Distributed Fiber Optic Sensing Method Based on Kalman Filtering in the Optical Frequency Domain

Abstract

The optical frequency domain distributed optical fiber sensing method based on Kalman filtering belongs to the field of distributed optical fiber sensing technology; the method uses Kalman filtering to process optical frequency domain distributed optical fiber sensing data, overcoming the problem of large amount of calculation of traditional cross-correlation algorithm under long-distance and large-range sensing. The present invention only adds Kalman filtering to the traditional cross-correlation algorithm for dynamic prediction without changing the structure and frequency sweep range of the optical frequency domain distributed optical fiber sensing system, thereby reducing the calculation time and improving the demodulation rate of OFDR; moreover, the prediction error of Kalman filtering and the signal-to-noise ratio of cross-correlation calculation are used as the judgment conditions of cross-correlation quality, thereby improving the accuracy of the demodulation result.

Description

Technical Field

[0001] The invention belongs to the technical field of distributed optical fiber sensing, and in particular relates to an optical frequency domain distributed optical fiber sensing method based on Kalman filtering. Background Art

[0002] Optical frequency domain reflectometry has become a typical high-precision fiber-optic distributed sensing technology due to its high spatial resolution, spectral resolution and continuous measurement characteristics. Distributed measurement of temperature and strain is of great significance in the structural health monitoring of large-scale key functional components in large-scale equipment.

[0003] With the increasing application of OFDR distributed sensing technology, the demand for OFDR technology's sensing density, sensing accuracy and continuous sensing has grown rapidly in recent years. The traditional method uses a weak fiber Bragg grating string as the sensing fiber. Only the position with the fiber Bragg grating on the fiber has the ability to sense strain or temperature, while the ordinary fiber between the fiber Bragg gratings does not have the sensing ability. Therefore, it still belongs to quasi-distributed sensing and cannot achieve true distributed sensing.

[0004] However, the current demodulation rate of OFDR is the main obstacle that limits its application in real-time monitoring of structural health. The main problem that limits the demodulation rate of OFDR is that the traditional cross-correlation algorithm will match the wrong measurement spectrum when measuring the temperature or strain change of the optical fiber segment and the nonlinear frequency sweep jitter of the light source. At this time, a neighborhood search is required to match the correct measurement spectrum. Each search requires a cross-correlation calculation between the reference spectrum and the measurement spectrum to determine whether the searched measurement spectrum is correct. If the neighborhood range is too small, it is difficult to search for the correct measurement spectrum. If the neighborhood range is too large, a large number of cross-correlation calculations are required, which makes the system solution time longer, limits the demodulation rate of the system, and further limits its application in rapid response in structural health monitoring. Summary of the invention

[0005] In order to solve the problem of low demodulation rate of the above-mentioned OFDR distributed sensing technology, the present invention proposes a method for improving the OFDR demodulation rate based on Kalman filtering to achieve the purpose of rapid response. In the traditional algorithm, the reference spectrum and the measured spectrum are cross-correlated, and the spectrum offset is obtained from the cross-correlation spectrum peak position. When the similarity between the measured spectrum and the reference spectrum is low, the maximum peak of the cross-correlation spectrum is close to the maximum value of the noise, and the low signal-to-noise ratio makes it difficult to accurately identify the cross-correlation spectrum peak position. Therefore, in the method for improving the OFDR demodulation rate based on Kalman filtering in the present invention, when obtaining the cross-correlation offset of the reference spectrum relative to the measured spectrum, Kalman filtering is first used to predict the offset estimate of the spatial position, and the error between the offset estimate and the cross-correlation offset is calculated. Only when the signal-to-noise ratio of the cross-correlation calculation result of a certain spatial position is lower than the signal-to-noise ratio threshold and the prediction error is higher than the prediction error threshold, the neighborhood search is entered. In the neighborhood search process, if the prediction error is lower than the prediction error threshold, the neighborhood search is jumped out, and a complete global search is not required, thereby reducing the number of cross-correlation calculations, reducing the demodulation time, and achieving the purpose of improving the OFDR demodulation rate.

