Fault positioning method, device and equipment for communication optical cable and medium

Abstract

The embodiment of the invention provides a fault positioning method and device for a communication optical cable, equipment and a medium, and the method comprises the steps: carrying out the preprocessing of distributed acoustic sensing data, and obtaining the processed distributed acoustic sensing data; detecting abnormal data in the processed distributed acoustic sensing data by using a dual-threshold threshold algorithm; performing analysis of different domains on the abnormal data to obtain multi-dimensional features; and identifying the multi-dimensional features by using a trained support vector machine model, and determining the fault position of the distributed acoustic sensing data, so as to solve the problem that the real-time dangerous case positioning of the communication optical cable line cannot be realized at present.

Description

Technical Field

[0001] The present application relates to the field of optical fiber sensing technology, and in particular to a method, device, electronic equipment and storage medium for locating a fault of a communication optical cable. Background Art

[0002] In order to cope with sudden failures and reduce maintenance costs when laying communication optical cables, existing communication operator service companies usually lay some more ordinary single-mode communication optical fibers as sensor fibers, also known as spare fibers or dark fibers.

[0003] In order to effectively prevent illegal construction and illegal excavation from damaging key infrastructure such as communication optical cables, traditional protection measures mainly rely on a combination of limited point sensors and manual inspections. This strategy usually adopts periodic inspections, including regular inspections by personnel or vehicles along the pipeline.

[0004] In the field of communication optical cable protection, infrared detection technology, pulse electronic detection technology, tension detection technology and intelligent video analysis technology are mainly used. However, the above four sensing technologies generally have the problem of high false alarm rate and are not suitable for long-distance continuous monitoring, which limits their wide application.

[0005] Therefore, existing monitoring solutions based on manual inspections and point sensors can only detect visible external risks on the inspection path, but traditional methods are often powerless against hidden or passive external interference. Many communication cable blocking incidents were not discovered until after they occurred, exposing the shortcomings of existing technologies in preventing sudden and hidden risks.

[0006] In addition, most of the existing monitoring solutions are post-event remedial measures and cannot achieve real-time monitoring, resulting in a significant lag in problem discovery and response. In addition, the manual maintenance model also faces challenges such as heavy workload, low efficiency and frequent line failures. Therefore, how to locate the real-time danger of optical cable lines has become one of the problems that need to be solved urgently. Summary of the invention

[0007] The technical problem to be solved by the embodiments of the present application is to provide a method, device, electronic device and storage medium for locating faults of communication optical cables, so as to solve the problem that it is currently impossible to locate dangerous situations of communication optical cable lines in real time.

[0008] In a first aspect, this embodiment provides a method for locating a fault of a communication optical cable, comprising:

[0009] Preprocessing the distributed acoustic sensing data to obtain processed distributed acoustic sensing data;

[0010] Detecting abnormal data in the processed distributed acoustic sensor data using a dual threshold algorithm;

[0011] Analyzing the abnormal data in different domains to obtain multi-dimensional features;

[0012] The multi-dimensional features are identified using the trained support vector machine model to determine the fault location of the distributed acoustic sensing data.

[0013] Optionally, the detecting abnormal data in the processed distributed acoustic sensing data by using a dual threshold algorithm includes:

[0014] Set short-time energy threshold and variance threshold;

[0015] Calculating the short-term energy value and variance of each frame of the processed distributed acoustic sensing data;

[0016] Compare the short-time energy value with the short-time energy threshold to obtain a first comparison result;

[0017] Compare the variance with the variance threshold to obtain a second comparison result;

[0018] Abnormal data is determined according to the first comparison result and the second comparison result.

[0019] Optionally, performing different domain analyses on the abnormal data to obtain multi-dimensional features includes:

[0020] Extracting a set number of time-domain features from the abnormal data;

[0021] Extracting a set number of frequency domain features from the time domain features;

[0022] Extracting a set number of wavelet domain features from the frequency domain features;

[0023] The multi-dimensional features are composed according to the time domain features, the frequency domain features and the wavelet domain features.

[0024] Optionally, the using the trained support vector machine model to identify the multidimensional features and determine the fault location of the distributed acoustic sensing data includes:

[0025] Matching the multidimensional features with vibration features in the support vector machine model;

[0026] If the multi-dimensional feature is consistent with the vibration feature, the position corresponding to the multi-dimensional feature is determined as the fault position of the distributed acoustic sensing data.

[0027] Optionally, the method further comprises: pre-training the support vector machine model;

[0028] The pre-training of the support vector machine model comprises:

[0029] Constructing a sample dataset based on anomaly data and distributed acoustic sensing data;

[0030] Calibrate the sample data set to obtain a calibrated sample data set;

[0031] Extracting multidimensional features of different domains from each data in the calibration sample data set;

[0032] Constructing a feature sample library according to the multidimensional features;

[0033] The multidimensional features in the feature sample library are input into a support vector machine model, and the support vector machine model is trained based on a least squares support vector machine algorithm of particle swarm optimization to obtain a trained support vector machine model.

