A signal processing method of a brillouin optical time domain analysis system and a terminal thereof

Abstract

The application discloses a signal processing method of a Brillouin optical time domain analysis system, and comprises the following steps: S1, collecting a Brillouin gain spectrum, performing two-dimensional variation modal decomposition and reconstruction; S2, performing spectrum type correction through time slot difference, and correcting the distorted Brillouin gain spectrum into a Lorentz spectrum type; and S3, extracting temperature information from the corrected Brillouin gain spectrum based on a data model. The three functions of noise reduction, spectrum type correction and temperature information extraction are realized, and the shortcomings of poor measurement accuracy and long processing time of a traditional fitting algorithm are overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to a Brillouin optical time domain analysis system signal processing method and terminal thereof. BACKGROUND

[0002] The distributed optical fiber sensing system based on Brillouin scattering has many advantages such as large sensing dynamic range, long monitoring distance and continuous distributed measurement, and is widely used in the health state diagnosis of fields such as bridges, oil pipelines and power transmission lines. The Brillouin optical time domain analysis system (BOTDA) realizes the monitoring of the optical fiber along the line according to the function relationship between the Brillouin frequency shift and the temperature / strain, and therefore it is very important to accurately obtain the Brillouin frequency shift. The long-distance Brillouin optical time domain sensing system is easily affected by non-local effects, and the Brillouin gain spectrum will be distorted. When the sensing signal is distorted and the signal-to-noise ratio is low, the conventional signal processing method has problems such as poor measurement accuracy, insufficient denoising performance, loss of spectral details, long processing time and the like. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a Brillouin optical time domain sensing system signal processing method and terminal thereof for the case of distorted sensing signal and low signal-to-noise ratio, and to improve the measurement accuracy of the temperature information along the optical fiber.

[0004] In order to solve the above technical problems, one technical solution adopted by the present application is:

[0005] A Brillouin optical time domain analysis system signal processing method, comprising the following steps:

[0006] S1, collecting the Brillouin gain spectrum, performing two-dimensional variational modal decomposition and reconstruction;

[0007] S2, performing spectral pattern correction through time slot difference to correct the distorted Brillouin gain spectrum into a Lorentz spectral pattern;

[0008] S3, extracting temperature information from the corrected Brillouin gain spectrum based on a data model.

[0009] In order to solve the above technical problems, another technical solution adopted by the present application is:

[0010] A Brillouin optical time domain analysis system signal processing terminal, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of the above Brillouin optical time domain analysis system signal processing method when executing the computer program.

[0011] The beneficial effects of the present application are: a signal processing method of a Brillouin optical time domain sensing system and a terminal thereof are provided, three functions of noise reduction, spectrum type correction and temperature information extraction are realized, the signal-to-noise ratio is improved while the edge characteristics of the Brillouin gain spectrum image signal are maintained; the distorted Brillouin gain spectrum is corrected by time slot difference spectrum type correction processing, and presents a good Lorentz shape, which helps to improve the temperature extraction accuracy; and the temperature information along the optical fiber is directly extracted based on the data model, which overcomes the shortcomings of poor measurement accuracy and long processing time of the traditional fitting algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A single-mode optical fiber stimulated Brillouin scattering schematic diagram of a signal processing method of a Brillouin optical time domain analysis system according to an embodiment of the present application;

[0013] Figure 2 A noise reduction flowchart of a signal processing method of a Brillouin optical time domain analysis system according to an embodiment of the present application;

[0014] Figure 3 An optimization flowchart of a signal processing method of a Brillouin optical time domain analysis system according to an embodiment of the present application;

[0015] Figure 4 A total flowchart of a signal processing method of a Brillouin optical time domain analysis system according to an embodiment of the present application;

[0016] Figure 5 An architecture diagram of a signal processing terminal of a Brillouin optical time domain analysis system according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] To explain the technical content, the achieved purposes and effects of the present application in detail, the following will be described in combination with the embodiments and the accompanying drawings.