[0006] Technical solution of the present invention

[0007] The optical frequency domain distributed optical fiber sensing method based on Kalman filtering is characterized in that the method comprises the following steps:

[0008] Step 1: Before strain or temperature acts on the sensing optical fiber, an optical frequency domain distributed optical fiber sensing instrument is used to obtain the original measurement data, and the original signal is subjected to Fourier transform to obtain the complex amplitude of Rayleigh backscattering in the distance domain;

[0009] Step 2: After the strain or temperature acts on the sensing optical fiber, the original measurement data is obtained by using an optical frequency domain distributed optical fiber sensing instrument, and the original signal is subjected to Fourier transform to obtain the complex amplitude of the backscattering Rayleigh in the distance domain;

[0010] Step 3: Complete the spatial resolution setting, perform inverse Fourier transform on the complex amplitude of backscattered Rayleigh scattering at each spatial position in the distance domain obtained in step 1, obtain the Rayleigh scattering spectrum at each spatial position before strain or temperature acts on the sensing optical fiber, and use it as the reference spectrum. At the center of the reference spectrum at each spatial position, select a local spectrum fragment as the local reference spectrum at each spatial position;

[0011] Step 4: Initialize the Kalman filter parameters, including the initial estimated a posteriori estimation error covariance, the initial a posteriori state estimate, the process noise variance, the measurement noise variance, and the prediction error threshold;

[0012] Step 5: Using the spatial resolution set in step 3, perform inverse Fourier transform on the complex amplitude of backscattered Rayleigh scattering at a certain spatial position in the distance domain in step 2, and obtain the Rayleigh scattering spectrum at a certain spatial position after strain or temperature acts on the sensing optical fiber as the measurement spectrum;

[0013] Step 6: perform cross-correlation calculation on the measured spectrum at a certain spatial position and the local reference spectrum corresponding to the spatial position, record the cross-correlation offset, and calculate the ratio of the amplitude of the highest peak to the amplitude of the second highest peak in the cross-correlation calculation result as the signal-to-noise ratio;

[0014] Step 7, substitute the cross-correlation offset as the measured value into the Kalman filter equation to calculate the estimated value of the offset of the measured spectrum at a certain spatial position relative to the reference spectrum;

[0015] Step 8: Calculate the absolute value of the difference between the offset estimate in step 7 and the cross-correlation offset as the prediction error. If the signal-to-noise ratio is greater than the signal-to-noise ratio threshold or the prediction error is less than the prediction error threshold, execute steps 14 to 15. If the signal-to-noise ratio is less than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, execute steps 9 to 15.

[0016] Step 9, enter the neighborhood search. After the temperature or strain acts on the sensing optical fiber, the current spatial position is offset by equal data length in the distance domain relative to the starting position according to +1, -1, +2, -2, ..., +N, -N, and the backscattered complex amplitude of the current spatial position after the offset is obtained. The integer x represents the offset. The backscattered complex amplitude of the current spatial position is inversely transformed by Fourier to obtain the measured spectrum of the current spatial position.

[0017] Step 10, performing cross-correlation calculation on the measured spectrum obtained in step 9 and the local reference spectrum, recording the cross-correlation offset, and calculating the ratio of the amplitude of the highest peak to the amplitude of the second highest peak in the cross-correlation calculation result as the signal-to-noise ratio;

[0018] Step 11, substituting the cross-correlation offset as the measured value into the Kalman filter equation to calculate the estimated value of the offset of the measured spectrum at the current spatial position relative to the reference spectrum;

[0019] Step 12, calculate the absolute value of the difference between the offset estimate value in step 11 and the cross-correlation offset as the prediction error, if the prediction error is less than the prediction error threshold, then record the cross-correlation offset of the measured spectrum at the current spatial position relative to the local reference spectrum and the offset x of the current spatial position relative to the starting position in the distance domain, exit the neighborhood search, and execute steps 13 to 15, if the signal-to-noise ratio is greater than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, then record the cross-correlation offset of the measured spectrum corresponding to the current spatial position relative to the local reference spectrum when the signal-to-noise ratio is maximum and the offset x of the current spatial position relative to the starting position in the distance domain, if the signal-to-noise ratio is less than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, then record the cross-correlation offset of the measured spectrum corresponding to the current spatial position with the local reference spectrum when the prediction error is minimum and the offset x of the current spatial position relative to the starting position in the distance domain, repeat steps 9 to 12 until the search from -N to +N is completed, and then execute steps 13 to 15;