[0034] In a second aspect, this embodiment provides a communication optical cable fault location device, including:

[0035] A data preprocessing module, used for preprocessing distributed acoustic sensing data to obtain processed distributed acoustic sensing data;

[0036] A detection module, used to detect abnormal data in the processed distributed acoustic sensor data using a dual threshold algorithm;

[0037] A feature extraction module, used to analyze the abnormal data and extract multi-dimensional features of different domains;

[0038] The identification module is used to identify the multi-dimensional features using the trained support vector machine model to determine the fault location of the distributed acoustic sensing data.

[0039] Optionally, the detection module includes:

[0040] A threshold unit, used to set a short-time energy threshold and a variance threshold;

[0041] A calculation unit, used to calculate the short-time energy value and variance of each frame of the processed distributed acoustic sensing data;

[0042] A first comparing unit, configured to compare the short-time energy value with the short-time energy threshold to obtain a first comparison result;

[0043] A second comparing unit, configured to compare the variance with the variance threshold to obtain a second comparison result;

[0044] A determining unit is used to determine abnormal data according to the first comparison result and the second comparison result.

[0045] Optionally, the feature extraction module includes:

[0046] A time domain unit, used to extract a set number of time domain features from the abnormal data;

[0047] A frequency domain unit, used to extract a set number of frequency domain features from the time domain features;

[0048] A wavelet domain unit, used to extract a set number of wavelet domain features from the frequency domain features;

[0049] A synthesis unit is used to compose the multi-dimensional feature according to the time domain feature, the frequency domain feature and the wavelet domain feature.

[0050] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0051] A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the communication optical cable fault location method when executing the computer program. In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program that implements the communication optical cable fault location method when executed by the processor.

[0052] Compared with the prior art, the embodiments of the present application have the following advantages:

[0053] In the embodiment of the present application, firstly, by analyzing the abnormal data in the processed distributed acoustic sensor data, multi-dimensional features of different domains are extracted. Compared with the single-domain features of the prior art, the multi-dimensional features of different domains can better represent the original distributed acoustic sensor data, making different communication optical cables more distinguishable. This significantly improves the recognition rate and accuracy of the subsequent support vector machine model, so that the fault of the communication optical cable can be accurately located, and the distributed acoustic sensor data is fault located by the support vector machine model, which solves the problem of extremely delayed dangerous situation response caused by limited monitoring devices and reliance on manual monitoring in the prior art, realizes real-time dangerous situation positioning of the communication optical cable line, and ensures the safe and reliable operation of the communication optical cable communication network.

[0054] Secondly, by using a dual threshold algorithm to detect abnormal data in the processed distributed acoustic sensor data, the complete pattern recognition processing of all distributed acoustic sensor data in the prior art can be avoided, thereby improving the real-time response of the system, computing and storage resources.

[0055] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A schematic diagram of the structure of a distributed acoustic sensing hardware system provided in an embodiment of the present application;

[0057] Figure 2 A flowchart of a method for locating a fault in a communication optical cable provided in an embodiment of the present application;

[0058] Figure 3 A flowchart of the steps of detecting abnormal data in the processed distributed acoustic sensing data using a dual threshold algorithm provided in an embodiment of the present application;

[0059] Figure 4 A flowchart of a method for locating a fault in a communication optical cable provided in an embodiment of the present application;

[0060] Figure 5 A discrete wavelet transform decomposition block diagram provided in an embodiment of the present application;

[0061] Figure 6 A flowchart of a method for locating a fault in a communication optical cable provided in an embodiment of the present application;

[0062] Figure 7 A schematic flow chart of the steps for pre-training the support vector machine model for this application;

[0063] Figure 8 A schematic structural diagram of a communication optical cable fault location device provided in an embodiment of the present application;

[0064] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0066] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0067] Reference Figure 1 , shows a schematic diagram of a distributed acoustic sensing hardware system provided in an embodiment of the present application, the hardware system specifically comprising:

[0068] The distributed acoustic sensing hardware system mainly includes: a narrow linewidth laser 1, an isolator 2, a first optical coupler 3, an acousto-optic modulator 4, a first erbium-doped fiber amplifier 5, a circulator 6, a sensing light 7, a second erbium-doped fiber amplifier 8, a fiber Bragg grating 9, a second optical coupler 10, a balanced photodetector 11, a data acquisition card 12, a host computer 13 and an arbitrary waveform generator 14.

[0069] The distributed acoustic sensing hardware system is based on the distributed acoustic sensing (DAS) technology, which is an advanced technology that uses optical fiber as a sensor to monitor sound waves and vibrations by detecting the scattered signals generated inside the optical fiber. This technology can perform continuous and real-time sound and vibration monitoring over long distances and large areas.

[0070] In this application, the core of the distributed acoustic sensing hardware system adopts an advanced phase-sensitive optical time domain reflectometry (φ-OTDR) system, which provides a solid physical basis for the safety monitoring of communication optical cables with its high sensitivity and high spatial resolution. Through this distributed acoustic sensing hardware system, it is possible to capture small changes along the optical cable and convert these changes into analyzable signals.

