Signal denoising method and device for optical fiber sensing vibration monitoring

Abstract

The invention provides a signal denoising method and device for optical fiber sensing vibration monitoring. The method comprises the following steps: collecting vibration data of a to-be-detected optical fiber, and judging the type (signal or noise) of the to-be-detected optical fiber according to a mahalanobis distance; then, eliminating signal data, filling noise data by using an interpolation method, and generating a noise signal; thirdly, amplitude and phase angle sequences of noise and vibration signals are calculated through Fourier transform, and then an amplitude difference and a phase angle difference are obtained; and finally, calculating a de-noised frequency spectrum according to the difference values, and obtaining a de-noised vibration monitoring signal through inverse Fourier transform. The method is small in calculation amount and rapid, the output signal-to-noise ratio and the algorithm accuracy can be improved, the false alarm rate and the missing report rate are reduced, and the engineering application requirement is met.

Description

Technical Field

[0001] This application belongs to the field of model construction, and particularly relates to a signal denoising method and device for fiber optic sensing vibration monitoring. Background Art

[0002] Fiber optic sensing technology can monitor vibration signals caused by various external events, and these signals contain the effects of all external disturbances and noises on the optical fiber. For the monitored signals, the analysis process is mainly divided into two stages: data processing and type recognition. In the data processing stage, the effect of suppressing system noise is directly related to the accuracy of subsequent signal type recognition. Therefore, it is necessary to perform denoising processing on the collected initial signals to ensure the effectiveness and stability of the system operation.

[0003] Common denoising methods include median filtering, moving average filtering, Gaussian filtering, wavelet threshold denoising, etc. However, in the actual engineering application site, the distributed fiber optic sensing system faces a variety of complex noises, and the traditional denoising methods for point sensor data have poor effects. Summary of the Invention

[0004] The purpose of this application is to overcome the defects in the prior art and provide a signal denoising method and device for fiber optic sensing vibration monitoring.

[0005] This application provides a signal denoising method for fiber optic sensing vibration monitoring, including:

[0006] Collect vibration data on the fiber optic to be measured, and judge the category of the vibration data according to the Mahalanobis distance of the vibration data. This category includes signal segment sample data and noise segment sample data;

[0007] Eliminate the data segments belonging to the signal segment sample data, and fill in the noise segment sample data using the interpolation method to generate a noise signal;

[0008] According to the real part sequence and imaginary part sequence obtained by performing Fourier transform on the signal sequences of the noise signal and the vibration data respectively, calculate the first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration signal;

[0009] Subtract the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtract the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference;

[0010] Calculate and generate a denoised spectrum according to the amplitude difference and the phase angle difference, and perform inverse Fourier transform to obtain the vibration monitoring signal after denoising.

[0011] Optionally, calculate a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal, with the expressions as follows:

[0012]

[0013]

[0014] where A0 is the real part sequence of the noise signal, B0*i is the imaginary part sequence of the noise signal, i is the unit imaginary number, Amp0 is the first amplitude sequence, and Angle0 is the first phase angle sequence;

[0015]

[0016]

[0017] where A is the real part sequence of the vibration signal, B*i is the imaginary part sequence of the vibration signal, i is the unit imaginary number, Amp is the second amplitude sequence, and Angle is the second phase angle sequence.

[0018] Optionally, calculate and generate a denoised spectrum based on the amplitude difference and the phase angle difference, with the expression as follows:

[0019] Y new = amp new sin(angle new ) + amp new cos(amp new )i;

[0020] where Amp new , Angle new are the amplitude and phase angle after denoising respectively, and i is the unit imaginary number.

[0021] Optionally, collect vibration data on the optical fiber to be measured, including:

[0022] Collect vibration data in 3 spatial dimensions of points;

[0023] Collect vibration data in 4096 time dimensions of points.

[0024] Optionally, the calculation expression of the Mahalanobis distance is as follows:

[0025]

[0026] where L k is the Mahalanobis distance, is each vibration data, m s is the clustering center of each category; C sis the covariance matrix of the pattern set, and W is the weight matrix.

[0027] Optionally, judging the category of the vibration data according to the Mahalanobis distance of the vibration data includes:

[0028] Calculating the Mahalanobis distances of the vibration data in two categories respectively;

[0029] Comparing the magnitudes of the calculated Mahalanobis distances in the two categories, and the vibration data belongs to the category with the smaller Mahalanobis distance.

