Rayleigh intensity pattern measuring device and Rayleigh intensity pattern measuring method

Abstract

A Rayleigh intensity pattern measurement device (100) comprises: a broadband wavelength variable LD(1); an optical coupler (4) which causes oscillated LD light to be incident on an optical fiber and emits Rayleigh scattering light from the optical fiber to a path that is different from the incident path of the LD light; a reception unit (13) which receives coherent light and the Rayleigh scattering light from the wavelength variable LD(1); and an RIP digital processing unit (14) which receives an output signal from the reception unit (13) through an AD converter (6), calculates a cross-correlation coefficient from two different Rayleigh intensity pattern signals obtained from the Rayleigh scattering light on the basis of data including phase information, and stores the cross-correlation coefficient obtained from the result of comparison with a given threshold value, wherein when the compared cross-correlation coefficient is smaller than the threshold value, a strain-distribution or a temperature distribution of a subject is measured by calculating the cross-correlation coefficient until the cross-correlation coefficient is equal to or greater than the threshold value and determining a Rayleigh frequency shift from the cross-correlation coefficient that is equal to or greater than the threshold value.

Description

[Technical field]

[0001] The present application relates to a Rayleigh intensity pattern measuring apparatus and a Rayleigh intensity pattern measuring method. [Background technology]

[0002] A feature of distributed fiber optic sensing (hereinafter referred to as DFOS) technology is that it is highly reliable and can provide valid measurements no matter what the optical fiber is installed in. The three most representative methods of distributed fiber optic sensing technology are Raman, Brillouin, and Rayleigh.

[0003] In particular, in civil engineering works, this DFOS technology is expected to be consistently applied from monitoring the quality of construction to maintenance management after the construction is completed. However, no DFOS technology has yet been manufactured and verified to have a lifespan of about 100 years.

[0004] For example, DAS used in civil engineering works, or other companies' prior technologies, can measure dynamic changes with high precision, but because they are based on electrical signals rather than optical fiber measurements, they lack reproducibility and are not suitable for long-term management.

[0005] On the other hand, the method using optical fiber is reproducible and is more suitable for long-term management. Of these, the three methods mentioned above are representative of the DFOS technology, and of these, the Brillouin method and the Rayleigh method are particularly effective. However, although the Brillouin method can measure absolute frequency, the accuracy is often insufficient for the required accuracy, and at present, the Rayleigh method is considered to be the most effective.

[0006] This Rayleigh method has recently been attracting attention due to its many features, including high stability, signal strength, usability in both multimode and singlemode, high accuracy, and high resolution of less than 1 mm. In other words, compared to the other two methods mentioned above, the Rayleigh method has three excellent features: (1) excellent measurement stability and signal strength, and usability in both singlemode and multimode, (2) higher measurement accuracy than the other two methods, and (3) high resolution of less than 1 mm.

[0007] A general explanation has been given above. Next, in order to compare the prior art related to the above-mentioned DFOS in detail, a specific example will be described below based on the example shown in FIG. FIG. 1 shows a table in which the representative performance of each of six prior art technologies listed vertically is divided into six indicators listed horizontally (see Patent Document 1 and Non-Patent Documents 1-5).

[0008] In Figure 1, the six indicators listed above are shown horizontally from left to right: absolute frequency (meaning the frequency of an LD where the frequency can be controlled and the frequency error is clear; the same applies below), sufficient measurement range (sufficient measurement range), effectiveness of non-uniform strain distribution (non-uniform strain distribution does not cause problems in measurement; in other words, it is determined that there is no problem of poor correlation in a relatively small strain range), speed (i.e., measurement speed), spatial resolution, and measurement distance (maximum measurable distance). In the above, measurement range refers to the measurement range of strain published in a paper or as a product value.

[0009] As a specific numerical example, in Figure 1, the TW-COTDR (Tunable Wavelength Coherent Optical Time Domain Reflectometry) technology shown as prior art number 1 controls the absolute frequency with a measurement accuracy of 2.3 MHz, but this is a case-by-case method that depends on the part, and there is no standard that continues to the next model. Also, in the OBR / Luna technology shown in the fourth row, the effectiveness of non-uniform strain distribution has become possible with a resolution of up to several tens of μm, so the non-uniformity of strain up to this value is not an issue, and the effectiveness is "yes." Also, in the TGD-DAS technology shown in the sixth row, if the phase change is the target, it can be converted directly into strain, so the effectiveness is "yes."

[0010] Figure 1 shows that none of the six prior art technologies listed in this table has a measurement distance of 10 km or more, has above-standard performance in terms of speed and spatial resolution, and also satisfies all of the requirements of absolute frequency, sufficient measurement range, and effectiveness of non-uniform strain distribution.

[0011] In addition, Figure 2 shows the above six prior art technologies related to DFOS, classifying the characteristic technologies used in these six prior art technologies into five types and comparing them (see Patent Document 1 and Non-Patent Documents 1-5). In this Figure 2, for frequency control, among the above five characteristic technologies, whether or not they have frequency control means (e.g., Etalon, Gas Cell, etc.) is shown.

[0012] By the way, the actual needs of this DFOS technology are that the bridges, oil wells, etc. that are monitored must have a lifespan of 100 years. Also, when measuring earthquakes, etc., it is necessary to reliably capture the amount of change over 100 years or more.

[0013] Therefore, we will consider whether the Rayleigh Backscattering Spectrum (hereinafter referred to as RBS) method, which corresponds to the above-mentioned Rayleigh method, can be applied to these strict requirements. When the Rayleigh Backscattering Spectrum (RBS) method is applied to the measurement of the managed object, it is expected to meet the above-mentioned requirements because this method has the following three features.

[0014] In other words, it has three features: optical fiber can be used as material coordinates (coordinates embedded in a moving object) and can be recognized as indicating position (location); it can be detected without using optical signals; and it can be used without practical limitations on distance resolution and measurement accuracy.

[0015] The Rayleigh backscattered spectrum (RBS) method applied to the measurement of managed objects is essentially the same as the Rayleigh intensity pattern (hereinafter sometimes referred to as RIP), which uses position and frequency as variables and the electric field strength (hereinafter the electric field is also referred to as the electric field), its absolute value, or the square of the absolute value as the functional value. Therefore, in the following description of this application, it will be referred to as a Rayleigh intensity pattern measuring device.

[0016] When applying the above-mentioned RIP as a measurement method for the object to be managed, the spatial positions are aligned, the cross-correlation between the RIP value at time t1 and the RIP value at time t2 is calculated, and the amount of Rayleigh frequency shift is calculated from the correlation peak position. In other words, the Rayleigh frequency shift is measured by the power correlation.

[0017] Here, the RIP measurement technology will be explained in more detail with reference to Fig. 3. Fig. 3 shows an example of the RIP distribution obtained by the Rayleigh backscattering light spectrum method using a full-band tunable wavelength distributed feedback LD (LD: Laser Diode). Here, the RIP measurement quantity is calculated by the formula (F S ) Rayleigh frequency shift ΔνR (See, for example, Patent Document 2). Δν R =C 21 Δε+C 22 ΔT+C 23 ΔP (F S ) In addition, the formula (F S ) in C 21 , C 22 , C 23 is the sensitivity coefficient given to the Rayleigh frequency shift, and Δε, ΔT, and ΔP indicate the strain, temperature change, and pressure change of the optical fiber at the measurement position, respectively.

[0018] The frequency shifts derived from the above-mentioned strain, temperature, and pressure changes are obtained by advanced correlation analysis (so the frequency shifts are relative values). The signal used for correlation analysis is the power spectrum of a specific distance range, and is not affected by phase. Therefore, the above frequency shift can be considered as a comparison of the distinct features according to frequency of the TW-COTDR method in two states (reference and observation) at the same location. Here, the TW-COTDR method and the PPP-BOTDA (Pulse Pre-Pump Brillouin Optical Time Domain Analysis) method, which is a high-precision Brillouin technique, can be decoded to harmonize the requirements of the hybrid measurement method using the two methods of Rayleigh backscattering light and Brillouin backscattering light. On the other hand, the Rayleigh backscattering method in the time domain has the following advantages over the Brillouin backscattering method: it only requires a single-ended system, it can accurately and easily detect small changes such as creep monitoring over long distances, and it has stability and high reproducibility because the signal from the optical fiber is independent of the phase. [Prior art documents] [Patent documents]

[0019] [Patent Document 1] Specifications of U.S. Patent No. 7440087

Patent document 2

Patent document 3

Non-licensed literature

[0020]

Non-licensed literature 1

Non-licensed Document 2

Non-licensed Document 4

[0021] However, the prior art including the TW-COTDR method has the following problems. First, the first index shown in Figure 1, "absolute frequency (absolute reference frequency)," has not been established. None of the prior arts shown in Prior Art No. 2 to Prior Art No. 6 can control the LD frequency or wavelength value to clarify the error, as shown in Figure 1. Also, the TW-COTDR method shown in Prior Art No. 1 controls the LD wavelength (optical frequency), but does not manage the absolute value. Therefore, when measuring between different devices (from the same manufacturer) or between devices from different manufacturers, only relative frequency shift values ​​can be obtained.

[0022] Next, the second indicator, the "measurement range," is insufficient (a measurement range sufficient for practical use is not ensured). In each technology other than Prior Art No. 1, the measurement range is too small. Specifically, for example, a temperature change of 100°C is a common case, but in each method using chirped pulse technology (each prior technology other than the TW-COTDR method in Prior Art No. 1), it is only possible to measure within a range of four digits or less of the required value.