[0071] The main working principles of the φ-OTDR system are as follows:

[0072] When the φ-OTDR system is working, first, the narrow linewidth laser 1 is used to emit laser light with a specific linewidth to ensure high coherence. The laser then passes through an isolator 2 to prevent the reflected light from returning to the narrow linewidth laser 1. Then, the light beam is sent to the first optical coupler 3. The first optical coupler 3 divides the light beam into two beams according to a specific ratio, one of which is used as a reference light and directly enters the second optical coupler 9; the other beam continues to move forward, and the forward continuous laser beam enters the acousto-optic modulator 4, which is modulated into a pulse sequence of specific width and shape by the acousto-optic modulator 4, and then amplified by the first erbium doped fiber amplifier 5 (Erbium Doped Fiber Application Amplifier, EDFA) to enhance the signal strength. The amplified pulse light enters the circulator 6 and is then injected into the sensing fiber 7. In the sensing fiber 7, the light pulse interacts with the optical fiber material to generate Rayleigh scattered light, which will propagate in all directions, and a part of it will propagate in the opposite direction of the original pulse light to form backward Rayleigh scattered light. The backward Rayleigh scattered light carries information along the optical fiber, including the attenuation characteristics of the optical fiber and the refractive index changes caused by any disturbance or event. The backward Rayleigh scattered light passes through the circulator 6 again. Due to the directional characteristics of the circulator 6, the scattered light comes out from different ports. The scattered light is amplified again by the second erbium-doped fiber amplifier 8 and filtered by the fiber Bragg grating 9 to reduce noise and improve signal quality. The filtered Rayleigh scattered light enters the second optical coupler 10, meets the reference light and generates beat frequency interference to form a beat frequency signal. The beat frequency signal is converted into an electrical signal by the balanced photodetector 11, and this electrical signal is collected by the data acquisition card 13. The collected processed signal is sent to the host computer 13 for further processing and analysis to determine the disturbance event along the optical fiber. The narrow linewidth laser 1 emits a laser with a specific linewidth and can also be processed by the arbitrary waveform generator 14. The processed electrical signal is collected by the data acquisition card 13 and sent to the host computer 13.

[0073] Based on the above distributed acoustic sensing hardware system, refer to Figure 2 , shows a flowchart of the steps of a method for locating a fault of a communication optical cable provided in an embodiment of the present application, specifically comprising:

[0074] Step 201: pre-process the distributed acoustic sensing data to obtain processed distributed acoustic sensing data.

[0075] In practical applications, distributed acoustic sensing data is collected through a distributed acoustic sensing hardware system, and the collected distributed acoustic sensing data is preprocessed. The preprocessing mainly includes: denoising, detrending and normalizing the distributed acoustic sensing data to improve the accuracy and efficiency of subsequent processing.

[0076] In a specific application, the original DAS data is first read, and the amplitude of the DAS data in each channel is normalized to [-1, 1], so that the DAS data is mapped to the [-1, 1] interval, thereby achieving data normalization.

[0077]

[0078] Among them, X N is the processed distributed acoustic sensing data, x i is the ith sequence value, x min and x max are the minimum and maximum values ​​of the sequence, respectively.

[0079] In addition, since the drift of distributed acoustic sensors can cause large optical path changes in optical fiber signals, the actual collected optical fiber event signals often show jitter and instability, which will affect the subsequent feature extraction and recognition and classification. Therefore, it is necessary to perform detrending processing on the optical fiber event signal. This detrending processing is to subtract a best fit curve from the original signal so that the mean value of the signal is zero, which can effectively filter out the signal jitter of the optical fiber event signal and eliminate the long-term change trend in the signal, so that the analysis is focused on the short-term changes or periodic characteristics of the signal.

[0080] For example, for a fiber event signal sequence with a length of N, the specific steps for detrending are as follows:

[0081] (1) Calculate the cumulative difference of the optical fiber event signal sequence as a new signal sequence. The new signal sequence can be calculated by the following formula:

[0082]

[0083] Among them, y(t) is the new signal sequence, x(i) is the original data value at time point i, is the mean value of the original data x(i).

[0084] (2) Divide the obtained new sequence y(t) into m signal sequence intervals of length n.

[0085] (3) Use the least squares method to perform linear fitting on each signal interval to obtain the local trend y of each interval n (t).

[0086] (4) Subtract y from the sequence y(t) n (t), and calculate the root mean square error.

[0087]

[0088] Among them, X RMSEstands for root mean square error.

[0089] (5) Change the interval length n and repeat steps 2 to 4 until a suitable optical fiber signal sequence is obtained.

[0090] Step 202: Detect abnormal data in the processed distributed acoustic sensing data using a dual threshold algorithm.

[0091] The dual threshold algorithm refers to a dual threshold algorithm using short-time energy and time domain variance to detect abnormal data in the processed distributed acoustic sensor data, that is, to detect the fluctuation segment of the vibration data in the processed distributed acoustic sensor data.

[0092] See also Figure 3 , a flowchart of the steps of using a dual threshold algorithm to detect abnormal data in the processed distributed acoustic sensor data is described in detail, specifically including the following steps:

[0093] Step 301: Setting the short-time energy threshold E T and variance threshold V T .

[0094] In specific applications, the short-time energy threshold E T The short-time energy threshold used to determine whether the signal may contain effective activity, so this application selects the short-time energy threshold E T As a threshold, since the variance can measure the degree of signal fluctuation, this application selects the short-time energy variance V T as another threshold.

[0095] Step 302: Calculate the short-term energy value and variance of each frame of the processed distributed acoustic sensing data.