[0030] Optionally, the steps of determining the clustering center include:

[0031] Conducting a vibration event simulation experiment;

[0032] Clustering the vibration data collected from the vibration event simulation experiment to generate signal segment sample data and noise segment sample data;

[0033] Calculating the clustering center according to the generated signal segment sample data and noise segment sample data.

[0034] The present application also provides a signal denoising device for fiber optic sensing vibration monitoring, including:

[0035] A collection module for collecting vibration data on a fiber optic to be measured, and judging the category of the vibration data according to the Mahalanobis distance of the vibration data, and the category includes signal segment sample data and noise segment sample data;

[0036] A noise module for removing the data segments belonging to the signal segment sample data and filling the noise segment sample data by interpolation to generate a noise signal;

[0037] A transformation module for calculating a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to a real part sequence and an imaginary part sequence obtained by respectively performing Fourier transforms on signal sequences of the noise signal and the vibration data;

[0038] A denoising module for obtaining an amplitude difference by subtracting the first amplitude sequence from the second amplitude sequence, and obtaining a phase angle difference by subtracting the first phase angle sequence from the second phase angle sequence;

[0039] An inverse transformation module for calculating and generating a denoised spectrum according to the amplitude difference and the phase angle difference, and performing an inverse Fourier transform to obtain a vibration monitoring signal after denoising.

[0040] The present application also provides a signal denoising device for fiber optic sensing vibration monitoring, including:

[0041] A memory for storing a computer-executable program for the method of signal denoising of the above-mentioned fiber optic sensing vibration monitoring;

[0042] A processor for retrieving the computer-executable program from the memory and performing: collecting vibration data on the fiber optic to be measured, judging the category of the vibration data according to the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; removing the data segments belonging to the signal segment sample data, and filling the noise segment sample data by using the interpolation method to generate a noise signal; calculating a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to the real part sequence and the imaginary part sequence obtained by respectively performing Fourier transform on the signal sequences of the noise signal and the vibration data; subtracting the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference; calculating and generating a denoising spectrum according to the amplitude difference and the phase angle difference, and performing inverse Fourier transform to obtain a vibration monitoring signal after denoising.

[0043] The present application also provides a storage medium storing a computer-executable program, which is used to be called by a processor to execute the steps of the method of signal denoising of the above-mentioned fiber optic sensing vibration monitoring.

[0044] Advantages and beneficial effects of the present application:

[0045] The present application provides a method for signal denoising of fiber optic sensing vibration monitoring, including: collecting vibration data on the fiber optic to be measured, judging the category of the vibration data according to the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; removing the data segments belonging to the signal segment sample data, and filling the noise segment sample data by using the interpolation method to generate a noise signal; calculating a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to the real part sequence and the imaginary part sequence obtained by respectively performing Fourier transform on the signal sequences of the noise signal and the vibration data; subtracting the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference; calculating and generating a denoising spectrum according to the amplitude difference and the phase angle difference, and performing inverse Fourier transform to obtain a vibration monitoring signal after denoising. This method has a small amount of calculation and is fast, can improve the output signal-to-noise ratio, and is applicable to the distributed fiber optic sensing engineering site. It can also improve the accuracy of the positioning and recognition algorithm, reduce the false alarm and missed alarm rates of vibration monitoring, so as to more effectively process the sensing data and meet the requirements of engineering applications. Description of the Drawings

[0046] Figure 1 It is a schematic diagram of the signal denoising process for fiber optic sensing vibration monitoring in this application.

[0047] Figure 2 It is a schematic diagram of the signal characteristics of the crusher operation of the disturbance signal sample in this application.

[0048] Figure 3 It is a schematic diagram of the signal characteristics of the excavator operation of the disturbance signal sample in this application.

[0049] Figure 4 It is a schematic diagram of the processing effect of the excavator signal data in this application.

[0050] Figure 5 It is a schematic diagram of the processing effect of the pile driver signal data in this application. Specific embodiments

[0051] The following further describes this application in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand this application and be able to implement it.

[0052] The following content are all examples of the specific implementation process provided to detail the technical solution to be protected by this application. However, this application can also be implemented in other ways different from the described ones. Under the guidance of the concept of this application, those skilled in the art can use different technical means to implement this application. Therefore, this application is not limited by the following specific embodiments.

