracted data are extracted from six input signals.

[0026] In the first case, an evaluation index e1 for evaluating one type of signal is used. The setting unit 7 converts the evaluation index e1 into an input criterion c1. The evaluation index e1 may be the same index as the input criterion c1.

[0027] For example, when the concentration of a certain aqueous solution changes by 50 mM (mmol / L), the SNR for the rate of change in absorbance required to detect this change may be required as the evaluation index e1. If the evaluation index e1 is assumed to be 20 dB, the setting unit 7 calculates the SNR (input standard c1) required for transmitted light using, for example, the following formula: ΔA(SNR)=20dB 20xLOG(ΔS / N)-6(dB)=20dB

[0028] Here, ΔA is the difference in absorbance. ΔS is the difference in transmitted light intensity. N is the base noise of the optical sensor. 6 dB is the SNR that deteriorates before ΔA is calculated from the transmitted light S. If ΔS = S x 1%, the SNR required for the transmitted light (S) is calculated to be 66 dB (= input standard c1). The setting unit 7 converts the evaluation index e1 into the input standard c1 by performing this calculation for the evaluation index e1. The setting unit 7 may always calculate up to the absorbance in preprocessing and match the evaluation index e1 with the input standard c1. Such a selection may be made on a case-by-case basis depending on the type of culture and the elements to be monitored.

[0029] In one example, a change in absorbance may be monitored based on the absorbance at the beginning of the culture, and the change may be small at the beginning of the culture but large from the middle to the later stages of the culture. In this case, the setting unit 7 may always maintain the same evaluation criterion for the absorbance change ΔA, but may separate the input criterion for the initial transmitted light, which serves as a reference value, from the input criterion for the transmitted light after a certain period of time has elapsed. For example, the setting unit 7 may extract data using an input criterion with a high SNR at the beginning of the culture, when the change is small, and lower the SNR for the transmitted light from the middle stage of the culture onward, when the change is large. In another example, a change in absorbance (ΔA) may be constantly monitored at regular intervals. In this case, the setting unit 7 may match the evaluation index e1 with the input criterion c1. By setting input criterion corresponding to each case in this way, optimal data extraction is possible.

[0030] The input unit 2 acquires an input signal x1. The input signal x1 is, for example, an optical spectrum. The preprocessing unit 3 performs preprocessing operations such as gain adjustment and calibration on the input signal x1 to generate a preprocessed signal f1. For example, gain adjustment for the optical spectrum absorbs differences in exposure time. Calibration for the optical spectrum is dark correction. In another example, at least one of gain adjustment and calibration may be performed depending on the type of input signal x1 or the type of sensor 30. The comparison unit 8 compares the preprocessed signal f1 with an input reference c1 to generate a comparison result.

[0031] If the pre-processed signal f1 does not satisfy the input standard c1, the post-processing unit 4 performs post-processing operations such as averaging or moving averaging on the pre-processed signal f1 to generate a post-processed signal p1. The comparison unit 8 compares the post-processed signal p1 with the input standard c1 to generate a comparison result.

[0032] If the pre-processed signal f1 satisfies the input criterion c1, the extraction unit 5 extracts the pre-processed signal f1 as extracted data d1. If the post-processed signal p1 satisfies the input criterion c1, the extraction unit 5 extracts the post-processed signal p1 as extracted data d1. If the post-processed signal p1 does not satisfy the input criterion c1, the extraction unit 5 removes the post-processed signal p1. For example, the extraction unit 5 may remove the post-processed signal p1 by replacing the post-processed signal p1 with None data. The extraction unit 5 may extract the None data as extracted data d1. The database 40 stores the extracted data d1.

[0033] In the second case, an evaluation index e2 is used to evaluate the results of calculating a plurality of types of signals. The setting unit 7 converts the evaluation index e2 into an input criterion c2.

[0034] The input unit 2 acquires input signals x2 and x3. The input signal x2 is, for example, the natural potential of a platinum electrode relative to GND. GND is the reference potential of the measurement system. The natural potential is the potential generated by the oxidation-reduction equilibrium of a metal electrode immersed in an aqueous solution. The input signal x3 is the reference electrode potential, which is also used in pH measurement, for example. This is achieved by immersing a silver / silver chloride electrode (an electrode in which the surface of silver is oxidized with silver chloride) in KCL (potassium chloride solution) and maintaining a constant concentration of Ag ions and Cl ions, resulting in a stable equilibrium potential "Ag(+) + Cl(-) = AgCl," which is the reference potential for potential measurement.

[0035] When the change in the platinum electrode potential (E) relative to the reference electrode from the initial stage of the culture is defined as ΔE, the SNR for the rate of change in potential required to determine the change in the culture solution can be required as the evaluation index e2. If the evaluation index e2 is assumed to be 20 dB, the setting unit 7 calculates the SNR (input standard c2) for the potential using, for example, the following formula: ΔE(SNR)=20dB 20xLOG(ΔE / N)-3(dB)=20dB

[0036] Here, ΔE is the difference in potential. N is the base noise of the potential sensor. 3 dB is the degradation in SNR due to the difference calculation with the reference electrode. If ΔE=Ex10%, the SNR required for the transmitted light (S) is calculated to be 43 dB (=c1). By performing this calculation on the evaluation index e2, the setting unit 7 converts the evaluation index e2 into the input standard c2.

[0037] The preprocessing unit 3 subtracts the input signal x3 from the input signal x2 to generate a preprocessed signal f2, which is the potential of the input signal x2 (platinum electrode) relative to the reference potential of the input signal x3. The input signal x2 and the extracted data d2 may each be converted into a multi-signal by arranging multiple metal electrodes. The comparison unit 8 compares the preprocessed signal f2 with the input reference c2 to generate a comparison result.

[0038] If the pre-processed signal f2 does not satisfy the input criterion c2, the post-processing unit 4 performs post-processing operations such as averaging or moving averaging on the pre-processed signal f2 to generate a post-processed signal p2. The comparing unit 8 compares the post-processed signal p2 with the input criterion c2 to generate a comparison result.

[0039] If the pre-processed signal f2 satisfies the input criterion c2, the extraction unit 5 extracts the pre-processed signal f2 as extracted data d2. If the post-processed signal p2 satisfies the input criterion c2, the extraction unit 5 extracts the post-processed signal p2 as extracted data d2. If the post-processed signal p2 does not satisfy the input criterion c2, the extraction unit 5 removes the post-processed signal p2. For example, the extraction unit 5 may remove the post-processed signal p2 by replacing the post-processed signal p2 with None data. The extraction unit 5 may extract the None data as extracted data d2. The database 40 stores the extracted data d2.

[0040] In the third case, an evaluation index e3 is used to evaluate each of multiple types of signals. The input unit 2 acquires input signals x4 to x6. Each of the input signals x4 to x6 is an RGB signal (for example, R, G, and B luminance values ​​in the RGB color model). When an RGB signal is input, the final signal may be extracted as a ΔE*ab value in the L*a*b* color space. In this case, the SNR required for the ΔE*ab value is set as the evaluation index e3. The setting unit 7 sets the SNR required for each of the RGB signals as input standards c31 to c33 (input standard c3).

[0041] The preprocessing unit 3 performs preprocessing calculations such as gain adjustment and white balance adjustment on each of the input signals x4 to x6, generating preprocessed signals f3 to f5. For example, white balance adjustment for R, G, and B luminance values ​​is a calibration for individual variations. The preprocessed signal f3 corresponds to the input reference c31. The preprocessed signal f4 corresponds to the input reference c32. The preprocessed signal f5 corresponds to the input reference c33. The comparison unit 8 compares the preprocessed signal f3 with the input reference c31, the preprocessed signal f4 with the input reference c32, and the preprocessed signal f5 with the input reference c33, and generates comparison results.

[0042] If at least one of pre-processed signals f3 to f5 does not satisfy the corresponding input criteria c31 to c33, post-processing unit 4 performs post-processing operations such as averaging or moving averaging on pre-processed signals f3 to f5 to generate post-processed signals p3 to p5. Post-processed signal p1 corresponds to input criteria c31. Post-processed signal p2 corresponds to input criteria c32. Post-processed signal p3 corresponds to input criteria c33. Comparing unit 8 compares post-processed signal p3 with input criteria c31, post-processed signal p4 with input criteria c32, and post-processed signal p5 with input criteria c33 to generate comparison results.

[0043] If the pre-processed signals f3 to f5 satisfy the corresponding input criteria c31 to c33, the extraction unit 5 extracts the pre-processed signals f3 to f5 as extracted data d3. If the post-processed signals p3 to p5 satisfy the corresponding input criteria c31 to c33, the extraction unit 5 extracts the post-processed signals p3 to p5 as extracted data d3. The extraction unit 5 may extract data obtained by performing an output calculation or the like on the pre-processed signals f3 to f5 or the post-processed signals p3 to p5 as extracted data d3. The extraction unit 5 may calculate a ΔE*ab value in the L*a*b* color space using the pre-processed signals f3 to f5 or the post-processed signals p3 to p5 and extract this as extracted data d3. If at least one of the post-processed signals p3 to p5 does not satisfy the corresponding input criteria c31 to c33, the extraction unit 5 removes the post-processed signals p3 to p5. For example, the extraction unit 5 may remove the post-processed signals p3 to p5 by replacing the post-processed signals p3 to p5 with None data. The extraction unit 5 may extract the None data as extracted data d3. The database 40 stores the extracted data d3.