[0096] The sliding window endpoint detection method based on double thresholds calculates the short-term energy value E of each frame of distributed acoustic sensor data after processing. K .

[0097]

[0098] Among them, x i is the i-th sample value of the abnormal data in the time window of the k-th frame, and N is the window size.

[0099] Calculate the variance of the processed distributed acoustic sensing data for each frame and define the time domain variance V K :

[0100]

[0101] Among them, μ k is the mean value of the k-th frame signal.

[0102] Step 303: Compare the short-time energy value with the short-time energy threshold to obtain a first comparison result.

[0103] The short-time energy value E k With the short-time energy threshold E T For comparison, if E k Greater than E T , then E k Greater than E T As a first comparison result, if E k Less than or equal to E T , the processed distributed acoustic sensing data is considered invalid and no processing is performed.

[0104] Step 304: Compare the variance with the variance threshold to obtain a second comparison result.

[0105] The variance V k With the variance threshold V T For comparison, if V k Greater than V T , then V k Greater than V T As the second comparison result, if V k Less than or equal to V T , the processed distributed acoustic sensing data is considered invalid and no processing is performed.

[0106] Step 305: Determine abnormal data according to the first comparison result and the second comparison result.

[0107] If E k Greater than E T , and V k Also greater than V T That is, if the short-term energy and variance of the processed distributed acoustic sensor data exceed the set threshold, the abnormal data is determined to be a valid active segment, and the output waveform, start time and end time of the abnormal data are obtained. The abnormal data is suitable for feature extraction; if it does not exceed the set threshold, the abnormal data is considered invalid and no further processing is performed. The above method can effectively screen out signal segments containing useful vibration information, laying the foundation for subsequent feature extraction and analysis.

[0108] In practical applications, the first comparison result and the second comparison result can be expressed using the following endpoint detection decision expression:

[0109]

[0110] Among them, D K Indicates the endpoint detection result of the Kth frame signal.

[0111] Step 203: Analyze the abnormal data in different domains to obtain multi-dimensional features.

[0112] The abnormal data is analyzed from the time domain, frequency domain and wavelet domain, and different features are extracted from the time domain, frequency domain and wavelet domain respectively. The features extracted from the time domain, frequency domain and wavelet domain form multi-dimensional features. Through multi-dimensional features, we can more comprehensively understand the structure and detailed information of the data to improve the accuracy of recognition.

[0113] Step 204: Use the trained support vector machine model to identify the multi-dimensional features and determine the fault location of the distributed acoustic sensing data.

[0114] In actual applications, the trained support vector machine model also sets a dynamically adjusted alarm threshold based on historical alarm data or environmental variables. If the prediction result output by the multi-dimensional feature through the support vector machine model is greater than the alarm threshold, a fault alarm is triggered.

[0115] In the embodiment of the present application, first, by analyzing the abnormal data in the processed distributed acoustic sensor data, multi-dimensional features of different domains are extracted. Compared with the single-domain features of the prior art, the multi-dimensional features of different domains can better represent the original distributed acoustic sensor data, making different communication optical cables more distinguishable. This significantly improves the recognition rate and accuracy of the subsequent support vector machine model, so that the fault of the communication optical cable can be accurately located, and the distributed acoustic sensor data is fault-located through the support vector machine model, which solves the problem of extremely delayed dangerous situation response caused by limited monitoring devices and reliance on manual monitoring in the prior art, realizes real-time dangerous situation positioning of the communication optical cable line, and ensures the safe and reliable operation of the communication optical cable communication network.

[0116] Secondly, by using a dual threshold algorithm to detect abnormal data in the processed distributed acoustic sensor data, the complete pattern recognition processing of all distributed acoustic sensor data in the prior art can be avoided, thereby improving the real-time response of the system, computing and storage resources.

[0117] Reference Figure 4 , shows a flowchart of the steps of a method for locating a fault of a communication optical cable provided in an embodiment of the present application, specifically comprising:

[0118] Step 401: pre-process the distributed acoustic sensing data to obtain processed distributed acoustic sensing data.

[0119] Step 402: Detect abnormal data in the processed distributed acoustic sensing data using a dual threshold algorithm.

[0120] Step 403: Analyze the abnormal data in different domains to obtain multi-dimensional features.

[0121] As an implementation method, step 403 includes the following sub-steps:

[0122] Sub-step 4031: extracting a set number of time domain features from the abnormal data according to a preset algorithm.

[0123] In practical applications, the following time domain features are mainly extracted from the abnormal data: mean absolute value MAV, mean absolute value slope MAVSlope, waveform length WL, slope sign change SSC, zero crossing rate ZC, standard deviation s, skewness SKE, kurtosis KUR, peak factor PF, pulse factor IF, etc., a total of 10 time domain features. Other features can also be selected, and this application does not make any specific restrictions on this.

[0124] Among them, x i is the amplitude of the signal value of the ith sampling point, N is the total number of signal sampling points, then the expression of the mean absolute value MAV is:

[0125]

[0126] The mean absolute value slope MAVSlope is used to describe the average steepness of the signal over time, which is expressed as

[0127]

[0128] The zero-crossing rate ZC refers to the number of times the signal changes from positive to negative or vice versa, expressed as:

[0129]

[0130] in,

[0131] Among them, sgn(x) represents the compliance function and threshold represents the threshold.