[0018] Please refer to Figures 1 to 4 The embodiment of the present application provides a signal processing method of a Brillouin optical time domain analysis system, which comprises the following steps:

[0019] S1, Brillouin gain spectrum is collected, two-dimensional variational modal decomposition and reconstruction are performed;

[0020] S2, spectrum type correction is performed by time slot difference, and the distorted Brillouin gain spectrum is corrected to a Lorentz spectrum type;

[0021] S3, temperature information is extracted from the corrected Brillouin gain spectrum based on a data model.

[0022] From the above description, the beneficial effects of the present application are that: a signal processing method of Brillouin optical time domain sensor system is provided, three functions of noise reduction, spectrum correction and temperature information extraction are realized, the signal-to-noise ratio is improved while the edge characteristics of the Brillouin gain spectrum image signal are maintained; the distorted Brillouin gain spectrum presents a good Lorentz shape after the spectrum correction processing of time slot difference, which helps to improve the temperature extraction accuracy; and the temperature information along the optical fiber is directly extracted based on the data model, which overcomes the shortcomings of poor measurement accuracy and long processing time of the traditional fitting algorithm.

[0023] Further, the Brillouin gain spectrum collected in step S1 is specifically:

[0024] The BOTDA system obtains m groups of one-dimensional Brillouin gain spectrum signals along the optical fiber through frequency sweeping, the one-dimensional Brillouin gain spectrum signal is a two-dimensional matrix M of m columns and n rows, and the matrix M includes local Brillouin gain spectra at each position of the optical fiber:

[0025]

[0026] In the formula, f(v i ,z j )(1≤i≤m,1≤j≤n) represents the Brillouin gain spectrum at the i-th frequency sweeping frequency point and the j-th sampling point in the length range of the optical fiber.

[0027] From the above description, corresponding to different scanning frequencies, the BOTDA system can obtain one-dimensional Brillouin gain curves of each spatial position point along the optical fiber. The two-dimensional Brillouin gain spectrum signal obtained by frequency sweeping is decomposed into multiple sub-modes, and the center frequencies of each mode are different, that is, the collected Brillouin gain spectrum can be regarded as a two-dimensional signal distributed with frequency and distance.

[0028] Further, the two-dimensional variational mode decomposition and reconstruction in step S1 are specifically:

[0029] Let f(v i ,z j )(1≤i≤m,1≤j≤n) be f(x), and let its constraint variational model be:

[0030]

[0031]

[0032] In the formula, u k ={u1,u2,…,u k} is the K mode function of decomposition, w k ={w1,w2,…,w k} is the corresponding center frequency;

[0033] A quadratic penalty factor and a Lagrange multiplier λ are introduced, and the constrained variational model is optimized by using an alternating direction multiplier method to obtain an unconstrained problem, an extended Lagrange function:

[0034]

[0035] Alternately updating Until the following condition is satisfied The iteration stops and each IMF component is output:

[0036]

[0037]

[0038]

[0039] where τ is a residual, and represent the frequency domain properties of u k and w k , respectively;

[0040] Each IMF component obtained by the decomposition is combined to obtain a reconstructed signal.

[0041] As can be seen from the above description, the reconstructed signal removes the noise component of the original signal and has a good signal-to-noise ratio.

[0042] Further, step S1 further includes, for a signal with a low signal-to-noise ratio, performing a transform domain noise reduction processing, specifically:

[0043] S101, inputting a noisy two-dimensional Brillouin gain spectrum signal collected by a BOTDA system, and setting an initial value of a decomposition layer number as 2;

[0044] S102, setting a correlation coefficient screening threshold as 0.3, calculating a correlation coefficient of IMF1 of the modal decomposition under a current value and the original signal, and judging whether the correlation coefficient of IMF1 and the original signal is greater than the threshold, when greater than the threshold, K=K-1 is the best decomposition number, otherwise K=K+1 and the step S102 is continuously executed;

[0045] S103, performing a decomposition process by using the value determined in step S102 to obtain a modal signal and reconstructing the modal signal.