[0023] Next, the third indicator, "effectiveness against non-uniform strain distribution," is lacking. In the TW-COTDR method of Prior Art No. 1, if non-uniform strain distribution occurs within the range of the spatial resolution, in other words, within half the pulse width, proper strain detection is not possible. This is because correlation with the reference data is lost due to distortion of the Rayleigh Intensity Pattern (RIP) caused by non-uniform strain distribution within the length range of the spatial resolution. In addition, once the correlation is lost during long-term monitoring, the relationship with the initial value is lost, making long-term monitoring impossible. This is because, in principle, the Rayleigh backscattering method accumulates data in the time axis direction to ultimately measure deformation.

[0024] An example of actual measurements in which no correlation was obtained will be explained with reference to Figure 4. Figure 4A shows the measurement results when a rod is used to measure the frequency shift of Rayleigh backscattered light (hereinafter also referred to as RFS) with the following measurement parameters: Measurement parameters: sweep start frequency: 192010GHz, sweep period: 400GHz, sweep step: 0.2GHz, spatial resolution: 2cm, sampling interval: 1cm.

[0025] In Fig. 4A, strain is not detected correctly in the area enclosed by the dotted circle (within the range of distance positions shown on the horizontal axis from 512.3 m to 513.1 m). Specifically, for example, at a distance position of 512.5 m, the correlation coefficient between the two data, i.e., the reference data and the measured data, is 0.2 or less as shown in Fig. 4B, indicating that there is no correlation. Conventionally, with the Rayleigh backscattering method, data is accumulated in the time direction to ultimately determine the deformation. Therefore, if the accumulation of one piece of data during data accumulation is unsuccessful, the subsequent measurements will lose correlation with the initial value (initial amount of strain change).

[0026] To examine the RFS data above, we will compare it with the data measured on a bar based on Brillouin scattered light, a measurement method that does not utilize the correlation between the two data. Figure 5 shows the data obtained by measuring the strain generated in the bar based on Brillouin scattered light using the same bar as above with the following measurement parameters: Measurement parameters: sweep start frequency: 10 GHz, sweep period: 1.45 GHz, sweep step: 5 MHz, spatial resolution: 5 cm, sampling interval: 2.5 cm.

[0027] Figure 5 shows the distribution of strain when distance (unit: m) is taken on the horizontal axis for the strain data of the bar measured using the Brillouin backscattering method. Figure 5 shows that the strain value, which was over 100με at a distance of about 512.2m, changed to -400με at a distance of about 512.5m, resulting in a net change in strain of over 500με.

[0028] Using this figure 5 to examine the strain change near a distance of 512.2 m, it is found that a strain change of 159.84 με occurs for a distance change of 4.1 cm, as shown in the figure. In other words, in an area within 4.1 cm (4 cm corresponds to the pulse length used in the Rayleigh backscattering light method) near a distance of 512.5 m, the strain change per meter was 3898.54 με.

[0029] As explained above, when there is a sudden strain change within the range of spatial resolution (when there is non-uniform strain), it has been found that when the strain change of the same bar is measured using the Rayleigh backscattering method, it is not possible to measure the accurate strain change or cross-correlation. Normally, the correlation is improved by interpolating the data, but in this case, the correlation was not improved. The unit step of the frequency (change) during data processing was 0.04 GHz.

[0030] Next, the fourth challenge is that in conventional technologies that use homodyne reception as the receiving method (the first and fourth technologies in the table; see Figure 2), there is a problem in that the phase information necessary for advanced signal analysis is lost from the start with the current homodyne reception.

[0031] Furthermore, as a fifth issue, if the speed indexed in the table is slow, not only will it be impossible to follow high-speed phenomena, but if there is a change in strain during measurement, a different frequency shift (Δν R ), which, as explained above, cannot be correlated. This will eventually lead to a loss of effectiveness of surveillance overall.

[0032] Next, we will explain the technical issues involved in realizing each of the prior arts shown in Figure 1. As for TW-COTDR (prior art number 1), since it uses a full-band LD, the measurement range is sufficiently wide, but there is a problem that the linewidth of the LD is too wide. For example, in the case of 4 MHz, the effective coherent length is only several tens of meters, and since the phase signal contains large phase noise in the TGD (Time Gated Digital) phase analysis, it cannot be used as is. In other words, since the TW-COTDR method uses homodyne reception for the reception method as shown in Figure 2, phase information may be lost from the beginning, and the phase information required for advanced signal analysis cannot be used.

[0033] In addition, in measurements using the TGD-DAS method (method shown in prior art number 6 in Figure 1), measurements were successfully performed using a light source with a line width of 1 KHz or less. However, in this case, there is a problem that the frequency cannot be changed (because there are almost no commercially available products).

[0034] Furthermore, to ensure a wide measurement range, the RIP's receiving frequency range (receiving bandwidth) needs to be 100 GHz or more, but the current situation is that it is nearly impossible to cover this as a receiving band using electrical signals.

[0035] The above-mentioned problems can be summarized as follows, with specific numerical examples for each of the representative prior arts. The TW-COTDR method of prior art No. 1 excels in static measurements such as strain and temperature, and is characterized by a high spatial resolution (2 cm) and a high accuracy measurement method, but has a drawback in that the measurement speed is slow, taking several tens of seconds. On the other hand, the TGD-DAS method of prior art No. 6 excels in dynamic measurements such as acoustic measurement, and is characterized by a high measurement speed of over 2000 times per second, but its spatial resolution is inferior, at around 1 m. In addition, although not shown in Figure 1, it is limited to measuring minute changes, for example, temperature changes are limited to 0.2°C, and it is not suitable for measuring large changes.

[0036] The present application has been made to solve the above-mentioned problems, and aims to provide a Rayleigh intensity pattern measurement device that can simultaneously achieve three parameters, spatial resolution, measurement speed, and sufficient measurement range, which are in a trade-off relationship when achieving high-speed, high-precision measurements, and that enables management of data acquired over a long period of time. [Means for solving the problem]

[0037] The Rayleigh intensity pattern measurement apparatus disclosed in the present application comprises: A wavelength-tunable LD, and a controller capable of changing the frequency of laser light oscillated from the wavelength-tunable LD; A local oscillator; A light source unit having an optical coupler that inputs the laser light oscillated from the light source unit into an optical fiber and outputs Rayleigh scattered light from the optical fiber through a path different from the input path of the laser light; a receiving unit that coherently receives the local light from the light source unit and the Rayleigh scattered light from the optical coupler; a Rayleigh intensity pattern digital processing unit comprising: an AD converter that performs AD conversion of a signal output from said receiving unit; a first calculation unit that performs a predetermined calculation on the signal converted by said AD converter; a memory database to which the signal calculated by said first calculation unit is input and stored; and a second calculation unit that performs a predetermined calculation on the basis of the data stored in said memory database, said Rayleigh intensity pattern digital processing unit determining a cross-correlation coefficient from two different Rayleigh intensity patterns at a predetermined measurement position obtained from the electric field signal of the Rayleigh scattered light after performing a predetermined correction for said measurement position in said second calculation unit, and storing in said memory database the cross-correlation coefficient when the determined cross-correlation coefficient is equal to or greater than a predetermined threshold value; and Matched Filter , Equipped with The absolute frequency of the tunable LD is controlled, and the frequency of the laser oscillated from the tunable LD is scanned in a stepwise manner, and the chirp signal pulsed at each step is received by the receiving unit and then synthesized by the first calculation unit; and The chirp signal is set by using a plurality of window functions according to the subbands extracted by the matched filter, and the Rayleigh intensity pattern including the subbands recombined by using the center frequency of the extracted subbands is preserved at the absolute frequency; The method is characterized in that the Rayleigh frequency shift is calculated based on the cross-correlation coefficient that is equal to or greater than a predetermined threshold value, thereby determining the strain distribution or temperature distribution of the object to be measured. Effect of the Invention

[0038] The Rayleigh intensity pattern measurement device disclosed in the present application aims to provide a Rayleigh intensity pattern measurement device that can simultaneously achieve three parameters, spatial resolution, measurement speed, and sufficient measurement range, which are in a trade-off relationship when achieving high-speed, high-precision measurements, and that is capable of managing data acquired over a long period of time. [Brief description of the drawings]