[0132] Slope sign change SSC, represents the frequency information of the signal, indicating the number of times the slope changes from positive to negative or vice versa, expressed as:

[0133]

[0134]

[0135] The waveform length WL provides information about the complexity of the signal and is defined as the cumulative length of the waveform over time, described as:

[0136]

[0137] The peak value refers to the global maximum value, that is, if there is no other point in the entire signal interval with a value greater than that of a certain point, then this point is the peak value of the entire signal, which is described as:

[0138] Peak=max(|x(n)|)

[0139] Among them, Peak is the global maximum value of the peak value

[0140] The mean is the sum of a set of values ​​divided by the number of values ​​in the set, and is described as:

[0141]

[0142] The standard deviation is used to describe the degree of deviation of a set of values ​​from its mean. The larger the standard deviation, the greater the difference between the data points and the mean; the smaller the standard deviation, the more concentrated the data points are around the mean.

[0143]

[0144] Skewness SKE is a statistic that measures the asymmetry of data distribution. It is used to describe which side of the mean the shape of the data distribution is biased towards. It is described as:

[0145]

[0146] Kurtosis KUR is a statistic that measures the peak degree and tail thickness of data distribution. It is used to describe the morphological characteristics of data distribution. Its formula is expressed as

[0147]

[0148] The peak factor PF, also known as the crest factor or peak factor, is a quantity used to describe the relationship between the maximum amplitude and the effective value of a signal. The peak factor measures the ratio of the maximum value of a signal to its effective value:

[0149]

[0150] The impulse factor IF, also known as the shock factor, is a measure of the relationship between the maximum amplitude of a signal and its average value. The impulse factor measures the ratio of the maximum value of a signal to its average value, expressed as:

[0151]

[0152] Sub-step 4032: extracting a set number of frequency domain features from the time domain features.

[0153] Among them, the frequency domain features include: frequency center, frequency entropy and autocorrelation features, etc. Other features can also be selected, and this application does not make specific restrictions on this.

[0154] For a discrete-time signal x(i), the frequency domain representation X(k) is obtained by the following discrete Fourier transform (DFT):

[0155]

[0156] Where k is the frequency and N is the total number of sampling points.

[0157] The frequency center is the "center of mass" of the spectrum, which indicates at which frequency the main frequency components of the signal are concentrated. The formula is as follows:

[0158]

[0159] Where N is the number of DFT points, k represents the frequency index, |X(k)| 2 is the power spectral density of the spectrum.

[0160] Frequency entropy refers to the uncertainty of the energy distribution of the spectrum. If the energy of a signal is concentrated in a few frequency components, its frequency entropy is low; conversely, if the energy is evenly distributed across all frequency components, the frequency entropy is high. The formula is as follows:

[0161]

[0162] in, is the normalized power spectral density.

[0163] The frequency domain autocorrelation function can be used to describe the relationship between different frequency components of the signal. The autocorrelation function calculated in the time domain can be converted to the frequency domain by Fourier transform. However, directly calculating the autocorrelation feature in the frequency domain usually involves multiplying the spectrum X(k) with its own complex conjugate and then performing an inverse Fourier transform. However, if only the autocorrelation feature in the frequency domain is considered without involving time domain conversion, the autocorrelation of the modulus of the spectrum can be calculated:

[0164]

[0165] Where τ is a delay or offset, and Rx(τ) represents the autocorrelation value of the spectrum |X(k)| at a delay of τ.

[0166] Sub-step 4033: extracting a set number of wavelet domain features from the frequency domain features.

[0167] The advantage of the discrete wavelet transform method is that it can decompose nonlinear and non-stationary signals at multiple scales. The signal is decomposed into a series of scale coefficients (approximate coefficients) and wavelet coefficients (detail coefficients) through discrete wavelet transform. For a one-dimensional signal x(i) of length N, the recursive formula for calculating the discrete wavelet transform using the Mallat fast algorithm is as follows:

[0168] Assume x(i) is the original signal, h(k) and g(k) are the coefficients of the low-pass filter (also called the analysis wavelet) and the high-pass filter (also called the detail wavelet), respectively, and their lengths are usually equal and less than N. Figure 5 The discrete wavelet transform (DWT) decomposition block diagram of the embodiment of the present application is shown. The DWT can be expressed as:

[0169] (1) First, the signal is convolved with the low-pass filter h(k) and the high-pass filter g(k) to obtain the approximation coefficient cA and detail coefficient cD:

[0170]

[0171] Where L is the length of the filter and j is the index of the downsampling.

[0172] (2) Then, the approximate coefficient cA is downsampled, that is, a value is taken at every other point to obtain the input signal of the next level:

[0173] x next (i) = cA(2i).

[0174] (3) Repeat the above steps until the signal length is not long enough to continue decomposition.

[0175] (4) Finally, all detail coefficients cD and the last approximation coefficient together constitute the result of discrete wavelet transform. Common wavelet basis functions include Haar wavelet, Daubechies wavelet, etc.

[0176] In practical applications, wavelet domain features usually include energy features, entropy features, mean absolute deviation, variance, kurtosis and skewness, a total of 6 features. Other features can also be selected, and this application does not make specific restrictions on this.

