Distributed optical fiber sound field sensor pulse repetition frequency optimization setting method

Abstract

The invention discloses a distributed optical fiber sound field sensor pulse repetition frequency optimization setting method which can be used as a reference for PRF setting of a DAS system. Performing segmented acquisition and analysis on noise at different moments and different positions along the optical cable from time and space domains; the demodulation fidelity performance of the system under different PRFs is analyzed in a down-sampling differential distribution statistics mode; establishing SNR degradation models under different downsampling rates; in data processing, anti-aliasing filtering and down-extraction are performed according to the upper limit of the frequency of event signals acquired by the DAS system, so that the data storage and calculation pressure of the system is reduced. According to the PRF optimization setting method provided by the invention, the balance of the signal fidelity acquisition capability, the data volume and the sensing distance of the DAS system can be realized, and the detection performance of the system can be fully exerted on the premise of not changing the hardware configuration of the system in the engineering application oriented to different scenes and working conditions.

Description

Technical Field

[0001] This invention relates to the field of distributed optical fiber sensing technology, and in particular to a method for optimizing the pulse repetition frequency of a distributed optical fiber acoustic field sensor. Background Technology

[0002] Distributed optical fiber acoustic sensing (DAS) enables real-time monitoring of the environment along the sensing optical cable by monitoring changes in parameters such as the phase and frequency of backscattered light caused by the external environment. It has advantages such as distributed sensing, high sensitivity, and resistance to electromagnetic interference. It performs outstandingly in continuous monitoring over large spaces and is often used in fields such as road traffic monitoring, pipeline safety monitoring, and perimeter security.

[0003] However, in DAS, the setting of the pulse repetition frequency (PRF) is constrained by the interplay between the system's sensing distance and amplitude-frequency response range. On one hand, to avoid aliasing of backscattered light from adjacent pulses, which would significantly reduce the demodulation accuracy of the phase signal, the upper limit of the system's PRF is limited by the sensing distance. On the other hand, the PRF directly determines the system's sampling rate, and due to limitations in the demodulation algorithm, the PRF also affects the system's response range to signal change rates. Therefore, a higher PRF provides a wider amplitude-frequency response range. Furthermore, when the sensing distance and amplitude-frequency response meet engineering requirements, the PRF should be reduced to decrease the data volume while maintaining signal fidelity, thereby reducing data storage and computational pressure. In practical engineering applications, complex factors such as differences in event amplitude and frequency, time-varying noise characteristics, and the distribution of noise sources relative to the sensing optical cable place higher demands on the PRF setting during DAS system signal acquisition.

[0004] In terms of optimizing parameters such as PRF in distributed optical fiber sound field sensing systems, some research has been conducted in the prior art, such as: (1) by numerical modeling DAS, it was found that the accuracy and precision of the sensing system are affected by the frequency and amplitude of the disturbance, as well as the width and linewidth of the probe pulse; (2) by using time division multiplexing technology, the pulse repetition frequency of the system is increased by introducing multiple wavelength division multiplexers; (3) Chinese invention patent with announcement number CN111044171B discloses a method and device for adaptive light source parameters of a distributed optical fiber sensing system, wherein the PRF setting is calculated according to the length L of the sensing optical cable to obtain the corresponding highest non-aliasing optical pulse repetition frequency; (4) by constructing an adjustable pulse width DAS system, the distributed dynamic strain value under different spatial resolutions is simulated and calculated, and the relationship between the measured noise, the measured true value and the spatial resolution is comprehensively considered to obtain the optimal spatial resolution to improve the sensing performance of the system.

[0005] However, in terms of PRF settings for DAS systems, the above-mentioned schemes typically set the highest sampling rate that does not cause aliasing based on the sensing distance, which can lead to a significant data storage and computational burden. Summary of the Invention

[0006] To address the current challenges in PRF settings for DAS technology, which involve constraints from multiple factors such as system sensing distance, demodulation fidelity, and data volume, this invention proposes a widely applicable method for optimizing DAS technology PRF settings. This method comprehensively considers system detection performance and data volume to determine the optimal pulse repetition frequency, thereby achieving a balance between the system's signal fidelity extraction capability and storage / computation pressure.

[0007] This invention adopts the following technical solution: a method for optimizing the pulse repetition frequency setting of a distributed fiber optic acoustic field sensor, comprising the following steps:

[0008] S1. For events to be tested in unknown frequency bands, determine whether to use a fast setup method or a segmented optimization method to analyze the system PRF settings, depending on whether there is sufficient analysis time.

[0009] S2, Regarding the frequency band where the event to be tested is located [F] min ,F max ], with a high oversampling pulse repetition frequency f h The event is sampled, and the initial pulse repetition frequency is required to be the maximum frequency that does not cause aliasing.