[0046] As can be seen from the above description, the low-frequency and high-frequency parts are separated by using the correlation coefficient, and the two-dimensional Brillouin gain spectrum signal after noise reduction can be obtained by using the low-frequency part for reconstruction.

[0047] Further, step S2 specifically includes:

[0048] Let the Brillouin gain spectrum after noise reduction be M':

[0049]

[0050] Select an arbitrary row from the time slot spectrum data which does not carry stimulated Brillouin scattering information, and record it as C:

[0051] C = [F(v1) F(v2) … F(v m )]

[0052] Subtract each row of matrix M' from C to obtain the corrected Brillouin gain spectrum data of Lorentz shape.

[0053] From the above description, for the distorted Brillouin gain spectrum caused by non-local effect, the time slot difference spectrum correction scheme is adopted to correct the distorted spectrum to Lorentz spectrum. Because the pump pulse repetition time of the system is greater than the time of the pulse traversing the fiber length and returning the scattered signal, there is a time slot during each sweep measurement, and the photodetector collects continuous light signals that have not met the pump pulse light, which does not carry SBS information. Subtracting the Brillouin gain spectrum of each position along the line from the continuous light spectrum without SBS can achieve the purpose of spectrum correction.

[0054] Further, the step S3 of directly extracting temperature information from the Brillouin gain spectrum based on the data model is specifically:

[0055] S301, a data model is established, and continuous learning and training are performed to obtain a mapping function relationship between Brillouin frequency shift and temperature;

[0056] S302, through simulation and laboratory calibration, L training samples (A i , T i ) are generated, i = 1, 2, …, L, wherein A i = [a i (v1), a i (v2), …, a i (v m )] is the Brillouin gain spectrum under different sweep frequencies of the i-th sample; T i is the corresponding output vector, i.e. the temperature value; the number of hidden layer nodes is N, and the output vector is as follows:

[0057]

[0058] In the formula, w j is the input weight between the input node and the hidden layer node, β j is the output weight between the hidden layer node and the output node, b j is the bias of the j-th hidden layer node, g(·) is an activation function, and β = [β1β2…β N ]T T = [T1 T2…T] L If ], then the output of the hidden layer node is:

[0059]

[0060] The output vector is expressed in matrix form as: Hβ = T;

[0061] S303. There are n sampling points along the optical fiber. The Brillouin gain spectrum at each sampling point of the optical fiber is selected and input into the trained model for testing to obtain the corresponding temperature, that is, the temperature information along the optical fiber.

[0062] As described above, during the initial measurement or laboratory calibration phase, by learning and training Brillouin spectrum data collected under different temperature conditions, a mapping function relationship between the Brillouin frequency shift spectrum and temperature can be established. Therefore, during actual monitoring, it is unnecessary to determine the Brillouin frequency shift; the temperature information along the optical fiber can be directly obtained through the trained function relationship. This improves measurement accuracy while effectively reducing data processing time.

[0063] Furthermore, step S3 also includes optimizing the input weights w and bias b: using the root mean square error of the prediction results of the training set as the fitness function, setting the objective of parameter optimization to minimize the root mean square error of the prediction results, and exiting the loop when the fitness value is minimized or the maximum number of iterations is met.

[0064] As described above, the prediction accuracy of the model can be improved by optimizing the input weights w and bias b.

[0065] Please refer to Figure 5 Another embodiment of the present invention provides a signal processing terminal for a Brillouin optical time-domain analysis system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps in the signal processing method of the Brillouin optical time-domain analysis system described above.