[0044] Figure 3 is a diagram showing an example of spectrophotometric measurement in biological production. In the example shown in Figure 3, spectrophotometric measurement is performed using a culture vessel 20, a light source 21, a flow cell 22, and a spectrometer 23. The light source 21 irradiates light onto the flow cell 22. The flow cell 22 draws and circulates the culture medium from the culture vessel 20. The transmitted light from the flow cell 22 is incident on the spectrometer 23. The spectrometer 23 outputs an optical spectrum.

[0045] Graph 24 shown in FIG. 3 schematically illustrates the output of spectrometer 23. Graph 24 illustrates signal level group L1 when spectroscopic measurements are performed multiple times, and signal level L2 in the dark. Even if the state of the culture medium flowing through flow cell 22 is constant, the signal level (signal output) fluctuates. As a result, signal level group L1 may have a fluctuation range F. This fluctuation range F is due to, for example, the installation conditions of hardware such as optical fiber related to spectroscopic measurements. In the example shown in FIG. 3, dark correction and relative intensity correction are performed as preprocessing calculations.

[0046] The pre-processing unit 3 performs dark correction on the optical spectrum as a pre-processing calculation. For example, the pre-processing unit 3 subtracts the signal level L2 in the dark from the signal level of the optical spectrum. This derives the signal level that changes depending on the state of the culture medium flowing through the flow cell 22.

[0047] The preprocessing unit 3 may further perform relative intensity correction on the dark-corrected signal as a preprocessing operation. For example, the preprocessing unit 3 calculates the overall relative intensity using a signal in a wavelength range R that is not responsive to changes in the medium and has a transmittance that is not zero (a non-zero signal level). "Not responsive to changes in the medium" means that the signal level does not fluctuate even when the medium changes. For example, the absorption pattern of each wavelength of light varies as it passes through an object. This is because the absorption wavelength differs depending on the molecular structure of the object. Wavelengths that do not react to any molecular structure are wavelengths that "do not respond to changes in the medium," i.e., wavelengths where "the signal level does not fluctuate even when the medium changes." An example of a wavelength range R that is not responsive to changes in the medium and has a transmittance that is not zero is a wavelength range around 1300 nm. The preprocessing unit 3 calculates the relative intensity relative to the signal level in the wavelength range R using the dark-corrected signal. That is, the preprocessing unit 3 generates the preprocessed signal by calculating the relative intensity relative to the signal level in a wavelength range that is not responsive to changes in the medium and has a transmittance that is not zero.

[0048] The optical spectroscopic measurement may be near-infrared spectroscopic measurement. Near-infrared spectroscopic measurement involves a wavelength range (e.g., 1250 to 1350 nm) that is not sensitive to organic molecular fluctuations and in which water absorption is not excessive. "Not sensitive to organic molecular fluctuations" means that the signal level does not fluctuate even when organic molecular fluctuations occur, or that the signal level is not affected by organic molecular fluctuations. "Water absorption is not excessively large" means that light is transmitted with a certain intensity, and the light intensity can be stably measured by the optical spectrometer 23. In other words, the light intensity (signal level) required for near-infrared spectroscopic measurement is not affected by water absorption. The appropriate absorbance range here depends on the measurement target, but is, for example, 0.05 to 1.5. A wavelength range in which water absorption is not excessively large may have an absorbance of 1.0 (absorbance of 90%) or less, for example. A wavelength range in which water absorption is not excessively large and that is not sensitive to organic molecular fluctuations is an example of wavelength range R. The pre-processing unit 3 calculates the relative intensity with respect to the signal level in the wavelength range using the dark-corrected signal. That is, the pre-processing unit 3 generates a pre-processed signal by calculating the relative intensity with respect to the signal level in the wavelength range R that does not respond to organic molecule fluctuations and has an excessively large absorption by water.

[0049] In the example shown in Figure 3, the input unit 2 and pre-processing unit 3 function as a partial data extraction system or sensing system. The data extraction system acquires an optical spectrum measured by optical spectroscopy in biological production. The data extraction system outputs a signal that calculates the relative intensity of the optical spectrum to the signal level of the optical spectrum in a wavelength range that does not respond to changes in the medium and where the transmittance is not zero.

[0050] Relative intensity correction will be described with reference to (a) to (c) of FIG. 4. (a) to (c) of FIG. 4 are graphs illustrating the extraction of changes in organic molecule concentration in near-infrared spectroscopy. In (a) to (c) of FIG. 4, the vertical axis represents signal level, and the horizontal axis represents wavelength. (a) to (c) of FIG. 4 show a wavelength range (e.g., 1250 to 1350 nm) that is not responsive to organic molecule fluctuations and where water absorption is not too large as the "non-fluctuating wavelength range." (a) to (c) of FIG. 4 also show the glucose calibration curve (1600 nm) as the "wavelength of the target substance."

[0051] Fig. 4(a) is a graph showing the signal level obtained by performing near-infrared spectroscopy measurement for several hours while tap water, instead of culture medium, was circulated through the flow cell 22. Fig. 4(a) shows that the signal level of the near-infrared spectroscopy fluctuates.

[0052] Figure 4(b) is a graph showing the signal level after dark correction has been performed on the signal level shown in Figure 4(a). In other words, Figure 4(b) is an example in which the signal level in dark conditions has been subtracted from the signal level shown in Figure 4(a). Comparing Figure 4(b) with Figure 4(a), it can be said that the offset has been eliminated by dark correction.

[0053] Figure 4(c) is a graph showing signal levels converted to relative intensities using the signal levels around 1300 nm (e.g., 1250 to 1350 nm) shown in Figure 4(b). Each of the signal levels shown in Figure 4(c) has a similar shape. In other words, it can be said that fluctuations in signal levels are eliminated by relative intensity correction. Therefore, the signal levels converted to relative intensities make it easy to capture changes in the wavelength range of the target substance.

[0054] FIG. 5 is a schematic diagram showing another example of the operation of the data extraction system. FIG. 5 shows an example of acquiring an input signal using the configuration shown in FIG. 3. In FIG. 5, evaluation indexes e4 and e5 for evaluating the SNR are used. The setting unit 7 sets an input criterion c4 corresponding to the evaluation index e4. The setting unit 7 sets an input criterion c5 corresponding to the evaluation index e5. The input unit 2 acquires an input signal x7 from the optical spectrometer 23. The input signal x7 is an optical spectrum. The setting unit 7 converts the evaluation indexes e4 and e5 into input criterion c4 and c5, similar to the method of calculating the input criterion c1 based on the evaluation index e1.

[0055] For example, the SNR for the rate of change in absorbance required to determine the change in turbidity of a certain aqueous solution may be required as the evaluation index e4. If the evaluation index e4 is assumed to be 10 dB, the setting unit 7 calculates the SNR (input standard c4) required for the transmitted light using, for example, the following formula: ΔA(SNR)=10dB 20xLOG(ΔS / N)-6(dB)=10dB

[0056] Here, ΔA is the difference in absorbance, ΔS is the difference in transmitted light intensity, and N is the base noise of the optical sensor. 6 dB is the SNR that deteriorates before ΔA is calculated from the transmitted light S. If ΔS = S x 4%, the SNR required for the transmitted light (S) is calculated to be 44 dB (= input standard c4). The setting unit 7 performs this calculation on the evaluation index e4 to convert it to the input standard c4. The setting unit 7 also performs a similar calculation on the evaluation index e5, for example, assuming ΔA (SNR) = 20 dB, to convert it to the input standard c5.

[0057] The pre-processing unit 3 performs dark correction on the input signal x7 to generate a pre-processed signal f6. If the pre-processed signal f6 does not satisfy the input criterion c4, the post-processing unit 4 performs post-processing on the pre-processed signal f6 to generate a post-processed signal p6. The input signal x7 that has undergone dark correction is finally extracted as extracted data d4.

[0058] The pre-processing unit 3 performs dark correction and relative intensity correction on the input signal x7 to generate a pre-processed signal f7. If the pre-processed signal f7 does not satisfy the input criterion c5, the post-processing unit 4 performs post-processing on the pre-processed signal f7 to generate a post-processed signal p7. The input signal x7 that has undergone dark correction and relative intensity correction is finally extracted as extracted data d5.

[0059] 5, extracted data d4 and d5 may be extracted by branching the process based on input signal x7. The extracted data d4 and d5 may be used depending on the purpose of data analysis when different substances are desired to be captured as changes in the culture medium.