[0010] If a segmented optimization method is adopted, the acquired data is first segmented according to the consistency of the phase signal change rate distribution and power spectral density in different time periods and different sections along the sensing optical cable. Then, demodulation fidelity performance analysis and SNR degradation analysis are performed segment by segment. The pulse repetition frequency analysis results of each segment obtained by the two analysis methods are combined to obtain the optimal pulse repetition frequency (PRF). op ;

[0011] If a rapid setup method is used, there is no need to segment the acquired signal, perform demodulation fidelity performance analysis and SNR degradation analysis on the acquired signal separately, and select the optimal pulse repetition frequency (PRF). op Set to the larger of the two analysis results. This can satisfy the demodulation fidelity performance and SNR threshold requirements;

[0012] S4. Use the preferred pulse repetition frequency PRF op After signal acquisition is completed, in the segmented optimization method, the upper limit f of the segmented measured event frequency obtained from the SNR degradation analysis is used. i,j,max The acquired signal is segmented for anti-aliasing filtering and down-dipping; in the fast setup method, the global event frequency upper limit F is used. maxThe demodulated phase signal is subjected to global anti-aliasing filtering and down-dipping to reduce the amount of system data while ensuring the detection capability.

[0013] Preferably, in step S2, the initially set pulse repetition frequency is calculated as follows:

[0014]

[0015] Where c is the speed of light in a vacuum, n is the refractive index of the core of the sensing fiber, and L is the length of the sensing cable.

[0016] Preferably, in step S3, the collected data is segmented according to the consistency and differences between the data difference distribution and the PSD results to ensure that the requirements for PRF settings are consistent for each segment. This specifically includes the following sub-steps:

[0017] S301. Based on known conditions, including but not limited to the characteristics of noise sources existing under different operating conditions and time periods along the optical cable, the sampling results are pre-segmented, and the time is divided into M segments. pre A characteristic time period T pre ={T 1pre ,T 2pre ,…,T Mpre Spatially, the sensing optical cable is divided into N sections. pre Section R pre ={R 1pre ,R 2pre ,…R Npre};

[0018] Let interval R ipre Characteristic time T jpre The signals collected inside are denoted as i∈{1,2,...,N pre},j∈{1,2,...,M pre};

[0019] S302. Calculate and compare the KL divergence matrix between the adjacent differential distribution results of each phase signal segment and the KL divergence matrix between PSDs to verify the rationality of the pre-segmentation.

[0020] For characteristic time T jpre Calculate the spatial segments i,j∈{1,2,...,N} pre Spatial KL divergence matrix of phase proximity difference distribution and the KL divergence matrix between PSDs

[0021]

[0022] in, Let i and j be the distributions of the temporal proximity differences, respectively. The PSD calculation results for spatial segments i and j are respectively, and D KL (P||Q) is the KL divergence calculation;

[0023] For space segment R ipre Calculate the time intervals i,j∈{1,2,...,M} pre Spatial KL divergence matrix of phase proximity difference distribution and the KL divergence matrix between PSDs

[0024]

[0025] in, Let i and j be the distributions of the time-domain proximity differences, respectively. The PSD calculation results are for time periods i and j, respectively.

[0026] S303. Based on the KL divergence matrix obtained in S32, perform a weighted summation to obtain the piecewise evaluation matrix:

[0027]

[0028] Wherein, λ1, λ2, λ3, and λ4 are the weight matrices of spatial segment differential distribution, spatial segment PSD, time segment differential distribution, and time segment PSD, respectively, which are set according to the degree of attention of the corresponding spatiotemporal segments;

[0029] Based on the set threshold, the pre-segmentation results are... Similar segments are merged, and segments with inconsistencies are split to obtain segments with different PRF requirements. i∈{1,2,...,N},j∈{1,2,...,M}, where N is the number of feature space segments and M is the number of feature time segments.

[0030] Preferably, in step S3, the demodulation fidelity performance analysis involves downsampling and temporal proximity differencing of the sampled data to different degrees, statistically analyzing the distribution of the differencing results, and determining the demodulation distortion ratio of the sampled data under different PRFs. Specifically, this includes the following sub-steps:

[0031] S311, Regarding the collected event signals For i∈{1,2,...,N},j∈{1,2,...,M}, the downsampling factor is W={W1,W2,...,W O Downsampling of} yields i∈{1,2,...,N},j∈{1,2,...,M},k∈{1,2,...,O}, where O is the number of downsampling factors;

[0032] S312, The l-th column in R i Timing signal of the l-th spatial point Represented as:

[0033]