[0039] [Figure 1] FIG. 2 is a diagram for explaining technical indicators that are problems to be solved by the Rayleigh intensity pattern measurement device according to the first embodiment. [Diagram 2] This is a diagram to explain the characteristic technologies of DFOS. [Diagram 3] FIG. 13 is a diagram showing an example of an RIP distribution obtained by a Rayleigh backscattering light method. [Figure 4] FIG. 13 is a diagram showing an example of the strain and correlation function of a bar measured by the Rayleigh backscattering light method. [Diagram 5] FIG. 1 shows strain data of a bar measured by Brillouin backscattering. [Figure 6] FIG. 1 is a diagram for explaining the overall configuration of a RIP measurement device according to a first embodiment. [Figure 7] FIG. 2 is a diagram for explaining the operation of the RIP measurement device according to the first embodiment. [Figure 8] 10 is a diagram for explaining the peak value of the cross-correlation coefficient of the Rayleigh frequency shift in the detection optimization analysis method for non-uniform distortion of the RIP measurement device according to the first embodiment. FIG. [Figure 9] FIG. 13 is a diagram for explaining the Fresnel integral used in the non-uniform distortion detection optimization analysis method. [Figure 10] 1A and 1B are diagrams for explaining a finite chirp waveform and its Fourier transform when the strain change of non-uniform strain is 0.1 GHz / m. [Figure 11] 1A and 1B are diagrams for explaining a finite chirp waveform and its Fourier transform when the distortion change of non-uniform distortion is 5 GHz / m. [Figure 12] 1A and 1B are diagrams for explaining a finite chirp waveform and its Fourier transform when the strain change of non-uniform strain is 20 GHz / m. [Figure 13] 1A and 1B are diagrams showing a distortion model and a distortion gradient model in a non-uniform distortion detection optimization analysis method. [Figure 14] FIG. 13 is a diagram showing a Rayleigh frequency shift model and its gradient model in an analysis method for optimizing detection of non-uniform distortion. [Figure 15] 11A and 11B are diagrams for explaining simulation results in a model that does not take into account the slope of the Rayleigh frequency shift in detecting non-uniform distortion. [Figure 16] 11 is a diagram for explaining a simulation result when a distribution of Rayleigh frequency shifts is obtained by applying an analysis method for optimizing detection of non-uniform distortion. FIG. [Figure 17]11 is a diagram for explaining a simulation result when a distribution of the slope of the Rayleigh frequency shift is obtained by applying an analysis method for optimizing the detection of non-uniform distortion. FIG. [Figure 18] 11A and 11B are diagrams for explaining the results of a simulation in which the distribution of strain is obtained by applying an analysis method for optimizing the detection of non-uniform strain. [Figure 19] 4 is a diagram for explaining strain distribution data for forming a Rayleigh intensity pattern used to explain examples in the RIP measurement device according to the first embodiment. FIG. [Figure 20] 1 is a diagram for explaining parameters used in explaining a scanning method of multiple chirp pulses simulating laser light, which is used to drive a receiving unit in a Rayleigh intensity pattern measurement device. FIG. [Figure 21] FIG. 11 is a model diagram for explaining a method of scanning a plurality of chirp pulses in a stepwise manner. [Figure 22] FIG. 1 illustrates a matched filter configuration for extracting subbands from each chirp pulse. [Figure 23] FIG. 13 is a diagram illustrating a series of methods for extracting a trace showing a specific frequency along the distance of the object from each chirp pulse. [Figure 24] FIG. 1 shows an example of RIP, the intensity of the Rayleigh backscattered signal at the corresponding laser frequency, divided into the laser linewidth. [Diagram 25] FIG. 13 is a diagram for explaining the data preparation process in another analysis method for optimizing the detection of non-uniform distortion. [Figure 26] FIG. 13 is a diagram for explaining a data decomposition process in another analysis method for optimizing detection of non-uniform distortion. [Figure 27] FIG. 13 is a diagram for explaining a correlation coefficient calculation and determination step in another analysis method for optimizing detection of non-uniform distortion. [Figure 28] FIG. 13 is a data processing flowchart for another analysis method for optimizing detection of non-uniform distortion. [Figure 29]FIG. 13 shows the results of processing data in which non-uniform distortion is detected using another analysis method for optimizing the detection of non-uniform distortion. [Diagram 30] FIG. 30 is a supplementary explanatory diagram for FIG. 29. [Diagram 31] FIG. 13 is a diagram for explaining fundamental frequency components used in a method similar to another analysis method for optimizing detection of non-uniform distortion. [Diagram 32] FIG. 13 is a diagram showing a data processing flow used in a method similar to another analytical method for optimizing the detection of non-uniform distortion. [Diagram 33] FIG. 13 is a diagram showing a correlation coefficient, which is an example of a result of applying a method similar to another analytical method for optimizing detection of non-uniform distortion. [Diagram 34] FIG. 13 is a diagram showing a strain distribution that is an example of the result of applying a method similar to another analytical method for optimizing the detection of non-uniform strain. [Diagram 35] 11A and 11B are diagrams for explaining the effect of a method similar to another analysis method for optimizing detection of non-uniform distortion. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0040] Embodiment 1 The overall configuration of a Rayleigh intensity pattern measurement apparatus (hereinafter, also simply referred to as a RIP measurement apparatus) according to the first embodiment will be described below with reference to FIGS. 6 and 7. FIG.

[0041] 6 is a diagram for explaining the overall configuration of the RIP measurement device 100 according to the embodiment 1. The RIP measurement device 100 according to the embodiment 1 is a measurement device of the TW (Tunable Wavelength)-TGD-RFS (Rayleigh Frequency Shift) type that uses a divided chirp pulse. The RIP measurement device 100 according to the first embodiment is roughly divided into four main components: a light source unit 11, a transmitting / receiving optical unit 12, a receiving unit 13, and a RIP determination digital processing unit 14 (hereinafter also referred to as a Rayleigh intensity pattern digital processing unit 14 or a RIP digital processing unit 14). The details of these components will be described below in order.

[0042] The light source unit 11 has a controller that controls the LD1, which is a wideband type wavelength-tunable laser (TW-LD), and the laser light emitted from the LD1, and performs automatic temperature control and automatic frequency control while performing absolute frequency control of the LD1 (here, absolute frequency control is based on the molecular absorption wavelength, for example, as in a gas cell. This information is publicly known, has low temperature dependency, and can be easily achieved within a range that does not affect measurement accuracy) and frequency step control using a built-in optical local oscillator (OLO) to oscillate divided chirp pulses as output. Note that the wavelength-tunable laser also includes external modulation types. That is, the light source unit 11 uses a chirped optical pulse to control the frequency of the tunable LD 1, and sweeps the frequency while changing it in multiple steps (for example, changing it in 12 steps at 200 GHz). In this case, the TW-LD is utilized to generate a high-precision optical pulse using a TGD (Time Gated Digital) method or the like. The sweep width in the TGD method is, for example, 4 GHz (sample rate is 8 GS / s).

[0043] The transmitting / receiving optical unit 12 has an optical switch 2 that pulses the laser signal, an erbium-doped fiber amplifier 3 (hereinafter abbreviated as EDFA3) that amplifies the output signal of this optical switch 2, and an optical coupler 4 that transmits the signal from the EDFA3 to a measurement object (hereinafter also referred to as the object to be measured (specifically, the optical fiber)) via a reference optical fiber 5, and receives backscattered light, which is scattered light from the measurement object of the transmitted signal, via the reference optical fiber 5 and then sends it to an external receiver described below.

[0044] The receiving unit 13 has a receiver that receives the backscattered light sent from the optical coupler 4 of the transmitting / receiving optical unit 12, and also receives the light source light from the light source unit 11. This receiver is configured as a heterodyne receiver that receives signals using a polarization diversity method, or a homodyne receiver that receives signals using a phase diversity method and a polarization diversity method, in order to improve reception sensitivity and obtain phase information.

[0045] The RIP digital processing unit 14 is a unit that performs signal strength or phase analysis by digital methods. Here, the output signal from the receiving unit 13 is converted to a digital signal at high speed by the 2ch AD converter 6 and received, and the signal is processed at high speed by four types of methods described later using an FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), etc. The RIP digital processing unit 14 also has a processor that performs arithmetic processing of digital signals and a memory database 8 (hereinafter also referred to as memory DB8) that stores and saves the signals processed by this processor, a first arithmetic unit 7 that performs arithmetic processing of the signal received from the 2ch AD converter 6 (for example, performs an arithmetic operation to synthesize multiple chirp signals) and outputs the signal to the memory database 8 as a signal different from the received signal, and a second arithmetic unit 8 that performs arithmetic processing of two times (for example, time t i and time t j ) data, and based on these two different data, the initial Rayleigh frequency shift Δν R‐I Then, the Rayleigh frequency shift Δν is determined by performing calculations such as resampling, distance correction, and correlation (correlation judgment), such as frequency-axis interpolation. R‐D (This second calculation unit 9 will be described in detail below with reference to FIG. 7 and the like. Hereinafter, this second calculation unit 9 may also be simply referred to as calculation unit 9.) Here, the 2ch AD converter 6 can convert the amplitude and phase components of the signal for each channel using a processor such as an FPGA chip. Also, the RIP digital processing unit 14 can perform signal strength and phase analysis using a digital method. In the figure, the determined Rayleigh frequency shift Δν R‐D The output of the above is included within the RIP digital processing unit 14, but may be external to the RIP digital processing unit 14.

[0046] Of the above-mentioned four main components, the three enclosed in solid line frames, namely the light source unit 11, the receiving unit 13, and the RIP digital processing unit 14, are particularly characteristic components of the RIP measurement device according to the present embodiment 1. On the other hand, for the transmitting / receiving optical unit 12 enclosed in a dotted line frame, existing OTDR technology can be applied, and it is not particularly new (for example, see Patent Document 3).

[0047] In addition, in the figure, Method 1, Method 2, Method 3, and Method 4 are each given different names in order to clearly explain the differences in operation in the RIP digital processing unit 14 depending on the receiving method in the receiving unit 13, and details of these will be explained in detail below using Figure 7.

[0048] Therefore, hereinafter, the operation of the entire RIP measurement device according to the first embodiment will be described in detail for each of the above methods 1, 2, 3, and 4, with particular focus on the above second calculation unit 9, with reference to Fig. 7. Whichever method is used, it is possible to improve the signal receiving sensitivity and achieve high-density wavelength multiplexing.

[0049] First, in method 1, the chirp signal, which is a signal received by the receiver 13 through heterodyne, is not subjected to any special processing in the 2ch AD converter 6 (the signal is passed through), and is stored as is in the memory DB8 of the RIP digital processor 14 (step S1). k Since the data is stored at time t i , and time t j The data at these two times are selected (step S4). Based on the data at these selected two times, the second calculation unit 9 executes the following series of operations.