[0177] The energy feature describes the energy distribution of the signal at different scales and can be used to distinguish different signal types. The formula is:

[0178]

[0179] Among them, E j is the energy of the j-th layer wavelet decomposition, cD j and cAj are the detail coefficient and approximation coefficient of the jth layer respectively.

[0180] The entropy feature reflects the uncertainty or complexity of the signal and is often used for signal classification. The formula is:

[0181]

[0182] in, is the probability of the i-th coefficient in the j-th layer, H j is the entropy of the jth layer.

[0183] Mean absolute deviation, this feature reflects the degree of fluctuation of the signal, and the formula is expressed as:

[0184]

[0185] in, is the average value of the detail coefficients of the jth layer.

[0186] Variance describes the dispersion of the signal, and the formula is:

[0187]

[0188] Kurtosis is a statistic that measures the peak degree and tail thickness of data distribution. It is used to describe the morphological characteristics of data distribution. The formula is:

[0189]

[0190] Skewness is a statistic that measures the asymmetry of data distribution. It is used to describe which side the shape of the data distribution is biased towards. The formula is:

[0191]

[0192] Sub-step 4034: compose the multi-dimensional features according to the time domain features, the frequency domain features and the wavelet domain features.

[0193] After the above steps, 10 time domain features, 3 frequency domain features and 6 wavelet domain features are extracted. The above time domain, frequency domain and wavelet domain features are combined into a 19-dimensional multi-dimensional feature. The multi-dimensional feature can better represent the original DAS event signal, making different communication optical cable warning events more distinguishable, and significantly improving the recognition rate and accuracy of the subsequent support vector machine model.

[0194] Step 404: Match the multi-dimensional features with the vibration features in the support vector machine model.

[0195] Step 405: If the multi-dimensional feature is consistent with the vibration feature, the position corresponding to the multi-dimensional feature is determined as the fault position of the distributed acoustic sensing data.

[0196] In the embodiment of the present application, features of abnormal data are screened from multiple dimensions such as time domain, frequency domain, and wavelet domain, and the multi-dimensional features of different domains are combined into a multi-domain feature vector. Compared with the single-domain feature extraction algorithm, the multi-domain feature vector can better represent the original DAS event signal, making different communication optical cable failure events more distinguishable, and significantly improving the recognition rate and accuracy of the subsequent support vector machine classification model.

[0197] Reference Figure 6 , shows a flowchart of the steps of a method for locating a fault of a communication optical cable provided in an embodiment of the present application, specifically comprising:

[0198] Step 601: pre-train the support vector machine model.

[0199] Among them, Support Vector Machine (SVM) is a supervised learning model widely used in classification and regression tasks. The basic principle of SVM is to find an optimal decision boundary (hyperplane) in the feature space to maximize the interval between positive and negative samples.

[0200] Both the standard SVM and the Least Squares Support Vector Machine (LSSVM) have two hyperparameters: regularization parameter and kernel function parameter. These two parameters will have a significant impact on the trained classification model. If the settings are unreasonable, it will have a relatively large impact on the classification accuracy. Therefore, in actual classification scenarios, if you want to use an efficient SVM model, you need to adjust the hyperparameters and select the best regularization parameter and kernel function parameter.

[0201] In order to improve the accuracy and training efficiency of the SVM model, this patent application adopts a classification method based on an improved particle swarm optimization support vector machine for classification and recognition. The PSO algorithm is a heuristic optimization algorithm based on swarm intelligence. Its basic idea is to simulate the interaction between individuals in a group (called particles) in the process of seeking a global optimal solution. When the algorithm is running, it first generates a population, which includes multiple particles. Each particle in the population has two properties, namely, the initial position and the initial velocity. After that, the algorithm begins to gradually optimize the particles in the population under the action of cooperation and competition mechanisms. This process is called iterative optimization. After completing a certain number of optimal solutions or reaching the set number of cycles through multiple iterative optimization processes, the cycle can be terminated and the result can be output. Simply put, under the particle swarm algorithm model, all feasible solutions to each optimization problem will be regarded as particles, and the state of the particles is jointly represented by the position vector and velocity vector of the particles at this time. Particles constantly adjust their own positions and seek the optimal solution through optimization learning.

[0202] Particle Swarm Optimization Least Squares Support Vector Machine (PSO-LSSVM) is an algorithm that combines particle swarm optimization (PSO) and least squares support vector machine (LSSVM). The PSO-LSSVM algorithm selects the PSO algorithm to find the optimal particles as the C and σ of LSSVM. The fitness is the accuracy of correctly classifying the test data set after substituting the C and σ of LSSVM into the LSSVM for training. The process of PSO-LSSVM is similar to that of PSO-SVM. The biggest difference is that PSO-LSSVM uses the classification accuracy of the LSSVM classification model as the fitness of the particle to iteratively update the particle speed and particle position in the population to find the optimal particle, that is, the optimal parameters C and σ.

[0203] See also Figure 7 , a flowchart of the steps of pre-training the support vector machine model in this application is described in detail, which specifically includes the following steps:

[0204] Step 701: construct a sample data set according to abnormal data and distributed acoustic sensing data.