[0034] in, W k As the downsampling factor, n0 = f h T j T represents the number of signal points when W=1 and no down-dipping is performed. j For characteristic time; x n for The l-th column of timing signals, s n For the collected event signals, η n For background noise, satisfy ζ represents the variance; n The noise introduced by the frequency folding of the spectrum into the signal band during the down-dipping process; A k Let φ be the amplitude of the k-th frequency component. k f is the initial phase of the k-th frequency component. s f is the sampling rate. k The frequency of the k-th frequency component;

[0035] right By performing proximity time difference, we obtain:

[0036]

[0037] in, Δζ n =ζ n+1 -ζ n ;

[0038] S313, Optical Cable R i The segment at characteristic time T j The sampling rate is f h / W k At that time, the demodulation distortion ratio is:

[0039]

[0040] Where length() is the number of points in the time or space segment within the parentheses, and I() is an indicator function that takes the value 1 only when the condition within the parentheses is met, and 0 otherwise;

[0041] S314. Traverse all characteristic time periods T = {T1, T2, ... T} M}, the interval R = {R1, R2, ... R N} and downsampling factor W = {W1, W2, ..., W O}, corresponding sampling rate Repeat steps S313 to S314 to obtain M×N×O demodulation distortion ratio data, which can be used as the demodulation distortion ratio of each data segment at different sampling rates.

[0042] S315. Based on the preset demodulation fidelity threshold P th Determine each interval R i T at different time periods j The lowest selectable PRF value i∈{1,2,...,N}, j∈{1,2,...,M}.

[0043] Preferably, in step S3, the SNR degradation analysis, based on the frequency band range and PSD of the event under test in the sampled data, calculates the noise power folded into the frequency band of the event under test at different sampling rates to obtain the SNR degradation at different sampling rates, specifically including the following sub-steps:

[0044] S321. Obtain the collected event signals. Power spectral density of M: For i∈{1,2,...,N},j∈{1,2,...,M}, the PSD of each position within each segment is approximately the same;

[0045] S322, Based on different down-conversion coefficients α and initial sampling rate f h The maximum aliasing order k under different frequency reduction coefficients α is obtained. max :

[0046]

[0047] S323. Based on the aliasing order and the original power spectral density, calculate the noise power P to be aliased. alised :

[0048]

[0049] S324. Calculate the signal-to-noise ratio degradation value under the current downsampling coefficient:

[0050]

[0051] S325. Repeat steps S323 to S325 to obtain the signal-to-noise ratio degradation curves of each of the N intervals at different time periods.

[0052] S326. Based on the set signal-to-noise ratio degradation threshold SNR th Determine the lowest PRF value that can be selected for each interval at different time periods. i∈{1,2,...,N}, j∈{1,2,...,M}.

[0053] Preferably, in step S3, the system demodulation fidelity threshold P is determined. th and SNR degradation threshold SNR th Set the system PRF, based on the calculated values ​​in the piecewise optimization method. and Obtain the preferred pulse repetition frequency (PRF) op :

[0054]

[0055] Furthermore, due to the influence of noise PSD and the event spectrum range, it is possible that in certain frequency bands [f abmin ,f abmax An abnormal SNR decrease occurs if the preferred pulse repetition frequency is in an abnormal frequency band: PRF op ∈[f abmin ,f abmax ], then the PRF op Moved up to outside the abnormal frequency band, causing PRF op =f abmax f abmin f abmax These are the lower and upper limits of the abnormal frequency band, respectively.

[0056] Preferably, in step S4, a preferred pulse repetition frequency PRF is used. op After signal acquisition is completed, the acquired signal is downsampled in segments, according to different intervals R. i With characteristic time period T j The collected signals Upper limit of effective signal frequency f i,j,max Perform anti-aliasing filtering and segmented downsampling with a downsampling factor of W(i,j):

[0057]

[0058] Where β is the frequency multiplication factor, and when β = 2, it can be guaranteed that the Nyquist law is satisfied after downsampling. In engineering, in order to ensure the signal demodulation effect, β is generally taken as 4 to 5.

[0059] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0060] 1. This invention provides a method for optimizing the PRF setting of a DAS system. Without changing the current hardware configuration of the system, this method can be used as a guide for setting the PRF in practical engineering applications. It also performs segmented anti-aliasing filtering and downsampling, effectively improving the signal fidelity acquisition capability of the DAS system, reducing the data storage and computing pressure, and extending the sensing distance.

[0061] 2. By using the signal demodulation fidelity performance analysis method of this invention, the neighbor difference distribution under different downsampling rates is statistically analyzed. The demodulation fidelity performance of the system can be analyzed under different PRFs with only one or a few probes, providing a basis for ensuring the demodulation fidelity performance of the system during the PRF optimization process.