[0020] Step 13, calculate the sum of the position offset cumulative amount of the starting position in the distance domain and the offset x of the current spatial position relative to the starting position as the new position offset cumulative amount;

[0021] Step 14: If the neighborhood search is performed, the cross-correlation offset is used as the spectral offset of the measured spectrum relative to the reference spectrum. If the neighborhood search is not performed, the cross-correlation offset is used as the spectral offset of the measured spectrum relative to the reference spectrum. According to the relationship between stress or temperature and the spectral offset, the stress or temperature at a certain spatial position is converted;

[0022] Step 15: According to the relationship between stress or temperature and the expansion coefficient of the sensing optical fiber, the sum of the position offset generated at a certain spatial position and the cumulative position offset of a certain spatial position in the distance domain is calculated as the cumulative position offset of the next sensing unit in the distance domain, and the position of the next sensing unit of the current sensing unit in the distance domain is obtained;

[0023] Step 16: Repeat steps 6 to 15 to obtain the stress or temperature at each spatial position in the sensing area.

[0024] Among them, the Kalman filter prior state equation expression in step 7 is The prior estimation error covariance equation expression is The Kalman gain equation is The expression of the posterior state estimation equation is The a posteriori estimation error covariance equation is i-1 and i represent the previous sensor unit and the current sensor unit respectively. represents the prior state estimate, represents the posterior state estimate, represents the prior estimation error covariance, P i represents the posterior estimation error covariance, K i represents the Kalman gain, z i represents the measured value (59) in step 10, which is the cross-correlation offset (45) in this method, Q represents the process noise variance, and R represents the measurement noise variance.

[0025] Among them, the a posteriori estimation error covariance of the initial estimate in step 4 should be 80 to 100 times the sensor accuracy, the initial a posteriori state estimate should be 0, the prediction error threshold (52) should be 3 to 5 times the sensor accuracy, and for temperature measurement, the process noise variance should be 1°C 2 ~2.5℃ 2 , the measurement noise variance should be 6℃ 2 ~10℃ 2 , for strain measurement, the process noise variance should be 10με 2 ~25με 2 , the measurement noise variance should be 60με 2 ~100με 2 .

[0026] Wherein, in step 9, N should be 10% to 200% of the number of data points contained in the length of the sensing unit and rounded up to an integer.

[0027] The advantages of the present invention are:

[0028] 1. Without changing the OFDR system structure and frequency sweep range, dynamic prediction is performed by adding Kalman filtering, which greatly reduces the number of cross-correlation calculations, reduces calculation time, and improves real-time performance.

[0029] 2. The prediction error of Kalman filtering and the signal-to-noise ratio of cross-correlation calculation are used as the judgment conditions of cross-correlation quality to improve the accuracy of the extraction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a schematic diagram of the local reference spectrum and the reference spectrum;

[0031] In the figure: 11. Reference spectrum, 12. Local reference spectrum.

[0032] Figure 2 It is a schematic diagram of selecting the sensing interval when entering the neighborhood search;

[0033] In the figure: 21. starting position, 21+1. sensing interval in the distance domain where the offset x of the current spatial position relative to the starting position is +1, 21-1. sensing interval in the distance domain where the offset x of the current spatial position relative to the starting position is -1, 21+N. sensing interval in the distance domain where the offset x of the current spatial position relative to the starting position is +N, 21-N. sensing interval in the distance domain where the offset x of the current spatial position relative to the starting position is -N.

[0034] Figure 3 It is a schematic diagram of the cross-correlation calculation results of the measured spectrum and the local reference spectrum;

[0035] In the figure: 31. The amplitude of the highest peak in the cross-correlation calculation results, 32. The amplitude of the second highest peak in the cross-correlation calculation results.