[0205] In practical applications, it is necessary to start the DAS demodulator and visualization host computer composed of the hardware system to manually collect vibration events (abnormal data) and environmental sample data (distributed acoustic sensor data). For example, to collect excavation events, it is necessary to manually tap the optical cable or use a machine to create vibration in advance to calibrate which point the data transmitted back to the host computer corresponds to the current position. For example, if point A is tapped continuously, 200 values ​​are returned to the host computer each time to form a 200*1 vector data. Each data represents the vibration value at each virtual sensor point. At this time, point A is tapped, and the 10th data has the highest peak value, which means that the 10th data corresponds to the data generated by the current tapping. Then, with the spatial resolution designed by the system, such as a spatial resolution of 5m, that is, a virtual sensor point (device) every 5m, the actual optical cable position corresponding to the remaining data can be presented. Then, various vibration events that need to be subsequently identified are created at this location, such as excavation events, and the host computer is used to automatically save this event signal to the corresponding event folder.

[0206] Step 702: Calibrate the sample data set to obtain a calibrated sample data set.

[0207] Use the dataset creation tool to manually select and intercept event signals in the event folder (generally the part with relatively large vibration values. Some events are intermittent during collection, such as knocking events, so they need to be intercepted one by one) to obtain a one-dimensional signal of unequal length and save it in .mat format and put it into the corresponding event folder, such as a folder for mining events, a folder for knocking events, and a folder for environmental events. Each folder contains several data saved in .mat. In this way, a series of event folders are obtained to save these sample data sets.

[0208] Step 703: extract multi-dimensional features of different domains from each data in the calibration sample data set.

[0209] In practical applications, it is necessary to use pre-written python code to extract time domain, frequency domain, and wavelet domain features from a series of one-dimensional signals in the calibration sample data set to form a 19-dimensional artificial feature. In this way, the vector signals of varying lengths in the data set are converted into one-dimensional feature vectors of equal length, for example: a 19*1 feature vector (if some features are solved and return a vector rather than a single value, it may also be larger than 19*1). The one-dimensional signal represents the detailed features of each signal. Therefore, the design and extraction of artificial features are very critical. It must describe the details of the signal itself that distinguish it from other signals so that it is easy to distinguish event types when it is sent to the SVM model to obtain excellent classification effects.

[0210] Step 704: construct a feature sample library according to the multi-dimensional features.

[0211] The feature sample library constructed based on multi-dimensional features includes both the data set obtained in the calibration data sample stage and the data set produced by long-term collection of event signals and multiple calibration data samples. The purpose of establishing this feature sample library is that after the classification model is trained for the first time, if the data set produced in the calibration data stage is too small or not representative, it is easy to have difficulty in characterizing the SVM model under all weather conditions and all interference events. Therefore, the feature sample library will produce data sets irregularly during the use of the entire system to train the SVM model again and obtain better classification accuracy.

[0212] Step 705: input the multidimensional features in the feature sample library into a support vector machine model, and train the support vector machine model based on a least squares support vector machine algorithm of particle swarm optimization to obtain a trained support vector machine model.

[0213] After obtaining multidimensional features from the special sample library, select the kernel function for LSSVM, determine the value range of the kernel function parameter σ and the regularization parameter C, and set the PSO parameters. Initialize the global position P of the particle swarm, the position pi and velocity vi of the current particle, determine the kernel function parameter σ and the regularization parameter C according to the initialized particle swarm, and build the support vector machine model.

[0214] According to the operation results of the support vector machine model, the current particle fitness is evaluated, the current global optimal position P of the particle swarm is determined, and it is judged whether the number of iterations exceeds the maximum iteration value or whether the optimal support vector machine model has been obtained. If the number of iterations exceeds the maximum value or the optimal support vector machine model is obtained, the optimal least squares support vector machine model is output.

[0215] The DAS data to be predicted is processed to obtain multi-dimensional features, and the multi-dimensional features are input into an optimal least squares support vector machine model to determine the fault location of the distributed acoustic sensor data.

[0216] If the number of iterations does not exceed the maximum value, or the optimal support vector machine model is not obtained, the position pi and velocity vi of particle i are updated, the LSSMV parameter σ and regularization parameter C are determined according to the new particles, and the support vector machine model is rebuilt. The current particle fitness is evaluated according to the operation results of the support vector machine model, and the current global optimal position P of the particle swarm is determined.

[0217] Step 602: pre-process the distributed acoustic sensing data to obtain processed distributed acoustic sensing data.

[0218] Step 603: Detect abnormal data in the processed distributed acoustic sensing data using a dual threshold algorithm.

[0219] Step 604: Analyze the abnormal data in different domains to obtain multi-dimensional features.

[0220] Step 605: Use the trained support vector machine model to identify the multi-dimensional features and determine the fault location of the distributed acoustic sensing data.

[0221] In the embodiment of the present application, first, the particle swarm algorithm is used to iteratively process the parameters of the least squares support vector machine to obtain the optimal least squares support vector machine model, and finally the extracted features are subjected to pattern recognition to achieve accurate classification and recognition of different types of dangerous situations. Secondly, the least squares support vector machine algorithm based on particle swarm optimization not only improves the recognition speed compared to the traditional SVM, but also iterates the optimal parameters of the SVM through the particle swarm algorithm, thereby improving the accuracy of pattern recognition.