[0062] 3. Using the SNR degradation analysis method of this invention, the effective signal frequency band folded down to different PRFs is calculated [f]. min ,f max The noise power of the system can be used to predict the SNR of the system under different PRFs, providing a basis for ensuring the system SNR during the PRF optimization process.

[0063] 4. After monitoring the preferred pulse repetition frequency obtained by the method of the present invention, the effective signal frequency band between segments [f] is then analyzed. min ,f max The differences can be segmented and filtered and downsampled to different degrees, thereby reducing the pressure on system data storage and computation. Attached Figure Description

[0064] Figure 1 This is an overall flowchart of the method of the present invention;

[0065] Figure 2 This is a waterfall chart of data from an embodiment of the present invention;

[0066] Figure 3 This is a schematic diagram of the neighboring differential distribution of phase signals in different segments and the KL divergence matrix between PSDs in an embodiment of the present invention;

[0067] Figure 4 This is a schematic diagram illustrating the demodulation distortion ratio of different segments under different PRFs in an embodiment of the present invention;

[0068] Figure 5 This is a schematic diagram illustrating the signal-to-noise ratio degradation of different segments under different PRFs in an embodiment of the present invention. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the application will be further described in detail below with reference to the accompanying drawings. The described embodiments are only a part of the embodiments involved in this invention. All non-innovative embodiments based on these embodiments by other researchers in the art are within the protection scope of this invention. Furthermore, the step numbers in the embodiments of this invention are only set for ease of explanation and do not limit the order of the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0070] In one embodiment of the present invention, a method for optimizing the PRF setting of a distributed fiber optic acoustic field sensing system is described in the following overall process: Figure 1 As shown, it includes two parts: a rapid setup method and a segmented optimization method. The segmented optimization method further includes two key steps: signal demodulation fidelity performance analysis and SNR degradation analysis.

[0071] For a test event in an unknown frequency band, firstly, a high oversampling PRF is set to sample it. After performing time-frequency analysis on the sampling results, the frequency band [F] of the global test event is determined. min ,F max ].

[0072] In the segmented optimization method, segmentation is required based on the inter-segment differences in the acquired data space, the phase change rate caused by the time-varying characteristics of the signal and noise, and the differences in noise power spectral density (PSD). Demodulation fidelity performance analysis and SNR degradation analysis are then performed on the segmented signals. Based on the segmentation analysis results, the optimal PRF (Programmable Rendering Function) is determined for each segment. op .

[0073] After using the PRF for signal acquisition, further analysis can be performed based on the upper limit f of the frequency of each segment of the event to be measured. max By performing segmented downsampling, a balance can be achieved between high signal-to-noise ratio acquisition of event signals and the pressure on sensing distance, data storage, and analysis.

[0074] In the PRF quick setup method, the acquired signal is not segmented for analysis. Instead, a better PRF is obtained based on the overall demodulation fidelity performance analysis and SNR degradation analysis results. Then, the upper frequency limit F of the global event under test is used. max The demodulated phase signal is subjected to global anti-aliasing filtering and de-sampling.

[0075] Specifically, in the process of determining the event frequency band, the event is first sampled using a high oversampled PRF. The system PRF can be set to the maximum frequency at which aliasing does not occur.

[0076]

[0077] Where c is the speed of light in a vacuum, n is the refractive index of the core of the sensing fiber, and L is the length of the sensing cable.

[0078] Because noise in the actual working environment may have time-varying characteristics and the characteristics of the sensing optical cable deployment route (including deployment method, geological conditions, distribution of nearby interference sources, etc.) are different, the time-domain phase change rate and noise PSD of the signals sampled in different areas along the sensing optical cable will vary at different times. Therefore, it is required to set the pulse repetition frequency to f. h The sampling process needs to cover multiple typical characteristic time periods, and the sampling results are pre-segmented based on known conditions (such as different operating conditions along the optical cable and the characteristics of noise sources in different time periods). The time period is divided into M segments. pre A characteristic time period T pre ={T 1pre ,T 2pre ,…,T Mpre Spatially, the sensing optical cable is divided into N sections. pre Section R pre ={R 1pre ,R 2pre ,…R Npre}, let the interval R ipre Characteristic time T jpre The signals collected inside are denoted as i∈{1,2,...,N pre},j∈{1,2,...,M pre}

[0079] To verify the rationality of the above pre-segmentation, it is necessary to compare the intra-segment consistency and inter-segment differences of the pre-segmentation results. Specifically, this is done by calculating and comparing the KL divergence matrices between adjacent differential distributions of the phase signal and between PSDs. The former affects demodulation fidelity performance, while the latter affects the SNR degradation index. During the calculation process, it is required to control a single variable: the characteristic time remains constant when verifying the rationality of the spatial segmentation, and the selected spatial segment remains constant when verifying the rationality of the characteristic time segmentation. The specific calculation process is as follows:

[0080] For characteristic time T jpre Calculate the spatial segments i,j∈{1,2,...,N} pre The spatial KL divergence matrix of the phase proximity difference distribution and the KL divergence matrix of the PSD distribution:

[0081]

[0082] in, Let i and j be the distributions of the temporal proximity differences, respectively. The PSD calculation results for spatial segments i and j are respectively, and D KL(P||Q) is the KL divergence calculation.