[0036] Figure 4 This is a workflow diagram for improving OFDR demodulation rate based on Kalman filtering. DETAILED DESCRIPTION

[0037] The specific implementation of the present invention is further described in detail below with reference to the accompanying drawings:

[0038] The optical frequency domain distributed optical fiber sensing method based on Kalman filtering is characterized in that the method comprises the following steps:

[0039] Step 1: Before strain or temperature acts on the sensing optical fiber, an optical frequency domain distributed optical fiber sensing instrument is used to obtain the original measurement data, and the original signal is subjected to Fourier transform to obtain the complex amplitude of Rayleigh backscattering in the distance domain;

[0040] Step 2: After the strain or temperature acts on the sensing optical fiber, the original measurement data is obtained by using an optical frequency domain distributed optical fiber sensing instrument, and the original signal is subjected to Fourier transform to obtain the complex amplitude of the backscattering Rayleigh in the distance domain;

[0041] Step 3, complete the spatial resolution setting, perform inverse Fourier transform on the complex amplitude of backscattered Rayleigh scattering at each spatial position in the distance domain obtained in step 1, obtain the Rayleigh scattering spectrum at each spatial position before strain or temperature acts on the sensing optical fiber, as the reference spectrum 11, select a local spectrum fragment at the center of the reference spectrum at each spatial position as the local reference spectrum 12 at each spatial position;

[0042] Step 4: Initialize the Kalman filter parameters, including the initial estimated a posteriori estimation error covariance, the initial a posteriori state estimate, the process noise variance, the measurement noise variance, and the prediction error threshold;

[0043] Step 5: Using the spatial resolution set in step 3, perform inverse Fourier transform on the complex amplitude of backscattered Rayleigh scattering at a certain spatial position in the distance domain in step 2, and obtain the Rayleigh scattering spectrum at a certain spatial position after strain or temperature acts on the sensing optical fiber as the measurement spectrum;

[0044] Step 6: perform cross-correlation calculation on the measured spectrum at a certain spatial position and the local reference spectrum 12 corresponding to the spatial position, record the cross-correlation offset, and calculate the ratio of the amplitude 31 of the highest peak to the amplitude 32 of the second highest peak in the cross-correlation calculation result as the signal-to-noise ratio;

[0045] Step 7, substituting the cross-correlation offset as a measured value into the Kalman filter equation to calculate an estimated value of the offset of the measured spectrum at a certain spatial position relative to the reference spectrum 11;

[0046] Step 8: Calculate the absolute value of the difference between the offset estimate in step 7 and the cross-correlation offset as the prediction error. If the signal-to-noise ratio is greater than the signal-to-noise ratio threshold or the prediction error is less than the prediction error threshold, execute steps 14 to 15. If the signal-to-noise ratio is less than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, execute steps 9 to 15.

[0047] Step 9, enter the neighborhood search. After the temperature or strain acts on the sensing optical fiber, the current spatial position 21x is offset by equal data length in the distance domain relative to the starting position 21 according to +1, -1, +2, -2, ..., +N, -N to obtain the back Rayleigh scattering complex amplitude of the current spatial position 21x after the offset. The integer x represents the offset. The back Rayleigh scattering complex amplitude of the current spatial position 21x is inversely transformed by Fourier transform to obtain the measured spectrum of the current spatial position 21x.

[0048] Step 10, performing cross-correlation calculation on the measured spectrum obtained in step 9 and the local reference spectrum 12, recording the cross-correlation offset, and calculating the ratio of the amplitude 31 of the highest peak to the amplitude 32 of the second highest peak in the cross-correlation calculation result as the signal-to-noise ratio;

[0049] Step 11, substitute the cross-correlation offset as the measured value into the Kalman filter equation to calculate the estimated value of the offset of the measured spectrum at the current spatial position 21x relative to the reference spectrum 11;