[0050] First, the initial Rayleigh frequency shift Δν R-I Next, the initial Rayleigh frequency shift Δν R-IFor the distance correction, a frequency resampling process including an interpolation process is performed (step S6). Next, a cross-correlation coefficient after distance correction is calculated based on the frequency resampled data (step S7). Next, it is determined whether the calculated cross-correlation coefficient is larger than a predetermined threshold value (step S8). This threshold value can be preset based on experience, for example, as 0.3. If the result of the above judgment is "Yes", the Rayleigh frequency shift Δν R Calculate the Rayleigh frequency shift Δν R The Rayleigh frequency shift Δν R‐D (Step S10).

[0051] On the other hand, if the result of the above judgment is "No", an analysis is performed to optimize the detection of non-uniform distortion. That is, data that is the basis for recalculating the cross-correlation coefficient after distance correction is obtained by analysis (step S9), and the cross-correlation coefficient after distance correction is calculated again (step S7). Next, it is determined whether the calculated cross-correlation coefficient value is greater than a predetermined threshold value (step S8). Then, the operations of step S9, step S7, and step S8 are repeated until the value of the cross-correlation coefficient becomes larger than a predetermined threshold value, and the Rayleigh frequency shift Δν is calculated based on the value of the cross-correlation coefficient when the value of the cross-correlation coefficient becomes larger than the predetermined threshold value. R Calculate the Rayleigh frequency shift Δν R The value of is calculated by the Rayleigh frequency shift Δν R-D (Step S10). In method 1, since there is no processing by the 2ch AD converter 6, high speed processing of the processing data is possible.

[0052] Next, method 2 will be described, focusing on the differences from method 1. In method 2, a chirp signal, which is a signal received by heterodyne reception at the receiving unit 13, is AD converted by the 2ch AD converter 6, and the signal after comb filtering processing by an FPGA or the like is stored in the memory DB8 of the RIP digital processing unit 14 (step S2). The subsequent processing is the same as in method 1, so a detailed description will be omitted here. In method 2, comb filtering can be used to remove periodic noise contained in the signal, improving the accuracy of the measurement data.

[0053] Next, method 3 will be described, focusing on the differences from methods 1 and 2. In method 3, after the processing of method 2, the processed signal is further squared to produce a digitally processed signal, which is then stored in the memory DB8 of the RIP digital processing unit 14 (step S3). The subsequent processing is the same as method 1, so a detailed description will be omitted here. Method 3 uses a squared signal, so the signal components are less affected by the phase, which leads to higher accuracy of the measurement data.

[0054] Finally, method 4 will be described, focusing on the differences from methods 1, 2, and 3. Methods 1 to 3 use a chirp signal, which is a signal received by the receiver 13 through heterodyne reception, but method 4 differs from these three methods in that it uses a signal received by the receiver 13 through homodyne reception, which is a signal containing phase information. In addition, a phase diversity method is adopted to prevent the loss of phase information. The signal after squaring this homodyne-received signal is stored in the memory DB8 of the RIP digital processor 14 (step S3). The operations from this point onwards are the same as those of method 1, so a detailed description will be omitted here. Compared with the methods 1 to 3 that use heterodyne reception signals, method 4 requires a narrow-linewidth LD and optical PLL with low phase noise, but it has the advantage of enabling the most sensitive reception and easier high-speed data processing.

[0055] Since the Rayleigh intensity pattern measurement device 100 according to the first embodiment is configured as described above, it is possible to effectively improve the pulse power, and since it is possible to simultaneously obtain scattering waveforms at a plurality of frequencies, it is possible to significantly reduce the number of optical pulses to be transmitted, and it is possible to shorten the measurement time. Note that in order to analyze the correlation of RIP, a large amount of analysis is performed in the RIP digital processing unit 14.

[0056] The above methods 1 to 4 are devised to enable proper detection of non-uniform distortion by using an analysis method for detection optimization as shown in Fig. 7. Next, the analysis method that can properly detect the non-uniform distortion and is commonly used in the above methods 1 to 4 will be described in detail below. These analysis methods are roughly divided into two types, analysis method N1 and analysis method N2. The following will explain them in order, starting with analysis method N1.

[0057] [Analysis method N1] Analysis method N1 is an analysis method that can properly detect the strain distribution of a measured object even when there is non-uniform strain, from the cross-correlation between a reference spectrum obtained from the electric field of the reference data of Rayleigh scattered light and various reference spectra obtained by modifying it, and an observed spectrum obtained from the electric field of the observed data. The reference spectrum is modified to deal with the strain gradient (explained in detail later), and the deformation differs depending on the magnitude of the gradient. The processing method (algorithm) used in analysis method N1 is explained in detail below.

[0058] In creating the above algorithm, first, parameters are determined based on the basic model. Here, we consider the case where a laser beam with a frequency ν and a pulse width D is input from the input end of an optical fiber, and the scattered light returning is observed, in particular, the Rayleigh scattered light P is observed. In such cases, the observed values ​​are the light intensity obtained by direct detection or the electric field intensity obtained by heterodyne reception, both of which are expressed by P = P(t, ν) and two parameters: time t and frequency ν. For example, spectral data of Rayleigh scattered light can be obtained by acquiring observed values ​​while scanning the frequency ν of the laser light source.

[0059] Next, consider the case where the strain or temperature in the optical fiber changes. In this case, the optical path length changes. In other words, the round-trip time τ(z) of light from the input end of the optical fiber to the distance z changes. The observed value in this case can be expressed as P = P(τ(z),ν) by replacing the above time t with the round-trip time τ(z) of light. To simplify the notation, hereafter, P(τ(z),ν) will be written as P(z,ν). Therefore, in the following, in order to first consider the spectrum before and after the change in strain or temperature, the spectrum before the change in strain or temperature is P 0 (z, ν), and the spectrum after the change is P 1 This is written as (z, ν).

[0060] Now, the distance interval [zl p / 2, z+l p / 2] (hereinafter, the distance section is simply called the section), the change in strain or temperature is assumed to be constant, and the Rayleigh frequency shift ν R (z), the Rayleigh frequency shift ν R (z) is a constant value in this interval. Then, the spectrum P after the change 1 (z, ν) and the spectrum before the change P 0 It is known that the following relationship holds between (z, ν):

number

[0061] [If the strain or temperature change is not constant within the pulse] Next, we consider the problem of the strain or temperature change being non-constant within the pulse. In this case, the relationship between the spectra is no longer expressed by the frequency shift in the above formula (1), which leads to a deterioration in detection performance, and will be examined in detail below.

[0062] [If the strain or temperature change is not constant within the pulse] For convenience of calculation, the frequency variable is changed from the frequency ν on the time axis described above to the corresponding spatial frequency (see equation (2)). Also, the Rayleigh frequency shift is expressed as equation (3) below.

number

number

[0063] Here, the electric fields of the scattered light observed before and after the change are Y 0 (z, u), Y 1 If we put (z, u), the square of their absolute values ​​gives the spectrum (see equations (4) and (5)).

number

number

[0064] Here, Y 0 (z, u), Y 1 (z, u) are given by equations (6) and (7), respectively.

number

number

[0065] It is assumed that the change in strain or temperature changes linearly within the interval corresponding to the pulse width, and the Rayleigh frequency shift is within the interval [zl p / 2, z+l p / 2], and assume that the following equation (8) is obtained.

number

number

[0066] Here, by utilizing equation (9), the following equation (10) holds from equation (7).

number

[0067] The right-hand side of equation (10) is composed of two functions ρ(z+ζ), -l p / 2≦ζ≦l p / 2 and exp(πiβζ 2 ), -∞<ζ<∞, is the Fourier transform with respect to ζ of the product. Here, ρ(z) is the Rayleigh backscattering coefficient at distance z, and the Fourier transform of the product of two functions is equal to the convolution of the Fourier transforms of each function. Also, ρ(z+ζ), -l p / 2≦ζ≦l p The Fourier transform of / 2 is given by Y 0 (z, u). Therefore, excluding the constant coefficient, the following equation (11) holds.

number

[0068] q(u;β) is the chirp exp(πiβζ 2 ) and is given by equation (12).

number

[0069] Therefore, the spectrum after the change can be expressed by equation (13), but within the interval u R Unlike the case where becomes constant, it cannot be expressed using the original spectrum.

number

[0070] In equations (4), (5), (6), and (10), for simplicity, α is set to 0, z is arbitrarily fixed to one, and the argument z is omitted, and the spectrum of scattered light can be expressed in simple terms as equations (14) and (15).

number

number

[0071] Here, the Rayleigh backscattering coefficient ρ(z) can be regarded as a spatially white Gaussian process, so by using the Dirac delta function, the following equation (16) holds: Here, E[·] represents the expectation value, and Q is a constant.

number

[0072] Using equation (16), P 0 and P 1 If we calculate the expectation value up to the second order for 0 and P 1 The peak value of the cross-correlation coefficient is expressed by equation (17).

number

number

[0073] Furthermore, the pulse width D is D=2l p / v g , the variation of the Rayleigh frequency shift within the pulse (the difference between the maximum and minimum) is Δν R =bl p =v g βl p / 2, so DΔν R =β(l p ) 2 Therefore, the peak value of the cross-correlation coefficient is finally given by equation (19).

number

[0074] 8A and 8B show the r calculated using equation (19). peak As can be seen from Figure 8A, the value of r peak is DΔν R = 0, DΔν R As increases, it decreases rapidly and approaches 0 (zero). In the method of detecting the magnitude of the Rayleigh frequency shift by cross-correlation between spectra, the peak value r peak Therefore, it is necessary to suppress the decrease in the peak value, and it is found that it is effective to reduce the pulse width as shown in FIG. 8B (in fact, when the pulse width is reduced, Δν R (Since the peak value decrease is also reduced, both effects suppress the decrease in the peak value.) So, for example, r peak DΔν when =0.3 R The value of is 3.11 (see Figure 8A). Applying this value, DΔνR The limit of applicability is ≦3.11. When converted to the slope of the Rayleigh frequency shift, it is expressed by the following equation (20).

number

[0075] [Method of detecting Rayleigh frequency shift when strain or temperature change has a slope within a section] If the strain or temperature change is constant within a section, the change can be detected as a shift in the spectrum, but if there is a slope, there are detection limitations as mentioned above. However, if the electric field data before calculating the optical intensity of the spectrum can be used, it becomes possible to detect the slope as well. This method is explained below.