[0053] This application provides a signal denoising method for fiber optic sensing vibration monitoring, including: collecting vibration data on the fiber to be measured, judging the category of the vibration data according to the Mahalanobis distance of the vibration data, and this category includes signal segment sample data and noise segment sample data; removing the data segments belonging to the signal segment sample data, and filling the noise segment sample data using the interpolation method to generate a noise signal; calculating the first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration signal according to the real part sequence and the imaginary part sequence obtained by performing Fourier transforms on the signal sequences of the noise signal and the vibration data respectively; subtracting the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference; calculating and generating a denoised spectrum according to the amplitude difference and the phase angle difference, and performing an inverse Fourier transform to obtain the vibration monitoring signal after denoising. This method has a small amount of calculation, is fast, can improve the output signal-to-noise ratio, and is applicable to the distributed fiber optic sensing engineering site. It can also improve the accuracy of the positioning and recognition algorithms, reduce the false alarm and missed alarm rates of vibration monitoring, so as to more effectively process the sensing data and meet the engineering application requirements.

[0054] Figure 1It is a schematic diagram of the signal denoising process of optical fiber sensing vibration monitoring in this application.

[0055] Please refer to Figure 1 As shown, a signal denoising method for optical fiber sensing vibration monitoring comprises the following steps:

[0056] S101 collects vibration data on the optical fiber to be tested, and determines the category of the vibration data according to the Mahalanobis distance of the vibration data, which category includes signal segment sample data and noise segment sample data.

[0057] First, an event simulation experiment is conducted to build a rich event sample library. In this application, a simulation scenario of a crusher and an excavator working near an optical cable is taken as an example to collect relevant vibration data and then build a disturbance signal sample database.

[0058] Under a specific disturbance event, the sample data obtained is expressed as Where k is the unique number of the sample. According to the characteristic value of the signal, these data can be divided into two main sets:

[0059]

[0060] S + : represents the sample set of the target signal segment, where The value range of is (-δ, +δ);

[0061] S0: represents the sample set of the noise segment, where The value range of is (-∞,-δ]∪[+δ,+∞).

[0062] It should be noted that It belongs to the R^n space and covers a variety of feature data, such as frequency domain features, short-time energy features, average amplitude and impact features, etc. In addition, for different types of disturbance signal events (such as the disturbance of a crusher or excavator), the value of the parameter δ will be different.

[0063] like Figure 2 and Figure 3 As shown in Figure 1, there are signal characteristics of the disturbance signal samples of crusher and excavator respectively.

[0064] S102 removes the data segment belonging to the signal segment sample data, and fills the noise segment sample data using an interpolation method to generate a noise signal.

[0065] Through cluster analysis, S + and S0 are considered as two pre-clustered pattern sets. In order to determine each sample To determine which pattern set a pattern belongs to, a classification method based on Mahalanobis distance is adopted.

[0066] Mahalanobis distance is a distance metric that takes into account the correlation between data, and it calculates the distance between a sample and the center m of each pattern set s . Specifically, the Mahalanobis distance L' between m s and each pattern set is given by the following formula: k

[0067]

[0068] Here, m s is the clustering center of each category s, s ∈ S + , S0; C s is the covariance matrix of this pattern set, and the W is the weight matrix. The elements in C s are represented as C i,j :

[0069]

[0070] In this formula, q is the sample number in the pattern set s; l is the number of samples of this pattern; x is an element in the vector , and i, j are the labels of this element, representing signal feature values such as frequency domain features, short-time energy features, average amplitude, and impulse features; is the average value of elements i and j.

[0071] By calculating the Mahalanobis distance, it is possible to determine which pattern set center a given sample is closer to, and thus classify it into the corresponding pattern set. This method helps to more accurately identify and classify different types of disturbance signal events. Specifically, the one with the minimum distance determines the category of the vibration data.

[0072] The matrix weight expression is as follows:

[0073] W = diag(ε1, ε2,..., ε j ,...);

[0074] This matrix weight indicates that the influence degrees of various features on the discriminant pattern set are different. Therefore, the weights of each feature factor should also be considered when calculating the Mahalanobis distance.

[0075] In this application, first, signals at various locations on the fiber to be measured are selected from a large amount of measured data. These signals are the vibration responses of the optical cable at different time and space points, containing rich information, including both signals of the normal operation of the optical cable and abnormal signals caused by external disturbances.

[0076] For each selected signal, the above formula for calculating the Mahalanobis distance is used to calculate their distances from S + ​The Mahalanobis distance between the two pattern sets S0 (signal segment sample set) and S1 (noise segment sample set). Specifically, for the signal to be tested, Can calculate and These two distance values respectively represent the similarity between the signal and the target sensor signal under external disturbance and the noise signal.