[0050] Step 12: Calculate the absolute value of the difference between the offset estimate in step 11 and the cross-correlation offset as the prediction error. If the prediction error is less than the prediction error threshold, record the cross-correlation offset of the measured spectrum at the current spatial position 21x relative to the local reference spectrum 12 and the offset x of the current spatial position 21x relative to the starting position 21 in the distance domain, exit the neighborhood search, and execute steps 13 to 15. If the signal-to-noise ratio is greater than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, record the corresponding measurement at the current spatial position 21x when the signal-to-noise ratio is maximum. The cross-correlation offset of the spectrum relative to the local reference spectrum 12 and the offset x of the current spatial position 21x relative to the starting position 21 in the distance domain. If the signal-to-noise ratio is less than the signal-to-noise ratio threshold and the prediction error is greater than the prediction error threshold, then record the cross-correlation offset of the measured spectrum corresponding to the current spatial position 21x and the local reference spectrum 12 when the prediction error is minimum and the offset x of the current spatial position 21x relative to the starting position 21 in the distance domain, repeat steps 9 to 12 until the search from -N to +N is completed, and then execute steps 13 to 15;

[0051] Step 13, calculate the sum of the position offset cumulative amount of the starting position 21 in the distance domain and the offset x of the current spatial position 21x relative to the starting position as the new position offset cumulative amount;

[0052] Step 14: If the neighborhood search is performed, the cross-correlation offset is used as the spectral offset of the measured spectrum relative to the reference spectrum 11; if the neighborhood search is not performed, the cross-correlation offset is used as the spectral offset of the measured spectrum relative to the reference spectrum 11, and the stress or temperature at a certain spatial position is converted according to the relationship between the stress or temperature and the spectral offset;

[0053] Step 15: According to the relationship between stress or temperature and the expansion coefficient of the sensing optical fiber, the sum of the position offset generated at a certain spatial position and the cumulative position offset of a certain spatial position in the distance domain is calculated as the cumulative position offset of the next sensing unit in the distance domain, and the position of the next sensing unit of the current sensing unit in the distance domain is obtained;

[0054] Step 16: Repeat steps 6 to 15 to obtain the stress or temperature at each spatial position in the sensing area.

[0055] Among them, the Kalman filter prior state equation expression in step 7 is The prior estimation error covariance equation expression is The Kalman gain equation is The expression of the posterior state estimation equation is The a posteriori estimation error covariance equation is i-1 and i represent the previous sensor unit and the current sensor unit respectively. represents the prior state estimate, represents the posterior state estimate, represents the prior estimation error covariance, P i represents the posterior estimation error covariance, K i represents the Kalman gain, z i represents the measured value 59 in step 10, which is the cross-correlation offset 45 in this method, Q represents the process noise variance, and R represents the measurement noise variance.

[0056] Among them, the a posteriori estimation error covariance of the initial estimate in step 4 should be 80 to 100 times the sensor accuracy, the initial a posteriori state estimate should be 0, the prediction error threshold 52 should be 3 to 5 times the sensor accuracy, and for temperature measurement, the process noise variance should be 1°C 2 ~2.5℃ 2 , the measurement noise variance should be 6℃ 2 ~10℃ 2 , for strain measurement, the process noise variance should be 10με 2 ~25με 2 , the measurement noise variance should be 60με 2 ~100με 2 .

[0057] Wherein, in step 9, N should be 10% to 200% of the number of data points contained in the length of the sensing unit and rounded up to an integer.

[0058] Working process of the present invention:

[0059] Before strain or temperature acts on the sensing fiber, the frequency-domain distributed optical fiber sensing instrument sweep light source performs a frequency sweep and measures the time-domain interference sensing signal of the sensing fiber as a reference signal. Then, before strain or temperature acts on the sensing fiber, the frequency-domain distributed optical fiber sensing instrument sweep light source performs another frequency sweep and measures the time-domain interference sensing signal of the sensing fiber as a measurement signal. The reference signal and the measurement signal are processed according to steps 1 to 16 of the distributed sensing information recovery method of the present invention to obtain strain or temperature information at different spatial positions along the sensing.