[0066] The signal processing method and terminal of the Brillouin optical time-domain sensing system described above can effectively improve the measurement accuracy of the Brillouin optical time-domain analysis system. The following detailed embodiments illustrate this method:

[0067] Example 1

[0068] Please refer to Figure 4 A signal processing method for a Brillouin optical time-domain analysis system includes the following steps:

[0069] S1. Acquire the Brillouin gain spectrum and perform two-dimensional variational mode decomposition and reconstruction;

[0070] In step S1, the Brillouin gain spectrum is collected, and the Brillouin gain spectrum is specifically as follows:

[0071] The BOTDA system obtains m groups of one-dimensional Brillouin gain spectrum signals along the fiber through frequency sweeping, and specifically, the BOTDA system can obtain one-dimensional Brillouin gain curves of each spatial position point along the fiber corresponding to different scanning frequencies:

[0072] The one-dimensional signals of the m groups of scanning frequency points are stored in columns to obtain a two-dimensional matrix M of m columns and n rows, and the matrix M contains the local Brillouin gain spectrum at each position of the fiber, so that:

[0073]

[0074] v1,v2,…,v m is the scanning frequency, z1,z2,…,z n is the fiber position, S1,S2,…,S m is the scanning frequency v1,v2,…,v m , and the m groups of one-dimensional signals are obtained.

[0075] The two-dimensional Brillouin gain spectrum signal obtained by frequency sweeping is decomposed into multiple sub-modes, and the center frequencies of the modes are different, that is, the collected Brillouin gain spectrum can be regarded as a two-dimensional signal distributed with frequency and distance.

[0076] The one-dimensional Brillouin gain spectrum signal is a two-dimensional matrix M of m columns and n rows, and the matrix M contains the local Brillouin gain spectrum at each position of the fiber:

[0077]

[0078] In the formula, f(v i ,z j )(1≤i≤m,1≤j≤n) represents the Brillouin gain spectrum at the jth sampling point in the fiber length range of the ith scanning frequency point.

[0079] In step S1, the two-dimensional variational mode decomposition and reconstruction are specifically as follows:

[0080] For simplicity of description, f(v i ,z j )(1≤i≤m,1≤j≤n) is denoted as f(x), and a constraint variational model thereof is:

[0081]

[0082]

[0083] In the formula, u k ={u1,u2,…,u k} is the K mode functions of decomposition, wk ={w1,w2,…,w k} are corresponding center frequencies;

[0084] Introducing a quadratic penalty factor and a Lagrange multiplier λ, and using the alternating direction multiplier method to optimize the constrained variational model, a non-constrained problem, an extended Lagrange function, is obtained:

[0085]

[0086] Alternately updating Until the condition is met The iteration stops and each IMF component is output:

[0087]

[0088]

[0089]

[0090] In the formula, τ is a residual, and respectively represent the frequency domain properties of u k and w k ;

[0091] Each IMF component obtained by decomposition is combined to obtain a reconstructed signal. The reconstructed signal removes the noise component of the original signal and has a good signal-to-noise ratio.

[0092] Wherein, please refer to Figure 2 , step S1 further includes, for a signal with a low signal-to-noise ratio, performing a transform domain noise reduction processing, specifically:

[0093] S101, inputting a noisy two-dimensional Brillouin gain spectrum signal collected by a BOTDA system, and setting an initial value of a decomposition layer number as 2;

[0094] S102, setting a correlation coefficient screening threshold value as 0.3, calculating a correlation coefficient of IMF1 of the modal decomposition under a current value and the original signal, judging whether the correlation coefficient of IMF1 and the original signal is greater than the threshold value, when greater than the threshold value, K=K-1 is the best decomposition number, otherwise K=K+1 and the step S102 is continuously executed;

[0095] S103, performing a decomposition process by using the value determined in step S102, obtaining a modal signal and reconstructing it.

[0096] Specifically, the sub-modal frequencies decrease in turn, so the center frequency of IMF1 is the highest, and when over-decomposition occurs, IMF1 corresponds to the decomposed pseudo-component. Among them, the effective information and the edge part are retained in the low-frequency mode, and the noise to be removed is concentrated in the high-frequency mode. The low-frequency and high-frequency parts are separated by using the correlation coefficient, and the two-dimensional Brillouin gain spectrum signal after noise reduction can be obtained by using the low-frequency part for reconstruction.