[0076] In other words, simply put, it is a method that utilizes spectral deformation by chirp. First, when reference data is obtained, spectra in cases where the Rayleigh frequency shift has various slopes due to spectral deformation by chirp are saved as reference spectra. Next, when observed data is obtained, the spectrum obtained from the observed data (hereinafter also referred to as the observed spectrum) is compared with the previous reference spectrum, and correlation is taken to obtain estimates of the Rayleigh frequency shift amount a and slope b.

[0077] [Spectral transformation by chirp] Each distance z(0≦z≦L f , L f The electric field data Y(z,ν k ), k = 0, 1, . . . , K-1 are given. k =kΔν, where Δν is the frequency interval. K is the number of data in the frequency direction, and KΔν is the frequency bandwidth of the data. The slope b of the Rayleigh frequency shift is also discretized, and b j , let j = 0, 1, ..., J-1.

[0078] The reference spectrum is the electric field data before the change Y 0 (z, ν) is processed to create the image. For the sake of convenience, the frequency ν is set to the spatial frequency u=2ν / v g (unit: cycle / m), and the slope b of the Rayleigh frequency shift is β=2b / v g (Unit: cycle / m 2 Then, the electric field data before the change is expressed by the following equation (21).

number

number

[0079] Next, taking note of the fact that the right-hand side of the above equation (22) is the Fourier transform of the product of functions, in the above-mentioned case [where the strain or temperature change is not constant within the pulse], it is expressed as the convolution of the Fourier transforms, as shown in the following equations (23) to (25).

number

number

number

[0080] q(u k , β j ) is the Fourier transform of an infinitely long chirp, and is expressed as an infinitely long chirp on the frequency axis, as shown in equation (12). However, this infinitely long chirp has an infinitely large vibration component, so it is difficult to discretize and handle it in signal processing. In other words, aliasing will occur no matter how high the sampling rate is.

[0081] So, the spatial resolution is lp Taking advantage of this, the following equivalent equations (26) and (27) are used instead of equations (23) and (24), respectively.

number

number

[0082] Here, q lp (u;β) is the Fourier transform of the infinite length chirp expressed by equation (28), and is expressed by equation (29) using the Fresnel integral function.

number

number

number

number

[0083] The Fresnel integral function is defined by the following equations (32) to (34), and changes as shown in FIG.

number

number

number

[0084] q required for calculation lp The length of (u;β) in the u direction varies depending on the bandwidth of s(z;β). Figures 10A, 10B, 11A, 11B, 12A, and 12B show the results for β=1, 50, and 200 cycles / m, respectively. 2 (b=0.1, 5, 20GHz / m) for the finite chirp waveform s(z;β) and its Fourier transform q lp (u;β). In addition, p = 0.5m. The spatial frequency bandwidth of s(z;β) is l p β. The maximum absolute value of β is max Then, the maximum bandwidth is l p β max It becomes.

[0085] [Detection method of Rayleigh frequency shift including slope] Next, we will show a method for detecting the Rayleigh frequency at each distance z, including the gradient in the distance direction. In this detection method, if sufficient correlation can be obtained without considering the gradient of the Rayleigh frequency shift, the gradient is not detected. Again, for convenience of calculation, the spatial frequency u=2ν / v is used instead of the frequency ν. g , instead of the Rayleigh frequency shift parameters a and b, α=2a / v g , β=2b / v g These relationships are expressed as follows: t=2z / v where z is the distance of the scattered light and t is the round-trip time. g The relationship is the same as above, and a similar conversion can be made.

[0086] [Detection method 1] In detection method 1, the reference spectrum is obtained in the frequency domain. (1) Electric field Y of the reference data 0 (z, u k ), k = 0, 1, . . . , K-1 are given. From this, for each β = β j , j=0, 1, . . . , J-1, the reference spectrum P 0 (z, u k ;βj ) is calculated in advance by the method described below. a) Fourier transform of chirp q lp (u;β) is found from the above equations (29) to (31), where Z(·) is the Fresnel integral function defined by equation (32). b) Convolve the Fourier transform of the chirp with the reference data in the frequency domain (see equation (35)).

number

number

number

number

number

number

number

number

number

number

number

number

number

[0087] [Detection method 2] In the above-mentioned detection method 1, the reference spectrum is obtained by convolution in the frequency domain, but in detection method 2, a method is used in which a product is calculated in the spatial domain using Fourier transform, and then the frequency is restored by inverse Fourier transform. This detection method 2 can reduce the calculation time. In detection method 2, the following (1a) is used instead of (1) in detection method 1. Note that (2) to (6) in detection method 2 are the same as (2) to (6) in the above-mentioned detection method 1, so a detailed explanation is omitted here. (1a) Electric field Y of the reference data 0 (z, u k ), k = 0, 1, . . . , K-1. k =kΔu is the spatial frequency and Δu is its step size. The step size of the spatial variable is Δζ=1 / (KΔu), and the spatial variable is ζ n =nΔζ, n=0, 1, . . . , K-1. Now, for each β = β j , for j=0, 1, . . . , J-1, the reference spectrum P 0 (z, u k ;β j ) is calculated in advance by the method described below. a) Each β = β j , j=0, 1, . . . , J-1, the chirp s(ζ n ;β j ) is calculated. First, we calculate s(ζ n ;β j ) is calculated according to the following equation (48).

number

number

number

number

number

number

[0088] [Verification by simulation] Here, we conduct simulations to verify the effect of dividing the signal into multiple frequencies on the receiving side using chirped optical pulses. The parameters used in the simulations are as follows:

[0089] a) Target fiber; The length of the optical fiber is 50 m. The assumed strain distribution is shown in Figures 13A and 13B. Figure 13A is a diagram showing the distribution of the magnitude of strain in the longitudinal direction of the optical fiber. Figure 13B is a diagram showing the distribution of the change in the magnitude of strain in the longitudinal direction of the optical fiber. In both diagrams, the horizontal axis indicates the position in the longitudinal direction of the optical fiber (unit: m). From these figures, it can be seen that there is distortion in two sections of the optical fiber: in the section from 20.7 m to 25.9 m, there is a uniform distortion of 10 με, and in the section from 31.0 m to 34.1 m, there is a distortion of 10 με to 100 με with a constant slope of 30 με / m. Here, the relationship between distortion and Rayleigh frequency shift is given by the following equation (54).

number

[0090] b) chirped light pulse; Pulse width: 80 ns, pulse shape: rectangular, frequency sweep width: 4 GHz, pulse power: 100 mW, light source line width: 0.3 MHz, number of transmitted pulses: 16.

[0091] c) Frequency allocation Frequency sweep range of individual chirped light pulses: ν n ~ν n +4GHz, frequency interval between pulses: ν n -ν n―1 = 3.825 GHz, number of subbands (width 200 MHz, spacing: 25 MHz) that can be set with individual chirped optical pulses: 153, number of subbands that can be set with 16 chirped optical pulses: 2448.

[0092] d) Signal Processing The Rayleigh scattered light from the chirp pulse is received and demodulated for each subband of the chirp to obtain the Fourier spectrum (both the reference spectrum and the observed spectrum). The Rayleigh frequency shift is obtained by taking the correlation between the transformed reference spectrum and the observed spectrum. The slope parameter in the simulation is Δb = 0.5 GHz / m (Δβ = 4.87 cycles / m 2 ), b j =-60~60GHz / m(β j =-584.4~584.4cycles / m 2 ), j = j = 0, 1, ···, J-1, J = 241.

[0093] The simulation results are explained below. For comparison, Fig. 15 shows the results of applying a conventional method for detecting the spectral shift without considering the inclination. In the section from 31.0 m to 34.1 m, the inclination is about -4.5 GHz / m, but it is not detected at all in this section. Since the spatial resolution is 0.5 m, the detection limit for the inclination predicted from equation (20) is 1.28 GHz / m, and this limit is exceeded in the section from 31.0 m to 34.1 m.

[0094] Next, the results of detecting the Rayleigh frequency shift by spectrum transformation when using this detection method that takes the slope into account are shown in Figures 16A, 16B, 17A, 17B, 18A, and 18B. Figure 16A shows the detection result of the Rayleigh frequency shift. It is detected at all points except for four points (20.7 m, 25.9 m, 31.0 m, and 34.0 m) where the Rayleigh frequency shift changes in a step shape. Figure 16B shows the estimation error of the Rayleigh frequency shift. Except for two points (25.2 m and 33.1 m), it is within the range of the quantization error. The quantization error occurs because the frequency interval of the Fourier spectrum is 25 MHz, and the lag when taking the cross-correlation is an integer multiple of 25 MHz.