[0060] The innovative point and technical effect of the optical frequency domain distributed optical fiber sensing method based on Kalman filtering are: using Kalman filtering to judge the accuracy of the reference spectrum and the measured spectrum offset, the number of cross-correlation calculations can be greatly reduced by adding Kalman filtering dynamic prediction without changing the OFDR system structure. Moreover, the Kalman filtering dynamic prediction result is used as an auxiliary judgment condition for the quality of the cross-correlation result, which improves the matching accuracy of the measured spectrum and the reference spectrum in the neighborhood search, and reduces the number of neighborhood searches when the cross-correlation result is within an acceptable range. The present invention can effectively reduce the demodulation time of optical frequency domain distributed optical fiber sensing signals and improve the real-time performance of the sensing system.

Claims

1. An optical frequency domain distributed optical fiber sensing method based on Kalman filtering, characterized in that The method comprises the following steps: Step 1: Before strain or temperature acts on the sensing optical fiber, an optical frequency domain distributed optical fiber sensing instrument is used to obtain the original measurement data, and the original signal is subjected to Fourier transform to obtain the complex amplitude of Rayleigh backscattering in the distance domain; Step 2: After the strain or temperature acts on the sensing optical fiber, the original measurement data is obtained by using an optical frequency domain distributed optical fiber sensing instrument, and the original signal is subjected to Fourier transform to obtain the complex amplitude of the backscattering Rayleigh in the distance domain; Step 3: Complete the spatial resolution setting, perform inverse Fourier transform on the complex amplitude of backscattered Rayleigh scattering at each spatial position in the distance domain obtained in step 1, obtain the Rayleigh scattering spectrum of the optical fiber at each spatial position before strain or temperature acts on the sensing optical fiber, and use it as the reference spectrum (11). At the center of the reference spectrum at each spatial position, select a local spectrum fragment as the local reference spectrum at each spatial position (12); Step 4, initializing Kalman filter parameters, including the initial estimated a posteriori estimation error covariance, the initial a posteriori state estimate, the process noise variance, the measurement noise variance, and the prediction error threshold (42); Step 5, using the spatial resolution set in step 3, performing an inverse Fourier transform on the complex amplitude of the backscattered Rayleigh scattering at a certain spatial position (43) in the distance domain in step 2, and obtaining the Rayleigh scattering spectrum at a certain spatial position (43) after the strain or temperature acts on the sensing optical fiber as the measurement spectrum (41); Step 6: perform cross-correlation calculation on the measured spectrum (41) at a certain spatial position (43) and the local reference spectrum (12) corresponding to the spatial position, record the cross-correlation offset (45), and calculate the ratio of the amplitude of the highest peak (31) to the amplitude of the second highest peak (32) in the cross-correlation calculation result as the signal-to-noise ratio (44); Step 7: Substitute the cross-correlation offset (45) as the measured value (49) into the Kalman filter equation to calculate the estimated offset (46) of the measured spectrum (41) at a certain spatial position (43) relative to the reference spectrum (11); Step 8: Calculate the absolute value of the difference between the offset estimate (46) in step 7 and the cross-correlation offset (45) as the prediction error (47); if the signal-to-noise ratio (34) is greater than the signal-to-noise ratio threshold (48) or the prediction error (47) is less than the prediction error threshold (42), execute steps 14 to 15; if the signal-to-noise ratio (44) is less than the signal-to-noise ratio threshold (58) and the prediction error (47) is greater than the prediction error threshold (42), execute steps 9 to 15; Step 9, enter the neighborhood search. After the temperature or strain acts on the sensing optical fiber, the current spatial position (21x) is offset by equal data length in the distance domain relative to the starting position (21) by +1, -1, +2, -2, ..., +N, -N, and the backscattered complex amplitude of the current spatial position (21x) after the offset is obtained. The integer x represents the offset. The backscattered complex amplitude of the current spatial position (21x) is inversely transformed by Fourier to obtain the measured spectrum (41x) of the current spatial position (21x). Step 10, perform cross-correlation calculation on the measured spectrum (41x) obtained in step 9 and the local reference spectrum (12), record the cross-correlation offset (45x), and calculate the ratio of the amplitude of the highest peak (31x) to the amplitude of the second highest peak (32x) in the cross-correlation calculation result as the signal-to-noise ratio (44x); Step 11, substitute the cross-correlation offset (45x) as the measured value (49x) into the Kalman filter equation to calculate the estimated offset value (46x) of the measured spectrum (41x) at the current spatial position (21x) relative to the reference spectrum (11); Step 12: Calculate the absolute value of the difference between the offset estimate (46x) in step 11 and the cross-correlation offset (45x) as the prediction error (47x). If the prediction error (47x) is less than the prediction error threshold (42), then record the cross-correlation offset (45x) of the measured spectrum (41x) at the current spatial position (21x) relative to the local reference spectrum (12) and the offset x of the current spatial position (21x) relative to the starting position (21) in the distance domain, exit the neighborhood search, and execute steps 13 to 15. If the signal-to-noise ratio (44x) is greater than the signal-to-noise ratio threshold (48) and the prediction error (47x) is greater than the prediction error threshold (42), then record the current spatial position (21x) when the signal-to-noise ratio (44x) is maximum. x) the cross-correlation offset (45x) of the measured spectrum (41x) corresponding to the local reference spectrum (12) and the offset x of the current spatial position (21x) relative to the starting position (21) in the distance domain; if the signal-to-noise ratio (44x) is less than the signal-to-noise ratio threshold (48) and the prediction error (47x) is greater than the prediction error threshold (42), then the cross-correlation offset (45x) of the measured spectrum (41x) corresponding to the current spatial position (21x) and the local reference spectrum (12) and the offset x of the current spatial position (21x) relative to the starting position (21) in the distance domain when the prediction error (47x) is minimum are recorded, and steps 9 to 12 are repeated until the -N to +N search is completed, and then steps 13 to 15 are executed; Step 13, calculate the sum of the position offset cumulative amount (52) of the starting position (21) in the distance domain and the offset x of the current spatial position (21x) relative to the starting position (21) as the new position offset cumulative amount (52); Step 14: If the neighborhood search is performed, the cross-correlation offset (45x) is used as the spectral offset of the measured spectrum (41x) relative to the reference spectrum (11); if the neighborhood search is not performed, the cross-correlation offset (45) is used as the spectral offset of the measured spectrum (41) relative to the reference spectrum (11); and according to the relationship between the stress or temperature and the spectral offset, the stress or temperature at a certain spatial position (43) is converted; Step 15: According to the relationship between stress or temperature and the expansion coefficient of the sensing optical fiber, the sum of the position offset (51) generated by a certain spatial position (43) and the cumulative position offset (52) of the certain spatial position (43) in the distance domain is calculated as the cumulative position offset (52) of the next sensing unit in the distance domain, and the position of the next sensing unit of the current sensing unit in the distance domain is obtained; Step 16: Repeat steps 6 to 15 to obtain the stress or temperature at each spatial position in the sensing area.