[0060] The example shown in FIG. 5 can be considered a process for extracting data obtained by performing different pre-processing operations on the same signal. The pre-processing unit 3 generates a first pre-processed signal (pre-processed signal f6) that does not calculate relative intensity, and generates a second pre-processed signal (pre-processed signal f7) that does calculate relative intensity. The setting unit 7 sets a first input criterion (input criterion c4) that corresponds to the first pre-processed signal, and a second input criterion (input criterion c5) that corresponds to the second pre-processed signal. If the first pre-processed signal does not satisfy the first input criterion, the post-processing unit 4 performs a post-processing operation on the first pre-processed signal to generate a first post-processed signal (post-processed signal p6). If the first pre-processed signal satisfies the first input criterion, the extraction unit 5 extracts the first pre-processed signal as first extracted data (extracted data d4). If the first post-processed signal satisfies the first input criterion, the extraction unit 5 extracts the first post-processed signal as first extracted data. The post-processing unit 4 performs a post-processing operation on the second pre-processed signal to generate a second post-processed signal (post-processed signal p7) when the second pre-processed signal does not satisfy the second input criterion. The extraction unit 5 extracts the second pre-processed signal as second extracted data (extracted data d5) when the second pre-processed signal satisfies the second input criterion, and extracts the second post-processed signal as the second extracted data when the second post-processed signal satisfies the second input criterion.

[0061] FIG. 6 is a schematic diagram showing yet another example of the operation of the data extraction system. FIG. 6 shows an example in which different input criteria are set in addition to the processing shown in FIG. 5. In FIG. 6, evaluation indexes e4 to e7 for evaluating the SNR are used. The setting unit 7 sets an input criterion c4 corresponding to the evaluation index e4. The setting unit 7 sets an input criterion c5 corresponding to the evaluation index e5. The setting unit 7 sets an input criterion c6 corresponding to the evaluation index e6. The setting unit 7 sets an input criterion c7 corresponding to the evaluation index e7. The input unit 2 acquires an input signal x7 from the optical spectroscope 23. The input signal x7 is an optical spectrum.

[0062] For example, the setting unit 7 calculates the SNR (input standards c4 to c7) required for the transmitted light using the following formula. -Input standard c4 ΔA(SNR)=10dB 20xLOG(ΔS / N)-6(dB)=10dB -Input standard c5 ΔA(SNR)=20dB 20xLOG(ΔS / N)-6(dB)=20dB -Input standard c6 ΔA(SNR)=0dB 20xLOG(ΔS / N)-6(dB)=0dB -Input standard c7 ΔA(SNR)=10dB 20xLOG(ΔS / N)-6(dB)=10dB

[0063] Here, ΔA is the difference in absorbance, ΔS is the difference in transmitted light intensity, and N is the base noise of the optical sensor. 6 dB is the SNR degradation that occurs before ΔA is calculated from the transmitted light S. For input standard c4, if ΔS = Sx 4%, the SNR required for the transmitted light (S) is calculated to be 44 dB. For input standard c5, if ΔS = Sx 1%, the SNR required for the transmitted light (S) is calculated to be 66 dB. For input standard c6, if ΔS = Sx 4%, the SNR required for the transmitted light (S) is calculated to be 34 dB. For input standard c7, if ΔS = Sx 1%, the SNR required for the transmitted light (S) is calculated to be 54 dB. The setting unit 7 converts the evaluation indexes e4 to e7 into input standards c4 to c7 by performing these calculations on the evaluation indexes e4 to e7.

[0064] The pre-processing unit 3 performs dark correction on the input signal x7 to generate a pre-processed signal f6. If the pre-processed signal f6 does not satisfy the input criterion c4, the post-processing unit 4 performs post-processing on the pre-processed signal f6 to generate a post-processed signal p6. The input signal x7 that has undergone dark correction is finally extracted as extracted data d4.

[0065] The pre-processing unit 3 performs dark correction and relative intensity correction on the input signal x7 to generate a pre-processed signal f7. If the pre-processed signal f7 does not satisfy the input criterion c5, the post-processing unit 4 performs post-processing on the pre-processed signal f7 to generate a post-processed signal p7. The input signal x7 that has undergone dark correction and relative intensity correction is finally extracted as extracted data d5.

[0066] The pre-processing unit 3 performs dark correction on the input signal x7 to generate a pre-processed signal f8. The pre-processed signal f8 is identical to the pre-processed signal f6. If the pre-processed signal f8 does not satisfy the input criterion c6, the post-processing unit 4 performs post-processing on the pre-processed signal f8 to generate a post-processed signal p8. The input signal x7 that has undergone dark correction is finally extracted as extracted data d6.

[0067] The pre-processing unit 3 performs dark correction and relative intensity correction on the input signal x7 to generate a pre-processed signal f9. The pre-processed signal f9 is identical to the pre-processed signal f7. If the pre-processed signal f9 does not satisfy the input criterion c7, the post-processing unit 4 performs post-processing operations on the pre-processed signal f9 to generate a post-processed signal p9. The input signal x7 that has undergone dark correction and relative intensity correction is finally extracted as extracted data d7.

[0068] Input standards c4 and c6 have in common the fact that they are both used for comparison with pre-processed signals that have undergone dark correction, but their values ​​are different. In other words, input standards c4 and c6 are input standards corresponding to pre-processed signals that have undergone dark correction, and can be said to be mutually different input standards. If the SNR of input standard c4 is greater than the SNR of input standard c6, the SNR of extracted data d4 is higher than the SNR of extracted data d6, resulting in a more accurate signal. The response of extracted data d6 is higher than the response of extracted data d4. The response is determined based on the time required for post-processing calculations.

[0069] Input standards c5 and c7 have in common the fact that they are both used for comparison with preprocessed signals that have undergone dark correction and relative intensity correction, but their values ​​are different. In other words, input standards c5 and c7 are input standards corresponding to preprocessed signals that have undergone dark correction and relative intensity correction, and can be said to be different input standards. If the SNR of input standard c5 is greater than the SNR of input standard c7, the SNR of extracted data d5 is higher than the SNR of extracted data d7, resulting in a more accurate signal. The response of extracted data d7 is higher than the response of extracted data d5.

[0070] The example shown in FIG. 6 can be said to be a process of extracting data of different properties by setting different input criteria for the same signal. The setting unit 7 further sets another input criterion corresponding to the pre-processed signal. If the pre-processed signal does not satisfy the other input criterion, the post-processing unit 4 performs a post-processing operation on the pre-processed signal to generate another post-processed signal. If the pre-processed signal satisfies the other input criterion, the extraction unit 5 further extracts the other pre-processed signal as another extracted data, and if the other post-processed signal satisfies the other input criterion, the extraction unit 5 further extracts the other post-processed signal as another extracted data.

[0071] With reference to (a) to (d) of Figure 7 and (a) to (d) of Figure 8, an example of the improvement in the quality of data used in analysis will be described as an effect of the present disclosure. (a) to (d) of Figure 7 are graphs showing an example for explaining the effect of pre-processing calculations and setting input standards. (a) to (d) of Figure 7 show an example in which a culture experiment was conducted over several days and lactic acid was observed as the target substance.

[0072] Figure 7(a) shows an example of sampling and measuring lactic acid concentration using high performance liquid chromatography (HPLC). In Figure 7(a), the vertical axis represents lactic acid concentration, and the horizontal axis represents time. In Figure 7(a), inflection points r1 and r2 appear.

[0073] Figures 7(b) to (d) show examples of data extracted under different conditions based on data observed in real time every minute by near-infrared spectroscopy. In Figures 7(b) to (d), the vertical axis represents absorbance (difference from the initial value) and the horizontal axis represents time.

[0074] Figure 7(b) shows an example of extracting the value at 1700 nm (lactic acid calibration curve) from extracted data that has been subjected to SNR as the input standard and dark correction and relative intensity correction as preprocessing calculations. Comparing Figure 7(a) and (b) reveals that inflection point t1, which corresponds to inflection point r1, and inflection point t2, which corresponds to inflection point r2, have been captured.

[0075] Figure 7(c) shows an example of extracting the value at 1700 nm (lactic acid calibration curve) from extracted data in which SNR was applied as the input standard and dark correction was performed as a pre-processing calculation. In Figure 7(c), the signal increases monotonically, and changes in the lactic acid concentration cannot be seen. However, the monotonic increase captures the turbidity. This monotonic increase in turbidity has an extremely high correlation with the amount of product produced by the culture.

[0076] Figure 7(d) shows an example of the 1700 nm (lactic acid calibration curve) value extracted from extracted data that was not subjected to input criteria and underwent dark correction and relative intensity correction as preprocessing calculations. In Figure 7(d), the inflection points t11 and t21 are buried in noise. Such signals can be said to be noisy but highly sensitive signals. In the case of cultures where large changes occur in an instant, it is also useful to use highly sensitive signals in data analysis.