[0095] Figure 17A shows the estimation result of the slope of the Rayleigh frequency shift. It can be seen that the magnitude of the slope was estimated well except for one point (33.1 m). Figure 17B shows the peak value and threshold value of the cross-correlation coefficient.

[0096] 18A shows the distortion estimate converted from the Rayleigh frequency shift to the distortion according to equation (54), and FIG 18B shows the estimation error of the distortion estimate converted from the Rayleigh frequency shift to the distortion according to equation (54).

[0097] The simulation results described above show that by using this detection method, when the spatial resolution is 0.5 m, the Rayleigh frequency shift can be detected with high accuracy even when the strain slope is large, such as 30 με / m (the Rayleigh frequency shift slope is −4.5 GHz / m).

[0098] As mentioned above, we have shown that the introduction of analysis method N1 makes it possible to detect the Rayleigh frequency shift even when the change in strain or temperature changes locally. In other words, by using analysis method N1, we have shown that it is possible to measure the amount of shift including the slope by deforming the reference spectrum with chirps of various chirp rates and taking the cross-correlation between the reference spectrum and the observed spectrum. In addition, we have also clarified that in the above analysis method N1, when the change in strain or temperature has a slope, not only does the spectrum shift, but it also undergoes deformation due to the chirp. As a concrete result, compared with the case of a spatial resolution of 0.15 m, it was revealed that it is possible to detect the shift amount even in cases with a large gradient, such as a gradient of 50 GHz / m, which could not be detected by conventional detection methods. As described above, it has been shown that analytical method N1 can realize a Rayleigh intensity pattern measurement device that is effective for non-uniform strain distribution.

[0099] Furthermore, in the Rayleigh intensity pattern measurement device of the first embodiment, the second problem of sufficient measurement range is realized by a method of scanning multiple chirp light pulses in a stepwise manner (each step is treated as a pulsed chirp signal). However, in order to generate this chirp light pulse, an LD with a narrow linewidth is generally required. For example, the TGD-DAS technology of prior art number 6 described above requires an LD with a linewidth of 100 Hz. In the Rayleigh intensity pattern measurement device of the first embodiment, this linewidth condition is relaxed, and it is possible to handle 300 kHz, which is a linewidth that can be handled by commercially available LDs. Therefore, the details of these are described below.

[0100] As previously described and shown in Figures 6 and 7, in the Rayleigh intensity pattern measurement device of embodiment 1, in methods 1 to 3, multiple chirp pulses are used to drive the receiving system and heterodyne detection is used to acquire the Rayleigh backscattering signal. More specifically, a matched filter technique is used to generate simulated data from multiple chirp pulses, from which a set of subbands are extracted and combined to form a Rayleigh Intensity Pattern (RIP) for cross-correlation. Therefore, the details of how the Rayleigh intensity pattern is formed in the first embodiment will be specifically described below with reference to examples.

[0101] [Example of Rayleigh intensity pattern formation] First, the strain distribution data used in this simulation is shown in Figure 19. It is basically similar to the strain distribution explained in Figure 14 above. Here, the model is a strain distribution with a constant strain of 10με in the section 20m to 25m, and a non-uniform strain with a gradient of 30με / m in the spatial resolution space in the section 30m to 33m. There is no strain in sections other than those mentioned above (strain is zero).

[0102] Next, a method of scanning 16 multiple chirp pulses in a stepwise manner to simulate the laser oscillation light used in strain distribution measurement will be explained with reference to Figures 20 and 21. Figure 20 shows the main parameters used in this simulation. Figure 21 specifically shows some aspects of the multiple chirp pulses used in this simulation.

[0103] As can be seen from Figures 20 and 21, the chirp optical pulses are set to overlap in frequency (the overlap frequency band between adjacent pulses is 0.175 GHz) to generate a RIP with no frequency step between the multiple pulses. By scanning 16 multiple chirp pulses in a stepped manner, a frequency range of 0.1 GHz to 61.275 GHz, which is sufficient for measurement, was achieved, as will be explained in detail later. Here, as shown in Fig. 20, the laser linewidth in this embodiment is 300 kHz and the pulse width is 80 ns. Note that the laser linewidth is one of the most important parameters because it induces phase noise in the Rayleigh backscattered optical signal.

[0104] Next, the matched filter of this embodiment will be described with reference to Fig. 22. The matched filter for subband extraction is defined in the form shown in Fig. 22. A single-pulse chirp signal sweeps the 4 GHz frequency band and is expressed by the following equation (55).

number

[0105] As can be seen from Fig. 20 and Fig. 22, the subband width is 200 MHz, and the interval between adjacent bands is 25 MHz. Therefore, the matched filter s n (t), n = 1, 2, ..., 153 is determined according to the definitions given above.

[0106] From this, traces showing specific frequencies along the distance can be calculated by performing a cross-correlation of the acquired data with a matched filter. This procedure is called subband extraction, and as a result, a total of 2448 traces can be extracted from all the simulation data (see Figure 23). The center frequency of the subband is then used to represent the frequency of the corresponding trace, calculated by the following equation (56):

number

[0107] That is, the Rayleigh backscattering spectrum consists of 2448 traces, covering frequencies from 0.1 GHz to 61.275 GHz calculated by Equation (56). The RIP is defined as the intensity of the Rayleigh backscattering signal at the corresponding laser frequency, but since the laser frequency before chirp is 193548.487 GHz, the absolute frequency of the RIP is expressed as 193548.587 GHz to 193609.762 GHz. Therefore, the RIP can be stored as an absolute frequency. As long as the condition of the optical fiber does not change, the RIP can be obtained by sweeping the same absolute frequency even if the measuring equipment is changed. Examples of RIP at a distance of 10 m are shown in Figures 24A, 24B, and 24C. Here, Figure 24A shows the RIP when the laser linewidth is 2 kHz, Figure 24B shows the RIP when the laser linewidth is 100 kHz, and Figure 24C shows the RIP when the laser linewidth is 300 kHz. From these figures, it is clear that the effect of the laser linewidth can be almost ignored when the linewidth is between 2 kHz and 300 kHz.

[0108] As described above, in the Rayleigh intensity pattern measurement device of the first embodiment, it is possible to realize a sufficient measurement range by scanning multiple chirp optical pulses in a stepped manner. Also, it was shown that a RIP usable for measurement can be obtained even with a laser linewidth of 300 kHz. As described above, it has been found that the Rayleigh intensity pattern measurement device of this embodiment 1 can provide a high-speed, high-precision strain or temperature measurement device that simultaneously achieves the three features of high spatial resolution, high-speed measurement, and sufficient measurement range. Furthermore, it is possible to manage RIP information for a long period of time, such as several decades, and highly accurate RIP information can be obtained continuously during the measurement period. Furthermore, even if a non-uniform strain distribution occurs in an area below the spatial resolution, the frequency shift can be obtained by using a detection optimization analysis method.

[0109] In addition to the above, when measuring RIP with guaranteed absolute frequency using optical fiber, the accuracy of the absolute frequency can be guaranteed to be about the linewidth of the LD, the range of measured RIP includes a range of temperature and strain changes that is sufficient for practical use, and by providing a means to correct correlation failures that occur non-uniformly in the spatial distribution, it is possible to continuously obtain reliable correlation values ​​for the RIP measured initially, and high-speed measurements that can also accommodate dynamic strain changes are possible. Moreover, even if an LD with an insufficient linewidth, for example, of about 300 kHz, is used, it is possible to measure dynamic strain.

[0110] Embodiment 2 Next, a Rayleigh intensity pattern measurement device according to embodiment 2 will be described below with reference to the drawings, focusing on the differences from embodiment 1. The Rayleigh intensity pattern measurement device according to embodiment 2 differs only in that it uses analysis method N2, which is different from analysis method N1 for eliminating non-uniform distortion, among the above descriptions. Therefore, this analysis method N2 will be described below.

[0111] [Analysis method N2] In analysis method N2, in order to obtain an appropriate correlation even when non-uniform distortion exists, an approximation component decoding method (hereinafter also referred to as the ACD method; ACD: Approximation Components Decoding) based on the discrete wavelet transform technique described below is used. Below, the details of this method and an example of verification of this method using measurement data of Rayleigh frequency shift (hereinafter simply referred to as RFS) are described.

[0112] The ACD method is roughly composed of the following four steps. (a) Data preparation process (preprocessing process: obtaining data to be decrypted) (b) Data decomposition process (a process of decomposing the data into an approximate data part and a detailed data part at a predetermined decomposition level by using a discrete wavelet transform) (c) Data reconstruction step (a step of reconstructing the data by setting the detailed data part to zero and performing wavelet inverse transform) (d) a correlation coefficient calculation and determination step (a step of calculating a cross-correlation coefficient based on the reconstructed data, and calculating a correlation coefficient when the calculated correlation coefficient is equal to or greater than a threshold value);

[0113] Analysis method N2 is called the ACD method because it is used in particular in a process that combines the above steps (b) and (c), in which the original data is decomposed into an approximate data portion and a detailed data portion using a wavelet transform, the detailed data portion is set to zero, and only the approximate data portion is subjected to an inverse wavelet transform to reconstruct the data. By using this ACD method, an appropriate RFS can be calculated even when the correlation coefficient is calculated based on the signal data of Rayleigh scattered light that was detected as non-uniform distortion by the conventional method. Then, the target distortion can be detected based on this calculated RFS. Next, the details of each of the above steps will be explained in order below with reference to the figures.