[0077] After getting the Mahalanobis distance, we will compare and According to the minimum distance principle, if Less than It is believed that the signal belongs to S + , that is, the target sensor signal under external disturbance; on the contrary, if If it is smaller, the signal is classified as a noise segment.

[0078] After pattern classification, it is possible to identify which data segments belong to the target sensor signal S under external disturbance + . In order to obtain a data set containing only noise information, it is necessary to remove the target sensor data segments under external disturbance from the original data. However, removing the data will cause discontinuous or incomplete data. To solve this problem, interpolation is used to fill in these removed data segments. Interpolation is a mathematical method that can estimate and fill in missing data based on known data points. After interpolation, a new data set X0 is obtained, which only contains noise information.

[0079] S103 calculates the first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration original signal data according to the real part sequence and the imaginary part sequence obtained by Fourier transforming the signal sequence of the noise signal and the vibration original signal data obtained by monitoring.

[0080] Performing a fast Fourier transform on the signal sequence X0(k,:) of the kth spatial point in X0 means that we want to analyze the characteristics of this spatial point in the frequency domain. Fast Fourier transform (FFT) is an efficient algorithm used to calculate the discrete Fourier transform (DFT) of a sequence. By performing FFT on a signal, the signal can be converted from the time domain to the frequency domain, so that the size and phase of different frequency components in the signal can be observed.

[0081] The transformed signal sequence is expressed as F0=A0+B0*i, which involves complex representation. In complex representation, a complex number consists of a real part and an imaginary part, where:

[0082] A0 represents the real part;

[0083] B0 represents the Imaginary Part;

[0084] i is the imaginary unit, satisfying i^2 = -1.

[0085] Therefore, F0 is a complex sequence, where each element is a complex number with a real part and an imaginary part. This complex sequence describes the characteristics of the signal at the k-th spatial point in the frequency domain. Specifically, A0 and B0 provide information on the amplitude and phase of this frequency component respectively.

[0086] When processing the FFT results, the amplitude sequence and phase sequence of each frequency component are further calculated, and the expressions are as follows:

[0087]

[0088]

[0089] where Amp0 is the first amplitude sequence of the transformed signal, and Angle0 is the first phase angle sequence of the transformed signal.

[0090] Performing a fast Fourier transform on the signal X(k, :) at the k-th spatial point in the vibration data X means performing a frequency-domain analysis on the time-domain vibration signal at this spatial point. After the fast Fourier transform, the obtained signal is represented as F = A + B*i, which is the result represented by a complex number.

[0091] A represents the real part sequence of the vibration signal after the fast Fourier transform. It is an array, and each element in it corresponds to a time point in the original signal X(k, :). The real part sequence A describes the distribution of the amplitudes of the various frequency components in the signal in the frequency domain, that is, the amplitude sequence.

[0092] B*i represents the imaginary part sequence of the signal after the fast Fourier transform. Similarly, B is also an array, and each element corresponds to the time point in the original signal X(k, :). The imaginary part sequence B*i describes the distribution of the phases of the various frequency components in the signal in the frequency domain, that is, the phase angle sequence. Here, i is the unit imaginary number, satisfying i^2 = -1.

[0093] Furthermore, the frequency-domain signal is represented by amplitude and phase angle, and the expressions are as follows:

[0094]

[0095]

[0096] where amp is the second amplitude sequence, representing the amplitude magnitude of each frequency component of the transformed signal. For each frequency component, angle is the second phase angle sequence, representing the phase angle of each frequency component of the transformed signal.

[0097] S104 Subtract the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtract the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference.

[0098] Calculating the difference between the original signal and the noise signal is to obtain a pure signal after removing the noise.

[0099] To obtain the denoised amplitude amp new , it is necessary to subtract the amplitude of the noise signal from the amplitude Amp of the original noisy signal. This process can help remove the influence of the noise, making the new amplitude sequence amp new closer to the true vibration signal.

[0100] Mathematically, this operation can be expressed as:

[0101]

[0102] The phase difference between the pure signal of the vibration and the noisy signal is usually very small. This is because the noise usually does not significantly change the phase of the signal. Therefore, in most cases, the phase angle angle new of the denoised signal can remain unchanged. This means that the phase angle Angle of the original noisy signal can be directly used as the phase angle angle new of the denoised signal, i.e., angle = Angle.