[0077] Figure 8 (a) to (d) are graphs showing another example for explaining the effect of pre-processing calculations and setting input standards. Figure 8 (a) to (d) show an example in which a culture experiment was conducted over several days and glucose was observed as the target substance.

[0078] Figure 8(a) shows an example of sampling and measuring glucose concentration by HPLC. In Figure 8(a), the vertical axis represents glucose concentration and the horizontal axis represents time. In Figure 8(a), inflection points r3 and r4 appear.

[0079] Figures 8(b) to (d) show examples of data extracted under different conditions based on data observed in real time every minute by near-infrared spectroscopy. In Figures 8(b) to (d), the vertical axis represents absorbance (difference from the initial value) and the horizontal axis represents time.

[0080] Figure 8(b) shows an example of extracting the value at 1600 nm (glucose calibration curve) from extracted data that has been subjected to SNR as the input criterion and dark correction and relative intensity correction as preprocessing operations. Comparing Figure 8(a) and (b) reveals that inflection point t3, which corresponds to inflection point r3, and inflection point t4, which corresponds to inflection point r4, have been captured.

[0081] Figure 8(c) shows an example of extracting the value at 1600 nm (glucose calibration curve) from extracted data in which SNR was applied as the input standard and dark correction was performed as a pre-processing calculation. In Figure 8(c), the signal increases monotonically, and changes in glucose concentration cannot be detected. However, the monotonic increase captures turbidity. This monotonic increase in turbidity has an extremely high correlation with the amount of product produced by the culture.

[0082] Figure 8(d) shows an example of the 1600 nm (glucose calibration curve) value extracted from extracted data that was not subjected to input criteria and underwent dark correction and relative intensity correction as preprocessing calculations. In Figure 8(d), the inflection points t31 and t41 are buried in noise. Such signals are noisy but highly sensitive. In the case of cultures where large changes occur in a short time, it is useful to use highly sensitive signals in data analysis.

[0083] [Example] FIG. 9 is an overall view showing an example of application of the data extraction system 1 according to the embodiment. The biological production system 50 shown in FIG. 9 includes a culture tank 20, a plurality of sensors 30 (a pH sensor 31, a DO sensor 32, a temperature sensor 33, a near-infrared sensor 34, an RGB sensor 35, and a potential sensor 36), a culture control device 51, a sensor control device 52, a control terminal 53, and a cloud system 54. The control terminal 53 is communicably connected to the culture control device 51 and the sensor control device 52 via a communication network 59. The cloud system 54 is communicably connected to the culture control device 51 and the sensor control device 52 via the communication network 59. The communication network 59 may be wired or wireless. The communication network 59 may be a non-dedicated line such as the Internet or a mobile communication network, or may be a dedicated line.

[0084] The pH sensor 31, the DO sensor 32, and the temperature sensor 33 acquire data essential for culturing (hereinafter referred to as "cultivation data"). The culture data may include the agitator rotation speed (Agit) acquired by a torque sensor or the like. The culture control device 51 controls the culture tank 20, the pH sensor 31, the DO sensor 32, and the temperature sensor 33. The culture control device 51 adds sensor data information and the like to signals output from the pH sensor 31, the DO sensor 32, and the temperature sensor 33, and transmits the signals to the cloud system 54. Examples of the sensor data information include, but are not limited to, a version number and a serial number.

[0085] The near-infrared sensor 34, RGB sensor 35, and potential sensor 36 are used to implement AI (Artificial Intelligence) in biological production. Implementing AI in biological production involves, for example, using a machine learning model to predict optimal control values ​​in real time and automatically controlling cultivation in accordance with the predictions. The sensor control device 52 controls the settings of the near-infrared sensor 34, RGB sensor 35, and potential sensor 36. The sensor control device 52 adds sensor data information and the like to the signals output from the near-infrared sensor 34, RGB sensor 35, and potential sensor 36 and transmits them to the cloud system 54.

[0086] The control terminal 53 is one or more computers used by a user of the biological production system 50. The control terminal 53 is, for example, but not limited to, a personal computer. The control terminal 53 accepts various settings and controls, for example, based on a Web UI (User Interface). The control terminal 53 changes the settings of the culture control device 51, the sensor control device 52, and the cloud system 54. The control terminal 53 controls the culture control device 51 and the sensor control device 52.

[0087] The cloud system 54 includes a data extraction system 1, a database 40, and a data analysis system 55. The database 40 may store signals received from the culture control device 51 and the sensor control device 52 as data. The data extraction system 1 may acquire and process signals read from the database 40 as input signals, and store the extracted data in the database 40.

[0088] There are multiple pieces of extracted data for each sensor 30. Furthermore, the extracted data is extracted as multiple pieces of data depending on the patterns of pre-processing operations, input criteria, and post-processing operations, resulting in a very large number of dimensional parameters. The data analysis system 55 processes the extracted data read from the database 40 and performs data analysis. The data analysis system 55 has, as functional elements, a dimension reduction unit 56, an explanatory variable extraction unit 57, and a machine learning unit 58.

[0089] The dimension reduction unit 56 reduces the dimensions of the extracted data, for example, by principal component analysis. The explanatory variable extraction unit 57 determines explanatory variables from the dimension-reduced signal based on correlation with the objective variable in machine learning and regression analysis. The machine learning unit 58 performs machine learning using the determined explanatory variables to create a machine learning model. The biological production system 50 automatically controls cultivation in real time using the machine learning model.

[0090] The extracted data of the present disclosure is data that meets input standards, so the quality of the data is guaranteed. By using such extracted data for machine learning, efficient machine learning can be achieved.

[0091] An example of the operation of the biological production system 50 will be described with reference to Fig. 10. Fig. 10 is a diagram showing an example of the overall operation of the biological production system 50 including the data extraction system 1.

[0092] In step S1, the control terminal 53 sets parameters for the culture control device 51, the sensor control device 52, and the cloud system 54. Examples of the parameters include, but are not limited to, the culture type, the culture time, the sensor setting values ​​(sampling interval, integral time, and gain), and an ID that identifies the culture data.

[0093] In step S2, the culture control device 51, the sensor control device 52, and the cloud system 54 each perform various settings in response to parameter settings from the control terminal 53.

[0094] In step S3, the control terminal 53 performs start control to start the operation of the culture control device 51 and the sensor control device 52. For example, the control terminal 53 may perform start control by transmitting a command to instruct the culture control device 51 and the sensor control device 52 to start the operation via a Web UI.

[0095] In step S4, the culture control device 51 and the sensor control device 52 each start operation in response to a start control from the control terminal 53. The culture control device 51 controls the pH sensor 31, the DO sensor 32, and the temperature sensor 33 to start culture. The sensor control device 52 controls the near-infrared sensor 34, the RGB sensor 35, and the potential sensor 36 to start measurement in the culture tank 20.

[0096] In step S5, the control terminal 53 performs stop control to stop the operation of the culture control device 51 and the sensor control device 52. For example, the control terminal 53 may perform stop control by transmitting a command to stop the operation to the culture control device 51 and the sensor control device 52 via a Web UI.

[0097] In step S6, the culture control device 51 and the sensor control device 52 transfer the collected data to the cloud system 54 in response to start control and stop control from the control terminal 53, respectively. For example, the culture control device 51 adds sensor data information and the like to signals output from the pH sensor 31, the DO sensor 32, and the temperature sensor 33, and transmits the signals to the cloud system 54. The sensor control device 52 adds sensor data information and the like to signals output from the near-infrared sensor 34, the RGB sensor 35, and the potential sensor 36, and transmits the signals to the cloud system 54. The culture control device 51 and the sensor control device 52 may transmit various data to the cloud system 54 at predetermined time intervals during the culture period.

[0098] In step S7, the database 40 stores various data received from the culture control device 51 and the sensor control device 52. When the cloud system 54 receives various data from the culture control device 51 and the sensor control device 52, the cloud system 54 may start subsequent processing in an event-driven manner.

[0099] In step S8, the data extraction system 1 reads various data as input signals from the database 40 and extracts extracted data. Detailed processing of the data extraction system 1 will be described later. The data extraction system 1 stores the extracted data in the database 40.

[0100] In step S9, the data analysis system 55 performs machine learning based on the extracted data as non-real-time processing. The dimension reduction unit 56 reduces the dimensions of the extracted data, for example, by principal component analysis. The explanatory variable extraction unit 57 determines explanatory variables from the dimension-reduced signals based on correlation with the target variable in machine learning and regression analysis. The machine learning unit 58 performs machine learning using the determined explanatory variables to create machine learning models such as a prediction model for predicting the amount of lactic acid, glucose, and product production, and a control model for controlling the culture.

[0101] The processes of steps S1 to S9 may be executed repeatedly. For example, the biological production system 50 may collect data for each parameter pattern and create a machine learning model.