2. The method according to claim 1, characterized in that: The Kalman filter prior state equation expression in step 7 is The prior estimation error covariance equation expression is The Kalman gain equation is The expression of the posterior state estimation equation is The a posteriori estimation error covariance equation is i-1 and i represent the previous sensor unit and the current sensor unit respectively. represents the prior state estimate, represents the posterior state estimate, represents the prior estimation error covariance, P i represents the posterior estimation error covariance, K i represents the Kalman gain, z i represents the measured value (59) in step 10, which is the cross-correlation offset (45) in this method, Q represents the process noise variance, and R represents the measurement noise variance.

3. The method according to claim 1, characterized in that: The error covariance of the initial estimate of the posterior estimation in step 4 should be 80 to 100 times the sensor accuracy, the initial posterior state estimate should be 0, the prediction error threshold (52) should be 3 to 5 times the sensor accuracy, and for temperature measurement, the process noise variance should be 1°C. 2 ~2.5℃ 2 , the measurement noise variance should be 6℃ 2 ~10℃ 2 , for strain measurement, the process noise variance should be 10με 2 ~25με 2 , the measurement noise variance should be 60με 2 ~100με 2 .

4. The method according to claim 1, characterized in that: In step 9, N should be 10% to 200% of the number of data points contained in the sensor unit length and rounded up to an integer.

Images