[0083] Preferably, in this embodiment, the KL divergence is calculated as follows:

[0084]

[0085] in, (k = 0, 1, ..., N-1), w(n) is a window function, F s Where is the sampling rate, N is the number of points in the discrete Fourier transform, and P(x) and Q(x) are the distribution functions of the two distributions used to calculate the KL divergence.

[0086] For space segment R ipre Calculate the time intervals i,j∈{1,2,...,M} pre The spatial KL divergence matrix of the phase proximity difference distribution and the KL divergence matrix between PSDs:

[0087]

[0088] in, Let i and j be the distributions of the time-domain proximity differences, respectively. The results are the PSD calculations for time periods i and j, respectively.

[0089] In this embodiment, the sampling data waterfall plot is as follows: Figure 2 As shown, after the above analysis and calculation, the KL divergence matrix of the difference distribution and PSD results for each segment is obtained, as follows: Figure 3 As shown in the example, the 11 points (front, middle, and rear) correspond to different working conditions. Based on the aforementioned pre-segmentation principle, they are pre-divided into 3 segments according to the different working conditions. The rationality of each segment is checked using the KL threshold. Similar segments in the pre-segmentation results are merged, and segments with poor intra-segment consistency are split. The pre-segmentation effect is found to be good, meeting the requirements of intra-segment consistency and inter-segment differences. Thus, the segmentation is obtained based on the phase signal proximity differential distribution and the phase signal PSD result. i∈{1,2,...,N},j∈{1,2,...,M} ensures that the PRF requirements of each segment tend to be consistent in the subsequent analysis process.

[0090] Furthermore, regarding the pulse repetition frequency f h The data collected below The demodulation fidelity performance analysis is performed as follows:

[0091] S311: For the collected event signals For i∈{1,2,...,N},j∈{1,2,...,M}, the downsampling factor is W={W1,W2,...,W O Downsampling of} i∈{1,2,...,N},j∈{1,2,...,M},k∈{1,2,...,O};

[0092] S312: The l-th column in R i The temporal signal of the l-th spatial point of segment can be represented as:

[0093]

[0094] in, W k As the downsampling factor, n0 = f h T j T represents the number of signal points when W=1 and no down-dipping is performed. j For characteristic time; x n for The l-th column of timing signals, s n For the collected event signals, η n For background noise, satisfy ζ represents the variance; n The noise introduced by the frequency folding of the spectrum into the signal band during the down-dipping process; A k Let φ be the amplitude of the k-th frequency component. k f is the initial phase of the k-th frequency component. s f is the sampling rate. k The frequency of the k-th frequency component;

[0095] right By performing proximity time difference, we obtain:

[0096]

[0097] in, Δζ n =ζ n+1 -ζ n ;

[0098] S313: Optical Cable R i The segment at characteristic time T j The sampling rate is f h / W k At that time, the demodulation distortion ratio is:

[0099]

[0100] Where length() is the number of points in the time or space segment within the parentheses, and I() is an indicator function that takes the value 1 only when the condition within the parentheses is met, and 0 otherwise;

[0101] S314: Traverse all characteristic time periods T = {T1, T2, ... T} M}, the interval R = {R1, R2, ... R N} and downsampling factor W = {W1, W2, ..., W O}(corresponding sampling rate) Repeat the calculation steps described in S313 and S314 to obtain M×N×O demodulation distortion ratio data, that is, the demodulation distortion ratio of each data segment at different sampling rates;

[0102] S315: Based on the preset demodulation fidelity threshold P th Each interval R can be determined. i T at different time periods j The lowest selectable PRF value i∈{1,2,...,N}, j∈{1,2,...,M}.

[0103] The data demodulation fidelity performance analysis results of this embodiment are as follows: Figure 4 As shown, the overall trend indicates that the demodulation failure probability gradually increases as the sampling rate decreases. However, for the first segment of data in the example, due to the small signal amplitude, even with a reduced sampling rate, it is impossible for the differential result to exceed [-π, π]. Therefore, it is impossible for the signal distortion to be caused by dewinding failure due to a decrease in the sampling rate. The differences between segments are mainly affected by the rate of change of the signal amplitude in each segment. Based on this demodulation fidelity performance analysis chart, the minimum PRF required to meet different demodulation fidelity performance requirements for each segment can be analyzed.