[0102] In step S10, the cloud system 54 performs the culture prediction as a real-time process. For example, the machine learning unit 58 performs the culture prediction using a prediction model.

[0103] In step S11, the cloud system 54 performs culture control as real-time processing. For example, the machine learning unit 58 uses the culture prediction results and the control model to create control variables for controlling the culture control device 51. The data analysis system 55 transmits the control variables to the culture control device 51.

[0104] In step S12, the culture control device 51 controls the amount of alkali to be added to control the pH, the amount of oxygen to be injected to control the DO, the temperature, Agit, etc. in response to the control variables received from the cloud system 54. The results are reflected in the culture tank 20, thereby enabling automatic control of the culture.

[0105] An example of the operation of the data extraction system 1 will be described with reference to Fig. 11. Fig. 11 is a diagram showing an example of the operation of the data extraction system 1. In Fig. 11, an example of measuring culture data, near-infrared light data, RGB data, and potential data will be described. In one example, the culture data is DO, pH, Agit, and temperature. The following description will be given assuming that the setting unit 7 has already set the input criteria.

[0106] In step S101, the multiple sensors 30 (for example, the pH sensor 31, the DO sensor, the torque sensor, and the temperature sensor 33) transmit signals indicating the measurement results of DO, pH, AgIt, and temperature to the culture control device 51. The culture control device 51 adds sensor data information of the pH sensor 31, the DO sensor, the torque sensor, and the temperature sensor 33 to the signals indicating the measurement results and transmits the signals to the cloud system 54. The database 40 stores the signals (culture data) received from the culture control device 51. The input unit 2 of the data extraction system 1 reads out the culture data from the database to obtain the culture data as an input signal.

[0107] In step S102, the near-infrared sensor 34 transmits a signal indicating the measurement result of the optical spectrum to the sensor control device 52. The sensor control device 52 adds sensor data information of the near-infrared sensor 34 and the like to the signal indicating the measurement result and transmits the signal to the cloud system 54. The database 40 stores the signal (near-infrared light data) received from the sensor control device 52. The input unit 2 of the data extraction system 1 reads the near-infrared light data from the database 40 to obtain the near-infrared light data as an input signal.

[0108] In step S103, the RGB sensor 35 transmits a signal indicating the RGB measurement result to the sensor control device 52. The sensor control device 52 adds sensor data information of the RGB sensor 35 to the signal indicating the measurement result and transmits the signal to the cloud system 54. The database 40 stores the signal (RGB data) received from the sensor control device 52. The input unit 2 of the data extraction system 1 reads the RGB data from the database 40 to obtain the RGB data as an input signal.

[0109] In step S104, the potential sensor 36 transmits a signal indicating the measurement result of the potential to the sensor control device 52. The sensor control device 52 adds sensor data information of the potential sensor 36 to the signal indicating the measurement result and transmits the signal to the cloud system 54. The database 40 stores the signal (potential data) received from the sensor control device 52. The input unit 2 of the data extraction system 1 reads the potential data from the database 40 to obtain the potential data as an input signal.

[0110] In step S105, the pre-processing unit 3 performs pre-processing calculations such as gain adjustment and calibration on the input signal representing the culture data.

[0111] In step S106, the preprocessing unit 3 performs dark correction on the input signal representing the near-infrared light data.

[0112] In step S107, the preprocessing unit 3 performs dark correction and relative intensity correction on the input signal, which is near-infrared light data.

[0113] In step S108, the preprocessing unit 3 performs preprocessing operations such as gain adjustment and calibration on the input signal representing RGB data.

[0114] In step S109, the pre-processing unit 3 performs pre-processing operations such as gain adjustment and calibration on the input signal indicating the potential data.

[0115] In step S110, the preprocessing unit 3 generates a preprocessed signal relating to the culture data by further calculating an SNR using a signal obtained by performing a preprocessing operation on an input signal representing the culture data.

[0116] In step S111, the pre-processing unit 3 generates a first pre-processed signal relating to the near-infrared light data by further calculating the SNR using a signal obtained by performing dark correction on the input signal representing the near-infrared light data.

[0117] In step S112, the pre-processing unit 3 generates a second pre-processed signal related to the near-infrared light data by further calculating the SNR using a signal that has been subjected to dark correction and relative intensity correction on the input signal representing the near-infrared light data.

[0118] In step S113, the preprocessing unit 3 generates a preprocessed signal relating to the RGB data by further calculating an SNR using a signal obtained by performing preprocessing operations on the input signal representing the RGB data.

[0119] In step S114, the preprocessing unit 3 generates a preprocessed signal relating to the potential data by further calculating an SNR using a signal obtained by performing a preprocessing operation on the input signal indicating the potential data.

[0120] The method of calculating the SNR in steps S110 to S114 is not limited. In one example, the pre-processing unit 3 may measure the noise level (N) of each sensor 30 for each range in advance, and calculate the SNR [dB] using the signal level (S) by 20log(S / N). In another example, the pre-processing unit 3 may obtain a signal (S1) obtained by removing noise from the actual signal level (S), extract the noise (N) by subtracting the signal (S1) obtained by removing noise from the actual signal level (S), and calculate the SNR [dB] by 20log(S1 / N).

[0121] In step S115, the comparison unit 8 compares the preprocessed signal relating to the culture data with the input reference corresponding to the signal to generate a comparison result relating to the culture data.

[0122] In step S116, the comparison unit 8 compares the first pre-processed signal relating to the near-infrared light data with a first input reference corresponding to the signal to generate a first comparison result relating to the near-infrared light data.

[0123] In step S117, the comparison unit 8 compares the second pre-processed signal relating to the near-infrared light data with a second input reference corresponding to the signal to generate a second comparison result relating to the near-infrared light data.

[0124] In step S118, the comparison unit 8 compares the preprocessed signal for the RGB data with the input reference corresponding to the signal to generate a comparison result for the RGB data.

[0125] In step S119, the comparison unit 8 compares the preprocessed signal relating to the potential data with the input reference corresponding to the signal, and generates a comparison result relating to the potential data.

[0126] In step S120, if the preprocessed signal for the culture data does not satisfy the input criteria corresponding to the signal, the postprocessing unit 4 performs a postprocessing operation to obtain an arithmetic average or a moving average of the preprocessed signal to generate a postprocessed signal for the culture data. If the preprocessed signal for the culture data satisfies the input criteria corresponding to the signal, the postprocessing unit 4 does not perform the postprocessing operation.

[0127] In step S121, the abnormality determination unit 9 determines an abnormality in the culture based on the input signal. The abnormality determination unit 9 may detect the abnormality based on a pre-processed signal related to the culture data or a post-processed signal related to the culture data. The processing of step S121 may be performed before step S120.

[0128] The abnormality determination unit 9 may detect an abnormality in the culture medium based on a change (decrease) in the amount of dissolved oxygen. For example, the abnormality determination unit 9 may detect a sudden drop in DO by comparing the fluctuation in DO over a predetermined time period with a predetermined threshold value, etc. A sudden drop in DO may indicate that solidification of the culture medium has occurred. However, if Agit is rotating at a low speed, this does not necessarily mean that solidification of the culture medium has occurred, so the abnormality determination unit 9 may compare Agit with a threshold value corresponding to Agit.

[0129] The abnormality determination unit 9 may detect an abnormality in the electrode potential based on the correlation between the amount of dissolved oxygen or pH and the electrode potential. The abnormality determination unit 9 may monitor the correlation between DO or pH and the electrode potential, and determine that an electrode potential that is out of correlation is in an abnormal state. For example, the smaller (decreasing) the DO value or the larger (increasing) the pH value is, the more alkaline the culture medium is becoming. In potential measurement, there is a correlation that the more alkaline the culture medium becomes, the lower the electrode potential tends to be. The abnormality determination unit 9 may determine that the electrode potential is in an abnormal state when it behaves in a manner that differs from this correlation.

[0130] When an abnormality is detected, the abnormality determination unit 9 may send an error to the control terminal 53 or control the sensor 30 as an abnormality processing. The abnormality determination unit 9 may replace all measurement data with None data as an abnormality processing. For example, the abnormality determination unit 9 replaces the culture data, near-infrared light data, RGB data, and potential data with None data.

[0131] In step S122, if the first pre-processed signal for the near-infrared light data does not satisfy the first input criterion corresponding to the signal, the post-processing unit 4 performs a post-processing operation to obtain an arithmetic average or a moving average for the signal to generate a first post-processed signal for the near-infrared light data. If the first pre-processed signal for the near-infrared light data satisfies the first input criterion corresponding to the signal, the post-processing unit 4 does not perform the post-processing operation.

[0132] In step S123, if the second pre-processed signal for the near-infrared light data does not satisfy the input criterion corresponding to the signal, the post-processing unit 4 performs a post-processing operation to obtain an arithmetic average or a moving average on the signal to generate a second post-processed signal for the near-infrared light data. If the second pre-processed signal for the near-infrared light data satisfies the second input criterion corresponding to the signal, the post-processing unit 4 does not perform the post-processing operation.