[0103] S105 Calculate and generate a denoised spectrum based on the amplitude difference and the phase angle difference, and perform an inverse Fourier transform to obtain the denoised signal.

[0104] Calculate and generate a denoised spectrum based on the amplitude difference and the phase angle difference, and its expression is as follows:

[0105] Y new = amp new sin(angle new ) + amp new cos(amp new )i

[0106] where Y new is the denoised spectrum.

[0107] Use the inverse Fourier transform algorithm to transform Y new . This process converts Y new from the frequency domain back to the time domain. Mathematically, this operation can be expressed as: X k = IFFT(Y new ), where IFFT represents the inverse fast Fourier transform.

[0108] After inverse Fourier transform, the denoised time domain signal X is obtained. k The signal is now a time series where each point is the denoised signal value.

[0109] X k is the denoised signal of X(k,:), which means that X k represents the denoised vibration signal of the k-th spatial point at all N moments. Therefore, X k Represented as a row vector of size 1×N.

[0110] Finally, k is 1, 2, 3, ..., M in turn, and the above steps are followed to obtain the denoised signals X1, X2, X3, ..., X M , then the signal X after denoising the two-dimensional array X filted It can be expressed as:

[0111]

[0112] According to the above method, the noise removal of different vibration detection data is completed.

[0113] like Figure 4 and Figure 5 As shown in Figure 2, they are schematic diagrams of the processed vibration signals of the crusher and excavator.

[0114] The present application also provides a signal denoising device for optical fiber sensing vibration monitoring, comprising:

[0115] A collection module, used for collecting vibration data on the optical fiber to be tested, and determining the category of the vibration data according to the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data;

[0116] A noise module, used to remove the data segment belonging to the signal segment sample data, and fill the noise segment sample data by interpolation method to generate a noise signal;

[0117] a transformation module, configured to calculate a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to a real part sequence and an imaginary part sequence obtained by performing Fourier transform on the signal sequences of the noise signal and the vibration data respectively;

[0118] a denoising module, configured to obtain an amplitude difference by subtracting the first amplitude sequence from the second amplitude sequence, and obtain a phase angle difference by subtracting the first phase angle sequence from the second phase angle sequence;

[0119] The inverse transformation module is used to calculate and generate a denoised spectrum according to the amplitude difference and the phase angle difference, and perform inverse Fourier transformation to obtain a denoised vibration monitoring signal.

[0120] The present application also provides a signal denoising device for optical fiber sensing vibration monitoring, comprising:

[0121] A memory for storing a computer executable program for the above-mentioned signal denoising method for optical fiber sensing vibration monitoring;

[0122] A processor is used to retrieve the computer executable program from the memory and execute: collecting vibration data on the optical fiber to be tested, judging the category of the vibration data according to the size of the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; eliminating the data segment belonging to the signal segment sample data, and filling the noise segment sample data by interpolation to generate a noise signal; calculating the first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration signal according to the real sequence and the imaginary sequence obtained by Fourier transforming the signal sequence of the noise signal and the vibration data respectively; subtracting the first amplitude sequence from the second amplitude sequence to obtain the amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain the phase angle difference; calculating and generating a denoised spectrum according to the amplitude difference and the phase angle difference, and performing an inverse Fourier transform to obtain a denoised vibration monitoring signal.

[0123] The present application also provides a storage medium storing a computer executable program, wherein the computer executable program is used to be called by a processor to execute the steps of the above-mentioned signal denoising method for optical fiber sensing vibration monitoring.

Claims

1. A signal denoising method for fiber optic sensing vibration monitoring, characterized in that, include: Collecting vibration data on the optical fiber to be tested, and determining the category of the vibration data according to the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; Eliminate the data segment belonging to the signal segment sample data, and fill the noise segment sample data by interpolation method to generate a noise signal; Calculate a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to a real part sequence and an imaginary part sequence obtained by Fourier transforming the signal sequences of the noise signal and the vibration data respectively; Subtracting the first amplitude sequence from the second amplitude sequence to obtain an amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain a phase angle difference; A de-noised spectrum is calculated based on the amplitude difference and the phase angle difference, and an inverse Fourier transform is performed to obtain a de-noised vibration monitoring signal.