[0097] S2, spectrum type correction is performed through time slot difference, and the distorted Brillouin gain spectrum is corrected to a Lorentz spectrum type;

[0098] In the step S2, the following steps are specifically included:

[0099] Let the Brillouin gain spectrum after noise reduction be M':

[0100]

[0101] From the time slot spectrum data without carrying the stimulated Brillouin scattering information, an arbitrary row is selected and recorded as C:

[0102] C = [F(v1) F(v2) … F(v m )]

[0103] Each row of data in the matrix M' is subtracted from C to obtain the corrected Brillouin gain spectrum data in the Lorentz shape. The corrected Brillouin gain spectrum data I in the Lorentz shape is:

[0104]

[0105] In the formula, f n (v i ,z j ) = f'(v i ,z j )-F(v i ), i = 1, 2, …, m, j = 1, 2, …, n.

[0106] Specifically, please refer to Figure 1 In the BOTDA system, the pump pulse light and the continuous light propagate in opposite directions, and the stimulated Brillouin scattering effect (SBS) occurs when they meet in the optical fiber. Let the pump pulse light repetition interval be t, and the sensing optical fiber length be l. A pump pulse light traverses to the tail end of the optical fiber, continuously meets the continuous light along the way, and carries the SBS information back to the head end to be collected by the photodetector, and the time used is set as t1. The time interval of t-t1 remaining before the next pulse light is emitted is called a time slot. The photodetector collects the continuous light signal in the time slot without meeting the pump pulse light, which does not carry SBS information. The Brillouin gain spectrum of each position along the line of the optical fiber is subtracted from the continuous light spectrum without SBS to achieve the purpose of spectrum type correction.

[0107] The data collected by the BOTDA system includes two parts: the time t1 in the time slot corresponds to the sampling point z1, z2, …, z Figure 1 in the fiber length, and the time t2 in the time slot corresponds to the sampling point z1, z2, …, z n , the SBS effect occurs; during the time slot t-t1, the SBS effect does not occur. The data of each frequency point along the fiber length where the SBS effect occurs constitutes M; the data of each frequency point at any time during the time slot constitutes C.

[0108] The core idea is to use the time slot to subtract the spectral signals with and without the SBS effect. Because the pump pulse repetition time of the system is greater than the time for the pulse to traverse the fiber length and return the scattered signal, there is a time slot during each sweep measurement. The spectral signal collected by the photodetector in this time slot is continuous light that does not meet the pulse light, and the SBS effect does not occur, which is named as the time slot spectrum. The difference operation between the Brillouin gain spectrum of each position along the line and the time slot spectrum can achieve the purpose of spectral correction.

[0109] S3, extracting temperature information from the corrected Brillouin gain spectrum based on a data model.

[0110] In step S3, the temperature information is directly extracted from the Brillouin gain spectrum based on a data model, and the specific process is as follows:

[0111] S301, a data model is established, and continuous learning and training are performed to obtain a mapping function relationship between the Brillouin frequency shift and the temperature; that is, during actual monitoring, the Brillouin frequency shift does not need to be determined, and the temperature information of the fiber along the line can be directly obtained through the trained function relationship.

[0112] S302, through simulation and laboratory calibration, L training samples (A i , T i ) are generated, i = 1, 2, …, L, where A i = [a i (v1), a i (v2), …, a i (v m )] is the i-th sample, that is, the Brillouin gain spectrum under different sweeps; T i is the corresponding output vector, that is, the temperature value; the number of hidden layer nodes is N, and the output vector is as follows:

[0113]

[0114] In the formula, w j is the input weight between the input node and the hidden layer node, β j is the output weight between the hidden layer node and the output node, and b jbias of the jth hidden layer node, g(·) is an activation function, preferably, a Sigmoid function is selected as the activation function. Let β = [β1β2…β N ] T , T = [T1 T2…T L ], the output of the hidden layer node is:

[0115]

[0116] The output vector is expressed in the form of a matrix as Hβ = T.

[0117] S303, there are n sampling points along the optical fiber, the Brillouin gain spectrum at each sampling point of the optical fiber is selected and input to the trained model for testing to obtain the corresponding temperature, i.e. the temperature information of the optical fiber along the line.