[0133] In step S124, if the pre-processed signal for RGB data does not satisfy the input criteria corresponding to the signal, the post-processing unit 4 performs a post-processing operation to obtain an arithmetic average or a moving average on the pre-processed signal to generate a post-processed signal for RGB data. If the pre-processed signal for RGB data satisfies the input criteria corresponding to the signal, the post-processing unit 4 does not perform the post-processing operation.

[0134] In step S125, if the pre-processed signal for the potential data does not satisfy the input criteria corresponding to the signal, the post-processing unit 4 performs a post-processing operation to obtain an arithmetic average or a moving average of the pre-processed signal to generate a post-processed signal for the potential data. If the pre-processed signal for the potential data satisfies the input criteria corresponding to the signal, the post-processing unit 4 does not perform the post-processing operation.

[0135] In one example, the input criterion is SNR. When the sensor 30 detects changes in the culture medium, the difference from the initial value is monitored, so the signal level of the input signal is expected to be very small. If the SNR of the input criterion is 10 dB and the SNR of the pre-processed signal is 0 dB, the pre-processed signal does not meet the input criterion. If an input signal is obtained once per minute, the post-processing unit 4 obtains an arithmetic average using the input signal for 10 minutes (10 times). This can eliminate the 10 dB difference in the comparison result.

[0136] In another example, if the SNR of the input standard is 20 dB and the SNR of the preprocessed signal is 0 dB, the preprocessed signal does not meet the input standard. If an input signal is obtained once per minute, the post-processing unit 4 obtains an arithmetic average using 100 input signals over 10 minutes. This eliminates the 20 dB difference in the comparison result. In other words, it takes 1000 minutes (approximately 16.7 hours) to obtain the input signal for post-processing calculation. If the signal level of the input signal becomes 10 times higher (the difference from the initial value is 10 times higher) after one hour, it becomes possible to obtain extracted data of 20 dB every 10 minutes.

[0137] In step S126, if the pre-processed signal for the culture data satisfies the input criteria corresponding to the signal, the extraction unit 5 extracts the pre-processed signal for the culture data as extracted data for the culture data. If the post-processed signal for the culture data satisfies the input criteria corresponding to the signal, the extraction unit 5 extracts the post-processed signal for the culture data as extracted data for the culture data. If the post-processed signal for the culture data does not satisfy the input criteria corresponding to the signal, the extraction unit 5 removes the post-processed signal for the culture data. For example, the extraction unit 5 may remove the post-processed signal for the culture data by replacing the post-processed signal for the culture data with None data. The extraction unit 5 may extract the None data as extracted data for the culture data.

[0138] In step S127, if the first pre-processed signal for the near-infrared light data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the absorbance using the first pre-processed signal for the near-infrared light data. The extraction unit 5 extracts the absorbance as first extracted data for the near-infrared light data. If the first post-processed signal for the near-infrared light data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the absorbance using the first post-processed signal for the near-infrared light data. The extraction unit 5 extracts the absorbance as first extracted data for the near-infrared light data. If the first post-processed signal for the near-infrared light data does not satisfy the first input criteria corresponding to the signal, the extraction unit 5 removes the first post-processed signal for the near-infrared light data. For example, the extraction unit 5 may remove the first post-processed signal for the near-infrared light data by replacing the first post-processed signal for the near-infrared light data with None data. The extraction unit 5 may extract the None data as the first extracted data related to the near-infrared light data.

[0139] In step S127, if the second pre-processed signal for the near-infrared light data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the absorbance using the second pre-processed signal for the near-infrared light data. The extraction unit 5 extracts the absorbance as second extracted data for the near-infrared light data. If the second post-processed signal for the near-infrared light data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the absorbance using the second post-processed signal for the near-infrared light data. The extraction unit 5 extracts the absorbance as second extracted data for the near-infrared light data. If the second post-processed signal for the near-infrared light data does not satisfy the second input criteria corresponding to the signal, the extraction unit 5 removes the second post-processed signal for the near-infrared light data. For example, the extraction unit 5 may remove the second post-processed signal for the near-infrared light data by replacing the second post-processed signal for the near-infrared light data with None data. The extraction unit 5 may extract the None data as second extracted data related to the near-infrared light data.

[0140] In step S128, if the pre-processed signal for the RGB data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the ΔE*ab value in the L*a*b* color space using the pre-processed signal for the RGB data. The extraction unit 5 extracts the ΔE*ab value as the extracted data for the RGB data. If the post-processed signal for the RGB data satisfies the input criteria corresponding to the signal, the extraction unit 5 calculates the ΔE*ab value in the L*a*b* color space using the post-processed signal for the RGB data. The extraction unit 5 extracts the ΔE*ab value as the extracted data for the RGB data. If the post-processed signal for the RGB data does not satisfy the input criteria corresponding to the signal, the extraction unit 5 removes the post-processed signal for the RGB data. For example, the extraction unit 5 may remove the post-processed signal for the RGB data by replacing it with None data. The extraction unit 5 may extract the None data as the extracted data for the RGB data.

[0141] In step S129, if the pre-processed signal related to the potential data satisfies the input criteria corresponding to the signal, the extraction unit 5 extracts the pre-processed signal related to the potential data as extracted data related to the potential data. If the post-processed signal related to the potential data satisfies the input criteria corresponding to the signal, the extraction unit 5 extracts the post-processed signal related to the potential data as extracted data related to the potential data. If the post-processed signal related to the potential data does not satisfy the input criteria corresponding to the signal, the extraction unit 5 removes the post-processed signal related to the potential data. For example, the extraction unit 5 may remove the post-processed signal related to the potential data by replacing the post-processed signal related to the potential data with None data. The extraction unit 5 may extract the None data as extracted data related to the potential data.

[0142] In step S130, the database 40 stores each piece of extracted data. The extracted data stored in the database 40 can be used for data analysis. The extracted data may be used for machine learning by the data analysis system 55 or the like. For example, if the extracted data is used as learning data in machine learning, the efficiency of building a predictive model is improved. The abnormality determination unit 9 may determine that an input signal, a pre-processed signal, or a post-processed signal deviates from the predictive model.

[0143] [Hardware configuration] Fig. 12 is a diagram showing an example of a hardware configuration related to the data extraction system 1. Fig. 12 shows a computer 100 that functions as the data extraction system 1. The computer 100 has a processor 101, a main memory unit 102, an auxiliary memory unit 103, a communication control unit 104, an input device 105, and an output device 106. The data extraction system 1 is configured by one or more computers 100 configured by these pieces of hardware and software such as programs.

[0144] When the data extraction system 1 is configured by a plurality of computers 100, these computers 100 may be connected locally or via a communication network 59 such as the Internet or an intranet.

[0145] The processor 101 is a CPU (Central Processing Unit) that executes an operating system, application programs, etc. The main memory 102 is composed of a ROM (Read Only Memory) and a RAM (Random Access Memory). The auxiliary memory 103 is a storage medium composed of a hard disk, flash memory, etc. The auxiliary memory 103 generally stores a larger amount of data than the main memory 102. The communication control unit 104 is composed of a network card or a wireless communication module. At least a part of the communication function with other devices in the data extraction system 1 may be realized by the communication control unit 104. The input device 105 is composed of a keyboard, a mouse, a touch panel, a microphone for voice input, etc. The output device 106 is composed of a display, a printer, etc.

[0146] The auxiliary storage unit 103 stores in advance a program 110 (data extraction program) and data necessary for processing. The program 110 causes the computer 100 to execute each functional element of the data extraction system 1. The program 110 causes, for example, processing related to the above-mentioned data extraction method to be executed in the computer 100. For example, the program 110 is read by the processor 101 or the main storage unit 102, and causes at least one of the processor 101, the main storage unit 102, the auxiliary storage unit 103, the communication control unit 104, the input device 105, and the output device 106 to operate. For example, the program 110 reads and writes data from and to the main storage unit 102 and the auxiliary storage unit 103.

[0147] The program 110 may be provided in the form of being recorded on a tangible storage medium such as a CD-ROM, a DVD-ROM, a semiconductor memory, etc. The program 110 may be provided via the communication network 59 as a data signal.

[0148] As described above, a data extraction system 1 according to one aspect of the present disclosure comprises a setting unit 7 that sets input criteria that indicate the performance required for data analysis; an input unit 2 that acquires an input signal that indicates the measurement results of an object to be measured in biological production; a pre-processing unit 3 that performs pre-processing operations on the input signal to compare it with the input criteria and generate a pre-processed signal; a post-processing unit 4 that, if the pre-processed signal does not satisfy the input criteria, performs post-processing operations on the pre-processed signal using another signal that indicates the measurement results of the object to be measured and generates a post-processed signal; and an extraction unit 5 that extracts the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracts the post-processed signal as extracted data if the post-processed signal satisfies the input criteria.