2. The signal denoising method for fiber optic sensing vibration monitoring according to claim 1, wherein The first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration signal are calculated, and the expressions are as follows: Where A0 is the real part sequence of the noise signal, B0*i is the imaginary part sequence of the noise signal, i is a unit imaginary number, Amp0 is the first amplitude sequence, and Angle0 is the first phase angle sequence; Among them, A is the real part sequence of the vibration signal, B*i is the imaginary part sequence of the vibration signal, i is a unit imaginary number, Amp is the second amplitude sequence, and Angle is the second phase angle sequence.

3. The signal denoising method for fiber optic sensing vibration monitoring according to claim 1, characterized in that, The de-noised spectrum is calculated and generated according to the amplitude difference and the phase angle difference, and the expression is as follows: Y new = amp new sin(angle new ) + amp new cos(amp new )i; Among them, Amp new and Angle new are the amplitude and phase angle after noise removal respectively, and i is the unit imaginary number.

4. The signal denoising method for fiber optic sensing vibration monitoring according to claim 1, wherein Collect vibration data on the fiber under test, including: Collect vibration data in three spatial dimensions; Collect vibration data at 4096 points in the time dimension.

5. The signal denoising method for fiber optic sensing vibration monitoring according to claim 1, characterized in that, The calculation expression of the Mahalanobis distance is as follows: Among them, L k is the Mahalanobis distance, are the respective vibration data, and m s is the clustering center of each category; C s is the covariance matrix of the mode set, and the said W is the weight matrix.

6. The signal denoising method for fiber optic sensing vibration monitoring according to claim 5, characterized in that, Determining the category of the vibration data according to the Mahalanobis distance of the vibration data includes: respectively calculating the Mahalanobis distances of the vibration data in two categories; Comparing the Mahalanobis distances calculated in the two categories, the vibration data belongs to the category with the smaller Mahalanobis distance.

7. The signal denoising method for optical fiber sensing vibration monitoring according to claim 5, wherein The step of determining the cluster center comprises: Conduct vibration event simulation experiments; Clustering is performed according to the vibration data collected by the vibration event simulation experiment to generate signal segment sample data and noise segment sample data; The cluster center is calculated according to the generated signal segment sample data and noise segment sample data.

8. A signal denoising device for fiber optic sensing vibration monitoring, characterized in that, include: A collection module, used for collecting vibration data on the optical fiber to be tested, and determining the category of the vibration data according to the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; A noise module, used to remove the data segment belonging to the signal segment sample data, and fill the noise segment sample data by interpolation method to generate a noise signal; a transformation module, configured to calculate a first amplitude sequence and a first phase angle sequence of the noise signal, and a second amplitude sequence and a second phase angle sequence of the vibration signal according to a real part sequence and an imaginary part sequence obtained by performing Fourier transform on the signal sequences of the noise signal and the vibration data respectively; a denoising module, configured to obtain an amplitude difference by subtracting the first amplitude sequence from the second amplitude sequence, and obtain a phase angle difference by subtracting the first phase angle sequence from the second phase angle sequence; The inverse transformation module is used to calculate and generate a denoised spectrum according to the amplitude difference and the phase angle difference, and perform inverse Fourier transformation to obtain a denoised vibration monitoring signal.

9. A signal denoising device for fiber optic sensing vibration monitoring, characterized in that, include: A memory for storing a computer executable program of a signal denoising method for optical fiber sensing vibration monitoring according to any one of claims 1 to 7; A processor, for retrieving the computer executable program from the memory, and executing: collecting vibration data on the optical fiber to be tested, judging the category of the vibration data according to the size of the Mahalanobis distance of the vibration data, the category including signal segment sample data and noise segment sample data; removing the data segment belonging to the signal segment sample data, and filling the noise segment sample data by interpolation to generate a noise signal; calculating the first amplitude sequence and the first phase angle sequence of the noise signal, and the second amplitude sequence and the second phase angle sequence of the vibration signal according to the real part sequence and the imaginary part sequence obtained by Fourier transforming the signal sequences of the noise signal and the vibration data respectively; subtracting the first amplitude sequence from the second amplitude sequence to obtain the amplitude difference, and subtracting the first phase angle sequence from the second phase angle sequence to obtain the phase angle difference; A de-noised spectrum is calculated based on the amplitude difference and the phase angle difference, and an inverse Fourier transform is performed to obtain a de-noised vibration monitoring signal.

10. A storage medium, characterized in that, A computer executable program is stored, and the computer executable program is used to be called by a processor to execute the steps of a signal denoising method for optical fiber sensing vibration monitoring according to any one of claims 1 to 7.

Images