[0114] [Data preparation process] Two frequency spectrum data (data 1, data 2) (see Fig. 25A and Fig. 25B) at a predetermined distance (for example, 512.5 m) are obtained from the raw data obtained at the non-uniform point in the area surrounded by the dotted long circle in Fig. 4, and after normalizing these data, if it is necessary to detect even smaller distortion, multiple-times interpolation processing is performed on the frequency step. For example, if the frequency step is changed from 0.2 GHz to 0.04 GHz, the minimum detectable distortion becomes 0.264 με from 1.32 με, and it becomes possible to detect even smaller values. Then, the cross-correlation coefficient (see Fig. 25C) is calculated. Here, for example, data 1 is the reference data, and data 2 is the measured data. Here, as shown in Fig. 25C, the cross-correlation coefficient shows a value below the empirically determined threshold (0.3), and it can be seen that detection is not performed well.

[0115] [Data decomposition process] The above frequency spectrum data can be regarded as a waveform containing many frequency components, but the non-uniform distortion distribution generates high-frequency components that distort the waveform and cause an inability to obtain proper correlation. Therefore, in order to remove the high-frequency components that distort the waveform, the above signal data is decomposed using the discrete wavelet transform (hereinafter, the discrete wavelet transform is also called DWT. Here, DWT: Discrete wavelet transform) described below. This method is a common method for decomposing signals in signal processing.

[0116] That is, as shown in FIG. 26, the signal data is divided into individual approximation data parts A, B, C, D ... n (n=1, 2, 3, . . .) and the high frequency part, the individual detailed data part D n (n=1, 2, 3, ...). More specifically, as shown in FIG. 26, each approximate data section A n (n=1, 2, 3, . . .) are the approximate data parts A of the higher levels. n+1 (n=1, 2, 3, ...) and detailed data section D n (n=1, 2, 3, ...). In this case, the maximum level depends on the signal length and the selected wavelet function (in wavelet transformation, the wavelet function to be used must be selected. Here, the orthogonal transform function coif2 was selected and used).

[0117] [Data reconstruction process] Next, the detailed data part D is obtained by decomposing the above data (using the wavelet function coif2). 1 , D 2 , D 3The detailed data portion can be easily removed by setting , ..., to 0 (zero). Therefore, the detailed data portion is reconstructed by performing an inverse wavelet transform using only the approximation data portion, which has less interference with the detailed data portion after the detailed data portion has been removed. By analyzing and extending the above ideas, we construct the main processing flow of the proposed ACD method.

[0118] [Correlation coefficient calculation and determination process] Next, based on the reconstructed data (two Rayleigh spectrum data), the cross-correlation of these data is taken to find the correlation coefficient. At this time, it should be noted that the correlation coefficient obtained differs depending on the decomposition level shown in FIG. 26 (see FIG. 27). The higher the decomposition level, the more signal data components are deleted, resulting in a higher correlation coefficient. On the other hand, taking into account actual cases, it is necessary to maintain as much detailed signal data as possible. In other words, it is necessary to keep the decomposition level low. In order to adjust this level, a threshold value for the correlation function is determined, and when the correlation coefficient exceeds this threshold, the calculation of the correlation coefficient is terminated.Then, by calculating the RFS based on the correlation coefficient thus determined, it becomes possible to measure the appropriate distortion value.

[0119] The entire process described above is shown in a data processing flowchart in the ACD method in FIG. First, the decomposition level n is set to n=1 (step S21). Next, two Rayleigh spectrum data are obtained (step S22). Next, the data at the decomposition level n of each data is decomposed by DWT, and the detailed data part of each data is set to zero and stored in the memory database (step S23). Next, the Rayleigh spectrum data of the previously stored data is reconstructed by inverse wavelet transform (step S24). Then, the cross-correlation coefficient of the two reconstructed data is calculated (step S25). The calculated cross-correlation coefficient value is compared with a threshold value (e.g., 0.4) (step S26), and if it is greater than the threshold value, the data processing is terminated. On the other hand, if the calculated cross-correlation coefficient value is smaller than the threshold value, the value of n is increased by 1 (step S27), and the process returns to step S23 of the above processing flow, and the processes from step S23 to step S27 are continued until the calculated cross-correlation coefficient value is greater than the threshold value.

[0120] This processing flow allows the data to be decomposed while retaining as many detailed components as possible. It also allows the distortion values ​​at all locations to be processed individually to determine the distortion value at the current measurement location, without using distortion values ​​at neighboring locations. This processing can save calculation time compared to the FCD method (fundamental components decoding method) that is based on FFT processing, which will be explained separately.

[0121] Next, in order to verify this ACD method, the results of processing the data in which non-uniform distortion was detected, as shown in FIG. 4A, will be described with reference to FIGS. 29A and 29B.

[0122] Figure 29A is a graph showing the distribution of strain generated in the optical fiber corresponding to the optical fiber position (see the distance shown on the horizontal axis) obtained by using the ACD method. In addition, in obtaining the strain distribution, a threshold value of 0.4 was given here (see Figure 28). Fig. 29B shows the distribution of correlation coefficients corresponding to the optical fiber position obtained using this ACD method. For reference, Fig. 29C shows a graph of the strain distribution obtained using the conventional method shown in Fig. 4A, and Fig. 29D shows the corresponding distribution of correlation coefficients.

[0123] From these figures (FIGS. 29A to 29D), it can be seen that the ACD method is effective even when there is non-uniform distortion. In addition, in FIG. 29B and FIG. 29D, the value 0.3 empirically used as the threshold value is indicated by a dotted line.

[0124] To examine the effectiveness of the ACD method from another perspective, the strain distribution shown in Figure 29A above was compared with data obtained using the Brillouin scattering method. The resulting data is shown in Figure 30A. Similarly, Figure 30B shows a graph in which the two types of data shown in Figure 30A were smoothed using the moving average method.

[0125] Comparing the two types of data from these graphs, it can be determined that the strain data processed by the ACD method described above based on Rayleigh scattered light is able to measure the strain distribution equally well when compared with the data obtained using the Brillouin scattering method. Furthermore, when this ACD method was applied to an area with non-uniform distortion, the success rate (unit: %) was calculated as the ratio of the number of points where treatment was successful to the total number of treatment points (the total number of points where treatment was successful and points where treatment was unsuccessful), and the success rate was examined, revealing a high value of 90.46%. From the above, it is clear that the ACD method is effective.

[0126] [FCD method] As the above analysis method N2, the Fundamental Components Decoding method (also called FCD method, FCD: Fundamental Components Decoding), which is an analysis method similar to the above ACD method, will be described below. That is, instead of the ACD method, the FCD method may be used as analysis method N2. Below, the contents of the FCD method will be described, focusing on the differences from the ACD method.

[0127] The ACD method uses the Discrete Wavelet Transform (DWT) as the basic analysis method, whereas the FCD method uses the Fast Fourier Transform (FFT). In principle, the ACD method extracts the approximate data portion of the RIP, whereas the FCD method uses a low-pass filter (hereafter also referred to as LPF) to extract elements of the low-frequency region of the RIP. Meanwhile, the ACD method requires the selection of the wavelet function to be used, whereas the FCD method does not. Conversely, the ACD method does not require an initial value, whereas the FCD method requires the initial value to be given. Therefore, the FCD method will be explained in detail below, starting from the above-mentioned fundamental aspects.

[0128] If the object being measured has a non-uniform strain distribution, its Rayleigh frequency spectrum will have high-frequency components, which will distort the strain waveform. On the other hand, when there is a clear correlation between the data of two Rayleigh frequencies, the spectra are considered to have a common fundamental component (basic frequency component. Specifically, see the frequency components at the points marked with four arrows in the figure), as shown in an example in Figure 31.

[0129] Therefore, this method seeks to obtain correlation using these common fundamental components, especially those in the low-frequency region. In order to obtain such common fundamental components, an LPF is used to cut the high-frequency components of the Rayleigh frequency signal (in this case, it is necessary to select the optimal cutoff frequency). In this case, the lower the cutoff frequency, the higher the correlation coefficient between the two Rayleigh frequency spectra, and the higher the cutoff frequency, the lower the correlation coefficient between the two Rayleigh frequency spectra. In order to select the optimal cutoff frequency (to find an appropriate value), the following two conditions were taken into consideration. The first is that the correlation coefficient must exceed a certain threshold, and the second is that the shift frequency at the current observation location must be close to the shift frequency at the previous observation location. The processing contents generally described above are shown in the form of a data processing flow in Fig. 32. In the following, the data processing flow in the FCD method will be described with reference to Fig. 32.

[0130] 32 shows a process flow for calculating a series of Rayleigh frequency shifts that have a correlation coefficient higher than a given threshold A. Then, these Rayleigh frequency shifts are compared with the Rayleigh frequency shifts at the previous location (location number: n-1), and the one with the smallest absolute difference is selected and set as the Rayleigh frequency shift at the current location (location number: n).

[0131] Specifically, as shown in FIG. 32, two Rayleigh spectrum data (data 1, data 2) are passed through an LPF (low pass filter) (step S31), data 1 after passing through the filter and data 2 after passing through the filter are obtained, and the correlation between these two data is calculated (step S32). After that, the correlation coefficient obtained by calculating the correlation is compared with a predetermined threshold value A (step S33). If the correlation coefficient is smaller than the predetermined threshold value A, If it is small Then, the cutoff frequency of the LPF is changed (step S34), and the two Rayleigh spectrum data (data 1, data 2) are passed through the LPF (low pass filter) again, and the correlation between data 1 and data 2 after passing through the filter is calculated again and compared with threshold A. If the correlation coefficient obtained here is equal to or greater than the predetermined threshold A, the value of the Rayleigh frequency shift is obtained from the correlation coefficient at this time (step S35) and stored as Rayleigh frequency shift data in a memory or the like. Then, the absolute value of the difference between the Rayleigh frequency shift data of the previous location obtained in advance and the Rayleigh frequency shift data obtained by the processing of steps S31 to S35 is obtained, and the one with the smallest value is selected (step S36) to be set as the Rayleigh frequency shift of the current location (position number: n).