[0118] Specifically, the present method directly extracts temperature information from the Brillouin gain spectrum, without determining the numerical value of the Brillouin center frequency, and can directly obtain the temperature information of the optical fiber along the line. The present method improves the measurement accuracy while effectively reducing the data processing time. The core idea is that, in the measurement starting or laboratory calibration stage, the Brillouin spectrum data collected under different temperature conditions are learned and trained to establish a mapping function relationship between the Brillouin frequency shift spectrum and the temperature. In the actual measurement process, the temperature information can be output based on the data model obtained by training.

[0119] In step S3, the input weight w and the bias b are also optimized: the root mean square error of the prediction results of the training set is taken as the fitness function, the target of parameter optimization is set to minimize the root mean square error of the prediction results, and the loop is exited when the fitness value is minimized or the maximum number of iterations is met. By optimizing the input weight w and the bias b, the prediction accuracy of the model is improved.

[0120] Specifically, please refer to Figure 3 The optimization steps are as follows:

[0121] Step 1: generate a training sample data set (A, T) by simulation experiment calibration, A is the Brillouin gain spectrum data as a feature vector; T is the prediction result.

[0122] Step 2: initialize the parameters, randomly generate the input weight w and the bias b, get the output weight H, and the prediction result vector is Hβ = T;

[0123] Step 3: calculate the fitness value and update the parameter value, the root mean square error is used as the fitness function in this paper;

[0124] Step 4: compare whether the current fitness value is better than the last result, if yes, update the parameter value, otherwise keep the parameter value unchanged;

[0125] Step 5: if the maximum number of iterations is reached, end, jump out of the loop to obtain the optimal input weight w and bias b; otherwise, return to step 3.

[0126] The optimal input weight w and bias b can be obtained through the above optimization process. Select a position point z of the optical fiber j , and the data after noise reduction and spectral shape correction processing is represented as [f"(v1,z j )f"(v2,z j )…f"(v m ,z j )], which is input into the well-trained model for testing, so that the corresponding temperature T j can be obtained, and other position points are the same, so that the temperature information along the optical fiber can be monitored.

[0127] Embodiment two

[0128] Please refer to Figure 5 , a signal processing terminal of a Brillouin optical time domain analysis system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements each step in embodiment one when executing the computer program.

[0129] In summary, the signal processing method and terminal of the Brillouin optical time domain analysis system provided by the application realize noise reduction, spectral shape correction and temperature information extraction by two-dimensional variational mode decomposition and reconstruction, time slot difference and the mapping function relationship between Brillouin frequency shift and temperature to directly obtain the temperature information along the optical fiber, thereby overcoming the shortcomings of poor measurement accuracy and long processing time of the traditional fitting algorithm.

[0130] It should be noted that, for the above-mentioned method embodiments, in order to facilitate description, they are all described as a series of action combinations, but those skilled in the art should know that the application is not limited by the described action sequence, because according to the application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily necessary for the application.

[0131] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0132] The above only describes the embodiments of the application, and does not limit the patent scope of the application, and any equivalent structure or equivalent process transformation according to the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the application.

Claims

1. A signal processing method for a Brillouin optical time-domain analysis system, characterized in that, Includes the following steps: S1. Acquire the Brillouin gain spectrum and perform two-dimensional variational mode decomposition and reconstruction; S2. After time-slot difference, spectral correction is performed to correct the distorted Brillouin gain spectrum to the Lorentz spectrum. S3. Extract temperature information from the corrected Brillouin gain spectrum based on the data model; The acquisition of the Brillouin gain spectrum in step S1 specifically involves: The BOTDA system obtains the fiber optic route information through frequency sweeping. m A one-dimensional Brillouin gain spectrum signal is formed, wherein the one-dimensional Brillouin gain spectrum signal is m List n A two-dimensional matrix of rows M The matrix M Includes the local Brillouin gain spectrum at various locations along the optical fiber: In the formula, f ( v i , z j ) indicates the first i The sweep frequency point, within the fiber length range of the first... j Brillouin gain spectrum at each sampling point, 1 ≤ i ≤ m , 1 ≤ j ≤ n ; Step S2 is as follows: Let the Brillouin gain spectrum after noise reduction be M’ : ； Randomly select one row from the time-slotted spectral data that does not carry stimulated Brillouin scattering information, and denote it as... C : matrix M’ Each row of data is related to C The difference is calculated to obtain the corrected Lorentz-shaped Brillouin gain spectrum data.