[0104] For pulse repetition frequency f h The data collected below The specific process for performing SNR degradation analysis is as follows:

[0105] S321: Obtain the sampled event signal Power spectral density of M: For i∈{1,2,...,N},j∈{1,2,...,M}, the PSD of each position within each segment is approximately the same;

[0106] S322: Based on different down-conversion coefficients α and initial sampling rate f h The maximum aliasing order k under different frequency reduction coefficients α is obtained. max :

[0107]

[0108] S323: Calculate the noise power to be aliased based on the aliasing order and the original power spectral density.

[0109]

[0110] S324: Calculate the signal-to-noise ratio degradation value under the current downsampling factor:

[0111]

[0112] S325: Repeat steps S323-S325 above to obtain the signal-to-noise ratio degradation curves of each of the N intervals at different time periods.

[0113] S326: Based on the set signal-to-noise ratio degradation threshold (SNR) th It can determine the lowest PRF value that can be selected for each interval at different time periods. i∈{1,2,...,N}, j∈{1,2,...,M}.

[0114] It should be noted that as PRF decreases, SNR does not necessarily decrease monotonically. It may vary in certain frequency bands due to factors such as noise PSD and the event's spectral range. abmin ,f abmax An abnormal decrease in SNR occurred.

[0115] The SNR degradation analysis results of this embodiment are as follows: Figure 5 As shown, the overall trend shows that the SNR gradually decreases as the sampling rate decreases in each segment, but there may be abnormal increases in some local areas. Due to differences in the distribution of noise sources relative to the sensing optical cable and the coupling effect of the sensing optical cable, the SNR degradation curves of each segment differ. The minimum PRF required to meet the SNR degradation requirements of each segment can be analyzed based on the SNR degradation curves.

[0116] The demodulation fidelity performance and SNR degradation analysis results are obtained as described above. and Subsequently, in order to ensure that the demodulation fidelity performance and SNR degradation are better than the set threshold P, th and SNR th The following formula is used for optimization:

[0117]

[0118] If the preferred pulse repetition frequency is in the abnormal frequency band, i.e., PRF op ∈[f abmin ,f abmax If it is moved up to outside the abnormal frequency band, then the PRF will be moved up. op =f abmax .

[0119] Using the optimized pulse repetition frequency PRF opAfter data acquisition is complete, segmented downsampling can further reduce the pressure on data storage and computation. The significance of segmented downsampling lies in the fact that the demodulation fidelity performance analysis mentioned above focuses on whether the phase change rate Δφ exceeds [-π,π], and its formula is shown below (ignoring noise terms):

[0120]

[0121] Among them, f s f is the sampling rate. k Let A be the frequency of the k-th frequency component. k Let be the amplitude of the k-th frequency component. It can be observed that the maximum phase change rate Δφ occurs under the conditions of low frequency with high amplitude and high frequency with low amplitude. max And the phase difference results may be distributed consistently. However, according to the Nyquist sampling theorem, a greater degree of down-dipping can be achieved in the case of low frequency and high amplitude.

[0122] For the preferred pulse repetition frequency PRF op The data collected below It can be determined based on the PSD results during the SNR degradation analysis. Upper limit of event frequency f i,j,max Anti-aliasing filtering and downsampling are performed. The maximum downsampling factor W(i,j) is calculated as follows:

[0123]

[0124] Where β is the harmonic coefficient.

[0125] As a preferred embodiment, β = 2 is chosen to ensure that the Nyquist theorem is satisfied after downsampling.

[0126] In summary, this invention proposes a widely applicable method for optimizing the PRF setting of DAS technology. Based on the phase change rate distribution and power spectral density (PSD) of signals sampled at different times and locations along the sensing optical cable, the signal is segmented to ensure that the PRF setting requirements for each segment are similar. Regarding demodulation fidelity performance analysis, the demodulation success rate under different PRFs is analyzed by performing temporal proximity differential analysis and statistically analyzing the differential distribution. Regarding signal-to-noise ratio (SNR) degradation analysis, the positive and negative folding to the effective signal frequency band under different PRFs is analyzed by calculating the PSD of the noisy signal. min ,f max The noise power of the signal is calculated to obtain the SNR degradation value under different PRFs. By combining the demodulation fidelity performance of each segment with the SNR degradation analysis results, the optimal PRF is selected. The signal acquired using the optimal PRF is then subjected to anti-aliasing filtering and segmented down-sampling to further reduce the amount of data, reduce the pressure on system data storage and computation, and achieve a balance between the system's signal fidelity extraction capability and storage and computation pressure.