[0149] A data extraction method according to one aspect of the present disclosure is executed by a data extraction system including at least one processor. The data extraction method includes the steps of: setting input criteria indicating performance required for data analysis; acquiring an input signal indicating a measurement result of a measurement target in biological production; performing a pre-processing operation on the input signal to compare it with the input criteria to generate a pre-processed signal; if the pre-processed signal does not satisfy the input criteria, performing a post-processing operation on the pre-processed signal using another signal indicating the measurement result of the measurement target to generate a post-processed signal; and extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input criteria.

[0150] A data extraction program according to one aspect of the present disclosure causes a computer to execute the following steps: setting input standards that indicate the performance required for data analysis; acquiring an input signal that indicates the measurement results of an object to be measured in biological production; performing a pre-processing operation on the input signal to compare it with the input standard and generate a pre-processed signal; if the pre-processed signal does not satisfy the input standard, performing a post-processing operation on the pre-processed signal using another signal that indicates the measurement results of the object to be measured and generate a post-processed signal; and extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input standard, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input standard.

[0151] In a data extraction system, a data extraction method, and a data extraction program according to one aspect of the present disclosure, a pre-processing operation is performed on an input signal to generate a pre-processed signal. Then, the pre-processed signal is compared with an input standard indicating performance required for data analysis. If the pre-processed signal satisfies the input standard, the pre-processed signal is extracted as extracted data. On the other hand, if the pre-processed signal does not satisfy the input standard, a post-processing operation is performed on the pre-processed signal to generate a post-processed signal. Then, the post-processed signal is compared with the input standard. If the post-processed signal satisfies the input standard, the post-processed signal is extracted as extracted data. This makes it possible to extract extracted data with guaranteed quality. As a result, it is possible to improve the quality of data used for analysis.

[0152] The input signal is an optical spectrum measured by spectrophotometric measurement in biological production. The preprocessing unit 3 generates a preprocessing signal by calculating the relative intensity of the signal level in a wavelength range that does not respond to changes in the culture medium and does not have zero transmittance. In conventional biological production, changes in the culture medium, etc., are captured while suppressing fluctuation noise by performing a second differentiation of the spectral signal in the wavelength direction. However, this method makes it difficult to directly capture wavelength changes because of the second differentiation. In contrast, the data extraction system disclosed herein calculates the relative intensity of the signal level in a wavelength range that does not respond to changes in the culture medium and does not have zero transmittance. The preprocessing signal indicating the relative intensity eliminates fluctuations in the signal level due to hardware installation conditions, etc. This improves the detection accuracy of target substances with low sensitivity. As a result, the quality of the data used for analysis can be improved.

[0153] The optical spectroscopic measurement is near-infrared spectroscopic measurement. The pre-processing unit 3 generates a pre-processing signal by calculating the relative intensity of the signal level in a wavelength range that is not sensitive to organic molecular fluctuations and in which the light intensity required for near-infrared spectroscopic measurement is not affected by water absorption. In this case, it is possible to detect target substances with low sensitivity, such as glucose concentration or lactic acid concentration.

[0154] The pre-processing unit 3 generates a first pre-processed signal in which the relative intensity has been calculated, and generates a second pre-processed signal in which the relative intensity has not been calculated. The setting unit 7 sets a first input criterion that is an input criterion corresponding to the first pre-processed signal, and sets a second input criterion that is an input criterion corresponding to the second pre-processed signal. The post-processing unit 4 performs a post-processing operation on the first pre-processed signal to generate a first post-processed signal if the first pre-processed signal does not satisfy the first input criterion. The extraction unit 5 extracts the first pre-processed signal as first extracted data if the first pre-processed signal satisfies the first input criterion, and extracts the first post-processed signal as first extracted data if the first post-processed signal satisfies the first input criterion. The post-processing unit 4 performs a post-processing operation on the second pre-processed signal to generate a second post-processed signal if the second pre-processed signal does not satisfy the second input criterion. The extraction unit 5 extracts the second pre-processed signal as second extracted data when the second pre-processed signal satisfies the second input criterion, and extracts the second post-processed signal as second extracted data when the second post-processed signal satisfies the second input criterion. In this case, extracted data based on the first pre-processed signal without calculating the relative intensity and second extracted data based on the second pre-processed signal with calculated relative intensity are extracted. In other words, data obtained by performing different pre-processing operations on the same signal are extracted. The first extracted data and the second extracted data can be used depending on the purpose of data analysis when different substances are desired to be detected as changes in the culture medium. This improves the convenience of data analysis.

[0155] The input criterion is the signal-to-noise ratio. By adopting SNR as the input criterion, unwanted noise components are removed, thereby improving the efficiency and accuracy of data analysis.

[0156] The post-processing unit 4 performs an averaging or a moving average as a post-processing operation, which improves the required accuracy of the post-processed signal.

[0157] The setting unit 7 further sets an input criterion different from the input criterion. If the preprocessed signal does not satisfy the different input criterion, the post-processing unit 4 performs a post-processing operation on the preprocessed signal to further generate another post-processed signal. If the preprocessed signal satisfies the different input criterion, the extraction unit 5 further extracts another preprocessed signal as another extracted data, and if the other post-processed signal satisfies the different input criterion, the extraction unit 5 further extracts another post-processed signal as another extracted data. In this case, extracted data based on the input criterion and another extracted data based on the other input criterion are extracted. In other words, by setting different input criteria for the same signal, data of different properties is extracted.

[0158] The system is equipped with an anomaly determination unit 9 that determines an anomaly based on the input signal, pre-processed signal, or post-processed signal, and invalidates the extracted data if an anomaly is detected. This allows extracted data with guaranteed quality to be extracted. As a result, the quality of the data used for analysis can be improved.

[0159] The abnormality determination unit 9 detects an abnormality in the culture medium based on a decrease in the amount of dissolved oxygen. In this case, solidification of the culture medium can be detected with high accuracy.

[0160] The abnormality determination unit 9 detects an abnormality in the electrode potential based on the correlation between the amount of dissolved oxygen or pH and the electrode potential. In this case, an abnormality in which the culture medium becomes alkaline can be detected with high accuracy.

[0161] A data extraction system according to one aspect of the present disclosure acquires an optical spectrum measured by optical spectroscopy in biological production, and outputs a signal that calculates the relative intensity of the optical spectrum to the signal level of the optical spectrum in a wavelength range that does not respond to changes in the culture medium and where the transmittance is not zero.

[0162] In a data extraction system according to one aspect of the present disclosure, the relative intensity is calculated relative to the signal level in a wavelength range that is not responsive to changes in the medium and does not have zero transmittance. The signal indicating the relative intensity eliminates fluctuations in the signal level due to factors such as hardware installation conditions. This improves the detection accuracy of target substances with low sensitivity. As a result, the quality of data used in analysis can be improved.

[0163] [Variations] The present disclosure is not necessarily limited to the above-described embodiments, and various modifications are possible without departing from the spirit of the present disclosure.

[0164] In bioproduction, the evaluation index may be constant or may change. For example, the evaluation index may change depending on the substance to be evaluated from the start to the end of the culture. In one example, the evaluation index may correspond to substance A in the early stage of the culture, substance B in the middle stage of the culture, and substance C in the later stage of the culture. In this case, the data extraction system extracts extracted data focusing on each process.