2. The signal processing method for a Brillouin optical time-domain analysis system according to claim 1, characterized in that, The two-dimensional variational mode decomposition and reconstruction described in step S1 specifically refers to: Will f ( v i , z j Set as f ( x Let its constrained variational model be: In the formula, u k ={ u 1, u 2, …, u k } represents the k modal functions of the decomposition. w k ={ w 1, w 2, , w k } represents the corresponding center frequency; By introducing a quadratic penalty factor and a Lagrange multiplier λ, and optimizing the constrained variational model using the alternating direction multiplier method, the unconstrained problem-extended Lagrange function is obtained: Alternate updates , , until satisfied When the iteration stops, each IMF component is output: ； ； ； In the formula, For residuals, and Represent u k and w k Frequency domain properties; The reconstructed signal is obtained by combining the individual IMF components obtained from the decomposition.

3. The signal processing method for a Brillouin optical time-domain analysis system according to claim 2, characterized in that, Step S1 also includes transform-domain noise reduction for signals with low signal-to-noise ratios, specifically: S101. Input the noisy two-dimensional Brillouin gain spectrum signal acquired by the BOTDA system, and set the initial value of the number of decomposition layers to 2. S102. Set the correlation coefficient filtering threshold to 0.3, calculate the correlation coefficient between the IMF1 of the modal decomposition and the original signal under the current value, and determine whether the correlation coefficient between the IMF1 and the original signal is greater than the threshold. If it is greater than the threshold, then k=k-1 is the optimal number of decompositions; otherwise, k=k+1 and continue to execute step S102. S103. Perform the decomposition process using the values ​​determined in step S102 to obtain each modal signal and reconstruct it.

4. The signal processing method for a Brillouin optical time-domain analysis system according to claim 1, characterized in that, The step S3, which involves directly extracting temperature information from the Brillouin gain spectrum based on the data model, specifically involves: S301. Establish a data model and conduct continuous learning and training to obtain the mapping function relationship between Brillouin frequency shift and temperature; S302. Through simulation and laboratory calibration, generate L training samples ( A i , T i ), i =1,2,…, L ,in For the first i Each sample, i.e., the Brillouin gain spectrum under different frequency sweeps; T i The corresponding output vector is the temperature value; the number of hidden layer nodes is... N The output vector is as follows: In the formula, w j The input weights are the input weights between the input nodes and the hidden layer nodes. β j This represents the output weights between hidden layer nodes and output nodes. b i For the first i The bias of each hidden layer node. g (•) is the activation function, let β =[ β 1 β 2… β N ] T , T =[ T 1 T 2… T L If ], then the output of the hidden layer node is: ； The output vector is expressed in matrix form as follows: Hβ=T ; S303, optical fiber along the route n Each sampling point is selected, and the Brillouin gain spectrum at each sampling point in the optical fiber is input into the trained model for testing to obtain the corresponding temperature, i.e., the temperature information along the optical fiber.

5. The signal processing method for a Brillouin optical time-domain analysis system according to claim 4, characterized in that, Step S3 also includes adjusting the input weights. w j and bias b i Optimization: Use the root mean square error of the prediction results in the training set as the fitness function, set the parameter optimization objective to minimize the root mean square error of the prediction results, and exit the loop when the fitness value is minimized or the maximum number of iterations is met.

6. A signal processing terminal for a Brillouin optical time-domain analysis system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the signal processing method of the Brillouin optical time-domain analysis system according to any one of claims 1-5.

Images