[0127] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor, characterized in that, Includes the following steps: S1. For events to be tested in unknown frequency bands, determine whether to use a rapid setting method or a segmented optimization method to analyze the system pulse repetition frequency setting, depending on whether there is sufficient analysis time. S2, Regarding the frequency band where the event to be tested is located [F] min ,F max The event is sampled at a pulse repetition frequency higher than the oversampling frequency, requiring the initial pulse repetition frequency f to be set. h The maximum frequency at which aliasing does not occur; S3. If a segmented optimization method is adopted, the acquired data is first segmented according to the consistency of the phase signal change rate distribution and power spectral density in different time periods and different sections along the sensing optical cable. Then, demodulation fidelity performance analysis and SNR degradation analysis are performed segment by segment. The pulse repetition frequency analysis results of each segment obtained by the two analysis methods are combined to obtain the optimal pulse repetition frequency (PRF). op ; If a rapid setup method is used, there is no need to segment the acquired signal, perform demodulation fidelity performance analysis and SNR degradation analysis on the acquired signal separately, and select the optimal pulse repetition frequency (PRF). op Set to the larger of the two analysis results; S4. Use the preferred pulse repetition frequency PRF op After signal acquisition is completed, in the segmented optimization method, the upper limit f of the segmented measured event frequency obtained from the SNR degradation analysis is used. i,j,max The acquired signal is segmented for anti-aliasing filtering and down-dipping; in the fast setup method, the global event frequency upper limit F is used. max The demodulated phase signal is subjected to global anti-aliasing filtering and down-dipping to reduce the amount of system data while ensuring the detection capability.

2. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 1, characterized in that, In step S2, the initial pulse repetition frequency f is set. h The calculation is as follows: Where c is the speed of light in a vacuum, n is the refractive index of the core of the sensing fiber, and L is the length of the sensing cable.

3. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 1, characterized in that, In step S3, the collected data is segmented based on the consistency and differences between the data difference distribution and the PSD results to ensure that the pulse repetition frequency setting requirements are consistent for each segment. This specifically includes the following sub-steps: S301. Based on known conditions, including but not limited to the characteristics of noise sources existing under different operating conditions and time periods along the optical cable, the sampling results are pre-segmented, and the time is divided into M segments. pre A characteristic time period T pre ={T 1pre ,T 2pre ,…,T Mpre Spatially, the sensing optical cable is divided into N sections. pre Section R pre ={R 1pre ,R 2pre ,…R Npre }; Let interval R ipre Characteristic time T jpre The signals collected inside are denoted as i∈{1,2,...,N pre },j∈{1,2,...,M pre }; S302. Calculate and compare the KL divergence matrix between the adjacent differential distribution results of each phase signal segment and the KL divergence matrix between PSDs to verify the rationality of the pre-segmentation. For characteristic time T jpre Calculate the spatial segments i,j∈{1,2,...,N} pre Spatial KL divergence matrix of phase proximity difference distribution and the KL divergence matrix between PSDs in, Let i and j be the distributions of the temporal proximity differences, respectively. The PSD calculation results for spatial segments i and j are respectively, and D KL (P||Q) is the KL divergence calculation; For space segment R ipre Calculate the time intervals i,j∈{1,2,...,M} pre Spatial KL divergence matrix of phase proximity difference distribution and the KL divergence matrix between PSDs in, Let i and j be the distributions of the time-domain proximity differences, respectively. The PSD calculation results are for time periods i and j, respectively. S303. Based on the KL divergence matrix obtained in S302, perform a weighted summation to obtain the piecewise evaluation matrix: Wherein, λ1, λ2, λ3, and λ4 are the weight matrices of spatial segment differential distribution, spatial segment PSD, time segment differential distribution, and time segment PSD, respectively, which are set according to the degree of attention of the corresponding spatiotemporal segments; Based on the set threshold, the pre-segmentation results are... Similar segments are merged, and segments that are inconsistent within a segment are split to obtain segments with different pulse repetition frequency requirements. i∈{1,2,...,N},j∈{1,2,...,M}, where N is the number of feature space segments and M is the number of feature time segments.

4. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 3, characterized in that, In step S302, the KL divergence is calculated using the following formula: in, (k = 0, 1, ..., N-1), w(n) is a window function, F s Where is the sampling rate, N is the number of points in the discrete Fourier transform, and P(x) and Q(x) are the distribution functions of the two distributions used to calculate the KL divergence.

5. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 3, characterized in that, In step S3, the demodulation fidelity performance analysis involves downsampling and temporal proximity differencing of the sampled data to different degrees, statistically analyzing the distribution of the differencing results, and determining the demodulation distortion ratio of the sampled data at different pulse repetition frequencies. Specifically, this includes the following sub-steps: S311, Regarding the collected event signals For i∈{1,2,...,N},j∈{1,2,...,M}, the downsampling factor is W={W1,W2,...,W O Downsampling of} yields i∈{1,2,...,N},j∈{1,2,...,M},k∈{1,2,...,O}, where O is the number of downsampling factors; S312, The l-th column in R i Timing signal of the l-th spatial point Represented as: in, W k As the downsampling factor, n0 = f h T j T represents the number of signal points when W=1 and no down-dipping is performed. j For characteristic time; x n for The l-th column of timing signals, s n For the collected event signals, η n For background noise, satisfy σ η 2 ζ represents the variance; n The noise introduced by the frequency folding of the spectrum into the signal band during the down-dipping process; A k Let φ be the amplitude of the k-th frequency component. k f is the initial phase of the k-th frequency component. s f is the sampling rate. k The frequency of the k-th frequency component; right By performing proximity time difference, we obtain: Among them, Dz n =ζ n+1 -g n ; S313, Optical Cable R i The segment at characteristic time T j The sampling rate is f h / W k At that time, the demodulation distortion ratio is: Where length() is the number of points in the time or space segment within the parentheses, and I() is an indicator function that takes the value 1 only when the condition within the parentheses is met, and 0 otherwise; S314. Traverse all characteristic time periods T = {T1, T2, ... T} M }, the interval R = {R1, R2, ... R N } and downsampling factor W = {W1, W2, ..., W O }, corresponding sampling rate Repeat steps S313 to S314 to obtain M×N×O demodulation distortion ratio data, which can be used as the demodulation distortion ratio of each data segment at different sampling rates. S315. Based on the preset demodulation fidelity threshold P th Determine each interval R i T at different time periods j Selectable minimum pulse repetition frequency value i∈{1,2,...,N}, j∈{1,2,...,M}.

6. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 5, characterized in that, In step S311, the collected event signals Represented as: Where length(·) is the number of points in the time or space segment within the parentheses; For the system acquisition interval R i With characteristic time period T j Signal Downsampling factor is After downsampling, we get Represented as: Where row is the number of rows. correspond The data collected by the system at that time.

7. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 3, characterized in that, In step S3, the SNR degradation analysis, based on the frequency band range and PSD of the event under test in the sampled data, calculates the noise power folded into the frequency band of the event under test at different sampling rates, and obtains the SNR degradation at different sampling rates. Specifically, this includes the following sub-steps: S321. Obtain the collected event signals. Power spectral density of M: For i∈{1,2,...,N},j∈{1,2,...,M}, the PSD of each position within each segment is approximately the same; S322, Based on different down-conversion coefficients α and initial sampling rate f h The maximum aliasing order k under different frequency reduction coefficients α is obtained. max : S323. Based on the aliasing order and the original power spectral density, calculate the noise power P to be aliased. alised : Among them, S n (f) represents the original power spectral density; S324. Calculate the signal-to-noise ratio degradation value under the current downsampling coefficient: Among them, P n,original For the original noise power, SNR alised SNR original These are the signal-to-noise ratio (SNR) after introducing folding noise and the original SNR, respectively. S325. Repeat steps S323 to S325 to obtain the signal-to-noise ratio degradation curves of each of the N intervals at different time periods. S326. Based on the set signal-to-noise ratio degradation threshold SNR th Determine the lowest pulse repetition frequency value that can be selected for each interval at different time periods.

8. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 7, characterized in that, In step S3, based on the system demodulation fidelity threshold P... th and SNR degradation threshold SNR th Set the system pulse repetition frequency as follows: The segmented optimization method is based on the calculated... and Obtain the preferred pulse repetition frequency (PRF) op : The rapid setup method is based on the demodulation fidelity performance analysis results. SNR degradation analysis results Obtain the preferred pulse repetition frequency (PRF) op :

9. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 8, characterized in that, In the segmented optimization method, due to the influence of noise PSD and the event spectrum range, it is possible that in certain frequency bands [f]... abmin ,f abmax An abnormal SNR decrease occurs if the preferred pulse repetition frequency is in an abnormal frequency band: PRF op ∈[f abmin ,f abmax ], then the PRF op Moved up to outside the abnormal frequency band, causing PRF op =f abmax f abmin f abmax These are the lower and upper limits of the abnormal frequency band, respectively.

10. The method for optimizing the pulse repetition frequency of a distributed fiber optic acoustic field sensor according to claim 8, characterized in that, In step S4, the preferred pulse repetition frequency PRF is used. op After signal acquisition is completed, the acquired signal is downsampled in segments, according to different intervals R. i With characteristic time period T j The collected signals Upper limit of effective signal frequency f i,j,max Perform anti-aliasing filtering and segmented downsampling with a downsampling factor of W(i,j): Where β is the harmonic coefficient, and when β = 2, it can be guaranteed that the Nyquist law is satisfied after downsampling.

Images