[0132] In order to determine the initial value (i.e., the value at position number 0 when n=1), the processing starts from the first location, which is the location of the known Rayleigh frequency shift. The correlation coefficient obtained through this processing by the FCD method is shown in Figure 33A. For comparison, the data obtained without processing by the FCD method is shown in Figure 33B. It can be seen that the correlation coefficient is clearly improved in the data processed by the FCD method.

[0133] Next, based on the frequency shift data obtained by processing using the FCD method, strain distribution data was obtained at distance positions including the distances where non-uniform strain was previously shown, and is shown in Figures 34A and 34B. Figure 34A shows the strain distribution data obtained by processing using the FCD method as is, plotted as is. Figure 34B shows the data obtained by smoothing the data in Figure 34A by taking a moving average. It can be seen that both data are significantly improved by using the FCD method, compared to the strain distribution data previously shown in Figure 4A. This is the same as the ACD method described above.

[0134] Finally, in order to demonstrate the improvement effect of using the FCD method, the spectral distribution data of the spectral magnitude and correlation coefficient obtained when the FCD method was applied, and the spectral distribution data of the spectral magnitude and correlation coefficient obtained when no processing was performed (FCD method was not applied) are shown in Figures 35A, 35B, 35C, and 35D, respectively.

[0135] As described above, it is understood that the analytical method N2 has the same effect as the analytical method N1 described in the first embodiment. Also, it has been described that both the ACD method and the FCD method are effective for the analytical method N2. The method to be selected can be determined based on the purpose of use, the availability of tools used for each method (e.g., FFT for the FCD method), and other factors. Even when this analysis method N2 is used, the same effects as those of the first embodiment can be obtained.

[0136] Although various exemplary embodiments and examples are described in this application, the various features, aspects, and functions described in one or more embodiments are not limited to application to a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless modified examples not illustrated are assumed within the scope of the technology disclosed in the present specification. For example, in the above description, the physical quantity to be measured is mainly strain, but the same discussion can be made even if strain is replaced with temperature. This includes cases in which at least one component is modified, added, or omitted, and further cases in which at least one component is extracted and combined with a component of another embodiment. [Explanation of symbols]

[0137] 1 LD, 2 optical switch, 3 erbium-doped optical fiber amplifier (EDFA), 4 optical coupler, 5 reference optical fiber, 6 2ch AD converter, 7 first calculation unit, 8 memory database (memory DB), 9 second calculation unit (calculation unit), 11 light source unit, 12 transmitting / receiving optical unit, 13 receiving unit, 14 Rayleigh intensity pattern digital processing unit (RIP digital processing unit), 100 Rayleigh intensity pattern measurement device (RIP measurement device)

Claims

1. a light source unit including a wavelength-tunable LD, a controller capable of changing the frequency of a laser beam emitted from the wavelength-tunable LD, and a local oscillator; an optical coupler that inputs the laser light oscillated from the light source unit into an optical fiber and outputs Rayleigh scattered light from the optical fiber through a path different from the input path of the laser light; a receiving unit that coherently receives the local light from the light source unit and the Rayleigh scattered light from the optical coupler; a Rayleigh intensity pattern digital processing unit having an AD converter that performs AD conversion of a signal output from the receiving unit, a first calculation unit that performs a predetermined calculation on the signal converted by the AD converter, a memory database to which the signal calculated by the first calculation unit is input and stored, and a second calculation unit that performs a predetermined calculation on the basis of the data stored in the memory database, wherein the second calculation unit performs a predetermined correction for the measurement position from two different Rayleigh intensity patterns at a predetermined measurement position obtained from the electric field signal of the Rayleigh scattered light, and stores in the memory database the cross-correlation coefficient when the calculated cross-correlation coefficient is equal to or greater than a predetermined threshold value; Matched filters, Equipped with The absolute frequency of the tunable LD is controlled, and the frequency of the laser oscillated from the tunable LD is scanned in a stepwise manner, and a chirp signal pulsed at each step is received by the receiving unit and then synthesized by the first calculation unit; and The chirp signal is set by using a plurality of window functions according to the subbands extracted by the matched filter, and the Rayleigh intensity pattern including the subbands recombined by using the center frequency of the extracted subbands is preserved at the absolute frequency; A Rayleigh intensity pattern measuring apparatus characterized in that a strain distribution or a temperature distribution of a measured object is determined by determining a Rayleigh frequency shift based on a cross-correlation coefficient that is equal to or greater than a predetermined threshold value.

2. The Rayleigh intensity pattern digital processing unit comprises an FPGA or an ASIC; The Rayleigh intensity pattern measurement device according to claim 1, characterized in that the first calculation unit acquires the frequency components of the subbands by digital processing, and the second calculation unit performs calculation processing including resampling, which is an interpolation process of the signal on the frequency axis, and calculation of the cross-correlation coefficient.

3. The Rayleigh intensity pattern measurement device according to claim 2, characterized in that the frequency components of the subbands acquired by the first calculation unit are formed based on a non-squared Rayleigh intensity pattern signal including phase information.

4. The second calculation unit includes:

4. The Rayleigh intensity pattern measurement device according to claim 1, wherein the measurement location and the Rayleigh frequency shift are analyzed by cross-correlation obtained using both distance and frequency as variables in each measurement based on the two Rayleigh intensity patterns.

5. 2. A Rayleigh intensity pattern measurement method for performing a Rayleigh intensity pattern measurement using the Rayleigh intensity pattern measurement apparatus according to claim 1, comprising: A Rayleigh intensity pattern measurement method characterized in that, when the calculated cross-correlation coefficient is smaller than a predetermined threshold, a Rayleigh intensity pattern different from the Rayleigh intensity pattern calculated by the second calculation unit is used to perform repeated calculations until a cross-correlation coefficient greater than or equal to a given threshold is obtained.

6. 2. A Rayleigh intensity pattern measurement method for performing a Rayleigh intensity pattern measurement using the Rayleigh intensity pattern measurement apparatus according to claim 1, comprising: When the calculated cross-correlation coefficient is smaller than a predetermined threshold value, A Rayleigh intensity pattern measurement method characterized in that, based on a reference spectrum obtained from the electric field of reference data for Rayleigh scattered light and an observed spectrum obtained from the electric field of observed data, the frequency of each spectrum is shifted to determine the cross-correlation coefficient of each spectrum, and a strain distribution of a measured object is detected based on the determined cross-correlation coefficient that is equal to or greater than a predetermined threshold value.

7. 2. A Rayleigh intensity pattern measurement method for performing a Rayleigh intensity pattern measurement using the Rayleigh intensity pattern measurement apparatus according to claim 1, comprising the steps of: When the calculated cross-correlation coefficient is smaller than a predetermined threshold value, A Rayleigh intensity pattern measurement method characterized by: modifying a reference spectrum obtained from the electric field of reference data for Rayleigh scattered light with chirp signals of various chirp rates; and measuring the frequency shift of the Rayleigh scattered light of a measured object by taking the cross-correlation between the modified reference spectrum and an observed spectrum.

8. 2. A Rayleigh intensity pattern measurement method for performing a Rayleigh intensity pattern measurement using the Rayleigh intensity pattern measurement apparatus according to claim 1, comprising the steps of: When the calculated cross-correlation coefficient is smaller than a predetermined threshold value, decomposing the data into an approximate data portion and a detailed data portion at a predetermined decomposition level; A step of setting the detailed data portion to zero and reconstructing the data by inverse wavelet transform; A step of calculating a cross-correlation coefficient based on the reconstructed data, and calculating a correlation coefficient when the calculated correlation coefficient is equal to or greater than the threshold value; A Rayleigh intensity pattern measuring method, comprising: measuring an amount of frequency shift of Rayleigh scattered light having a slope by using a discrete wavelet transform technique including:

9. 2. A Rayleigh intensity pattern measurement method for performing a Rayleigh intensity pattern measurement using the Rayleigh intensity pattern measurement apparatus according to claim 1, comprising the steps of: When the calculated cross-correlation coefficient is smaller than a predetermined threshold value, a Rayleigh intensity pattern measurement method comprising: passing two Rayleigh spectrum data for a specified measurement location through a low-pass filter, calculating the correlation between the two Rayleigh spectrum data after passing through the low-pass filter, comparing the magnitude of the obtained correlation coefficient with another threshold different from the predetermined threshold, and if the correlation coefficient is smaller than the other threshold, changing the cutoff frequency of the low-pass filter, passing the two Rayleigh spectrum data through the low-pass filter again, calculating the correlation between the two Rayleigh spectrum data after passing through the low-pass filter and comparing it with the other threshold, and if the obtained correlation coefficient is equal to or greater than the other threshold, determining a Rayleigh frequency shift value from the obtained correlation coefficient and saving it in the memory database, and also calculating multiple absolute values ​​of the difference between the Rayleigh frequency shift data stored before the currently saved Rayleigh frequency shift data and the currently saved Rayleigh frequency shift data, selecting the smallest absolute value among the obtained absolute values ​​as the Rayleigh frequency shift for the specified measurement location.

Images