[0165] [Note] The following is a summary of the present disclosure. [1] a setting unit for setting input criteria indicating performance required for data analysis; an input unit that acquires an input signal indicating a measurement result of a measurement object in biological production; a pre-processing unit that performs a pre-processing operation on the input signal for comparison with the input reference and generates a pre-processed signal; a post-processing unit that, when the pre-processed signal does not satisfy the input criterion, performs a post-processing operation on the pre-processed signal using another signal that indicates the measurement result of the measurement object, thereby generating a post-processed signal; an extraction unit that extracts the pre-processed signal as extracted data when the pre-processed signal satisfies the input criterion, and extracts the post-processed signal as extracted data when the post-processed signal satisfies the input criterion, Data extraction system. [2] The input signal is an optical spectrum measured by optical spectroscopy in biological production; The pre-processing unit generates the pre-processed signal by calculating a relative intensity to a signal level in a wavelength range that does not respond to a change in the culture medium and does not have zero transmittance. [1] The data extraction system described in [1]. [3] the optical spectroscopic measurement is near-infrared spectroscopic measurement; the pre-processing unit generates the pre-processed signal by calculating a relative intensity to a signal level in a wavelength range that does not respond to organic molecule fluctuations and in which the light intensity required for the near-infrared spectroscopic measurement is not affected by water absorption. [2] The data extraction system described in [2]. [4] the pre-processing unit generates a first pre-processed signal without calculating the relative intensity, and generates a second pre-processed signal with the relative intensity calculated; the setting unit sets a first input criterion that is an input criterion corresponding to the first preprocessed signal, and sets a second input criterion that is an input criterion corresponding to the second preprocessed signal; the post-processing unit performs the post-processing operation on the first pre-processed signal to generate a first post-processed signal when the first pre-processed signal does not satisfy the first input criterion; the extraction unit extracts the first pre-processed signal as first extracted data when the first pre-processed signal satisfies the first input criterion, and extracts the first post-processed signal as first extracted data when the first post-processed signal satisfies the first input criterion; the post-processing unit performs the post-processing operation on the second pre-processed signal to generate a second post-processed signal when the second pre-processed signal does not satisfy the second input criterion; the extraction unit extracts the second pre-processed signal as second extracted data when the second pre-processed signal satisfies the second input criterion, and extracts the second post-processed signal as second extracted data when the second post-processed signal satisfies the second input criterion; [3] The data extraction system described in. [5] The data extraction system according to any one of [1] to [4], wherein the input criterion is a signal-to-noise ratio. [6] The data extraction system according to any one of [1] to [5], wherein the post-processing unit performs an arithmetic average or a moving average as the post-processing calculation. [7] the setting unit further sets an input criterion different from the input criterion; the post-processing unit performs the post-processing operation on the pre-processed signal to further generate another post-processed signal when the pre-processed signal does not satisfy the another input criterion; the extraction unit further extracts the other pre-processed signal as other extracted data when the pre-processed signal satisfies the other input criterion, and further extracts the other post-processed signal as other extracted data when the other post-processed signal satisfies the other input criterion; The data extraction system according to any one of [1] to [6]. [8] The data extraction system according to any one of [1] to [7], further comprising an abnormality determination unit that determines an abnormality based on the input signal, the pre-processed signal, or the post-processed signal, and invalidates the extracted data when an abnormality is detected. [9] The data extraction system according to [8], wherein the abnormality determination unit detects abnormalities in the culture medium based on a decrease in the amount of dissolved oxygen.

[10] The data extraction system according to [8], wherein the abnormality determination unit detects an abnormality in the electrode potential based on a correlation between the amount of dissolved oxygen or pH and the electrode potential.

[11] 1. A data extraction method performed by a data extraction system comprising at least one processor, comprising: setting input criteria indicative of desired performance for data analysis; A step of acquiring an input signal indicating a measurement result of a measurement object in biological production; performing a pre-processing operation on the input signal for comparison with the input reference to generate a pre-processed signal; If the pre-processed signal does not satisfy the input criteria, performing a post-processing operation on the pre-processed signal using another signal that indicates the measurement result of the measurement object to generate a post-processed signal; extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input criteria. Data extraction methods.

[12] setting input criteria indicative of desired performance for data analysis; A step of acquiring an input signal indicating a measurement result of a measurement object in biological production; performing a pre-processing operation on the input signal for comparison with the input reference to generate a pre-processed signal; If the pre-processed signal does not satisfy the input criteria, performing a post-processing operation on the pre-processed signal using another signal that indicates the measurement result of the measurement object to generate a post-processed signal; extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input criteria; A data extraction program that causes a computer to execute the above.

[13] Acquiring a light spectrum measured by optical spectroscopy in bioproduction, and outputting a signal obtained by calculating the relative intensity of the optical spectrum to the signal level of the optical spectrum in a wavelength range that does not respond to changes in the culture medium and does not have zero transmittance. Data extraction system. [Explanation of symbols]

[0166] 1...data extraction system, 2...input unit, 3...preprocessing unit, 4...postprocessing unit, 5...extraction unit, 6...control unit, 7...setting unit, 8...comparison unit, 9...abnormality determination unit, 10...control block, 20...culture tank, 21...light source, 22...flow cell, 23...spectrophotometer, 24...graph, 30...sensor, 40...database, 50...biological production system, 51...culture control device, 52...sensor control device, 53...control terminal, 54...cloud system, 55...data analysis system, 56...dimensionality reduction unit, 57...explanatory variable extraction unit, 58...machine learning unit, 59...communication network, 100...computer, 101...processor, 102...main memory unit, 103...auxiliary memory unit, 104...communication control unit, 105...input device, 106...output device, 110...program

Claims

1. A setting unit for setting input criteria indicating performance required for data analysis; An input unit for acquiring an input signal indicating a measurement result of a measurement object in biological production; a pre-processing unit that performs a pre-processing operation on the input signal for comparison with the input reference to generate a pre-processed signal; a post-processing unit that performs a post-processing operation on the pre-processed signal using another signal that indicates a measurement result of the measurement object when the pre-processed signal does not satisfy the input criterion, and generates a post-processed signal; an extraction unit that extracts the pre-processed signal as extracted data when the pre-processed signal satisfies the input criterion, and extracts the post-processed signal as extracted data when the post-processed signal satisfies the input criterion, Data extraction system.

2. The input signal is an optical spectrum measured by optical spectrometry in biological production; The pre-processing unit generates the pre-processing signal by calculating a relative intensity to a signal level in a wavelength range that does not respond to a change in the medium and has a non-zero transmittance. The data extraction system according to claim 1 .

3. the optical spectroscopic measurement is near-infrared spectroscopic measurement; The pre-processing unit generates the pre-processed signal by calculating a relative intensity to a signal level in a wavelength range that does not respond to organic molecular fluctuations and in which the light intensity required for the near-infrared spectroscopic measurement is not affected by absorption by water. The data extraction system according to claim 2.

4. The pre-processing unit generates a first pre-processed signal in which the relative intensity is not calculated, and generates a second pre-processed signal in which the relative intensity is calculated, the setting unit sets a first input criterion that is an input criterion corresponding to the first preprocessed signal, and sets a second input criterion that is an input criterion corresponding to the second preprocessed signal; The post-processing unit performs the post-processing operation on the first pre-processed signal to generate a first post-processed signal when the first pre-processed signal does not satisfy the first input criterion; the extraction unit extracts the first pre-processed signal as first extracted data when the first pre-processed signal satisfies the first input criterion, and extracts the first post-processed signal as first extracted data when the first post-processed signal satisfies the first input criterion; the post-processing unit performs the post-processing operation on the second pre-processed signal to generate a second post-processed signal when the second pre-processed signal does not satisfy the second input criterion; the extraction unit extracts the second pre-processed signal as second extracted data when the second pre-processed signal satisfies the second input criterion, and extracts the second post-processed signal as second extracted data when the second post-processed signal satisfies the second input criterion. The data extraction system according to claim 3.

5. The data extraction system of claim 4 , wherein the input criterion is a signal-to-noise ratio.

6. The data extraction system according to claim 5 , wherein the post-processing unit performs an arithmetic operation of an additive average or a moving average.

7. The setting unit further sets an input criterion different from the input criterion, The post-processing unit performs the post-processing operation on the pre-processed signal to further generate another post-processed signal when the pre-processed signal does not satisfy the another input criterion; the extraction unit further extracts the other pre-processed signal as another extracted data when the pre-processed signal satisfies the other input criterion, and further extracts the other post-processed signal as another extracted data when the other post-processed signal satisfies the other input criterion. The data extraction system according to claim 1 .

8. The data extraction system according to claim 1 , further comprising an abnormality determination unit that determines an abnormality based on the input signal, the pre-processed signal, or the post-processed signal, and invalidates the extracted data when an abnormality is detected.

9. The data extraction system according to claim 8 , wherein the abnormality determination unit detects an abnormality in the culture medium based on a decrease in an amount of dissolved oxygen.

10. The data extraction system according to claim 8 , wherein the abnormality determination unit detects an abnormality in the electrode potential based on a correlation between the amount of dissolved oxygen or pH and the electrode potential.

11. 1. A method of data extraction performed by a data extraction system comprising at least one processor, comprising: setting input criteria indicative of performance required for data analysis; Acquiring an input signal indicating a measurement result of a measurement object in biological production; performing a pre-processing operation on the input signal for comparison with the input reference to generate a pre-processed signal; performing a post-processing operation on the pre-processed signal using another signal indicating a measurement result of the measurement object when the pre-processed signal does not satisfy the input criterion, thereby generating a post-processed signal; extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input criteria. Data extraction methods.

12. setting input criteria indicative of performance required for data analysis; Acquiring an input signal indicating a measurement result of a measurement object in biological production; performing a pre-processing operation on the input signal for comparison with the input reference to generate a pre-processed signal; performing a post-processing operation on the pre-processed signal using another signal indicating a measurement result of the measurement object when the pre-processed signal does not satisfy the input criterion, thereby generating a post-processed signal; extracting the pre-processed signal as extracted data if the pre-processed signal satisfies the input criteria, and extracting the post-processed signal as extracted data if the post-processed signal satisfies the input criteria; A data extraction program that causes a computer to execute the above.

13. Acquire a light spectrum measured by optical spectroscopy in biological production, A signal is output that is calculated as a relative intensity to a signal level of the optical spectrum in a wavelength range that does not respond to a change in the medium and does not have a transmittance of zero with respect to the optical spectrum. Data extraction system.

Images