[0222] Reference Figure 8 , shows a schematic structural diagram of a communication optical cable fault location device provided in an embodiment of the present application, specifically comprising:

[0223] The data preprocessing module 801 is used to preprocess the distributed acoustic sensing data to obtain processed distributed acoustic sensing data.

[0224] The detection module 802 is used to detect abnormal data in the processed distributed acoustic sensing data using a dual threshold algorithm.

[0225] The feature extraction module 803 is used to analyze the abnormal data and extract multi-dimensional features of different domains.

[0226] The identification module 804 is used to identify the multi-dimensional features using a support vector machine model to determine the fault location of the distributed acoustic sensing data.

[0227] Optionally, the detection module includes:

[0228] A threshold unit, used to set a short-time energy threshold and a variance threshold;

[0229] A calculation unit, used to calculate the short-time energy value and variance of each frame of the processed distributed acoustic sensing data;

[0230] A first comparing unit, configured to compare the short-time energy value with the short-time energy threshold to obtain a first comparison result;

[0231] A second comparing unit, configured to compare the variance with the variance threshold to obtain a second comparison result;

[0232] A determining unit is used to determine abnormal data according to the first comparison result and the second comparison result.

[0233] Optionally, the feature extraction module includes:

[0234] A time domain unit, used to extract a set number of time domain features from the abnormal data according to a preset algorithm;

[0235] A frequency domain unit, used to extract a set number of frequency domain features from the time domain features;

[0236] A wavelet domain unit, used to extract a set number of wavelet domain features from the frequency domain features;

[0237] A synthesis unit is used to compose the multi-dimensional feature according to the time domain feature, the frequency domain feature and the wavelet domain feature.

[0238] Optionally, the identification module includes:

[0239] A matching unit, used for matching the multi-dimensional features with the vibration features in the support vector machine model;

[0240] A fault location unit is used to determine the position corresponding to the multi-dimensional feature as the fault position of the distributed acoustic sensing data if the multi-dimensional feature is consistent with the vibration feature.

[0241] Optionally, the device further comprises: a training module, used for pre-training the support vector machine model;

[0242] The training module includes:

[0243] Constructing a sample unit, which is used to construct a sample data set according to the abnormal data and the distributed acoustic sensing data;

[0244] A calibration unit, used for calibrating the sample data set to obtain a calibrated sample data set;

[0245] An extraction unit, used for extracting multi-dimensional features of different domains from each data in the calibration sample data set;

[0246] A construction unit, used to construct a feature sample library according to the multi-dimensional features;

[0247] The input unit is used to input the multidimensional features in the feature sample library into the support vector machine model, and train the support vector machine model based on the least squares support vector machine algorithm of particle swarm optimization to obtain the trained support vector machine model.

[0248] In the embodiment of the present application, firstly, by analyzing the abnormal data in the processed distributed acoustic sensor data, multi-dimensional features of different domains are extracted. Compared with the single-domain features of the prior art, the multi-dimensional features of different domains can better represent the original distributed acoustic sensor data, making different communication optical cables more distinguishable. This significantly improves the recognition rate and accuracy of the subsequent support vector machine model, so that the fault of the communication optical cable can be accurately located, and the distributed acoustic sensor data is fault located by the support vector machine model, which solves the problem of extremely delayed dangerous situation response caused by limited monitoring devices and reliance on manual monitoring in the prior art, realizes real-time dangerous situation positioning of the communication optical cable line, and ensures the safe and reliable operation of the communication optical cable communication network.

[0249] Secondly, by using a dual threshold algorithm to detect abnormal data in the processed distributed acoustic sensor data, the complete pattern recognition processing of all distributed acoustic sensor data in the prior art can be avoided, thereby improving the real-time response of the system, computing and storage resources.

[0250] An embodiment of the present application further provides an electronic device, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the communication optical cable fault location method when executing the computer program.

[0251] Fig. 9 FIG. 1 is a schematic diagram showing the structure of an electronic device 900 according to an embodiment of the present invention. Fig. 9 As shown, the electronic device 900 includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) 902 or computer program instructions loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 can also be stored. The CPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0252] Multiple components in the electronic device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, a microphone, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the electronic device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0253] The various processes and processing described above may be performed by the processing unit 901. For example, the method of any of the above embodiments may be implemented as a computer software program, which is tangibly contained in a computer-readable medium, such as the storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the CPU 901, one or more actions in the method described above may be performed.

[0254] In addition, an embodiment of the present application further provides a computer-readable storage medium, which stores a computer program that implements the communication optical cable fault locating method when executed by a processor.

[0255] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0256] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0257] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminals (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0258] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0259] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal so that a series of operating steps are executed on the computer or other programmable terminal to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable terminal. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0260] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0261] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or terminal including the elements.

[0262] The above is a detailed introduction to a communication optical cable fault location method, a communication optical cable fault location device, an electronic device and a computer-readable storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for a person skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for locating a fault in a communication optical cable, characterized in that: include: Preprocessing the distributed acoustic sensing data to obtain processed distributed acoustic sensing data; Detecting abnormal data in the processed distributed acoustic sensor data using a dual threshold algorithm; Analyzing the abnormal data in different domains to obtain multi-dimensional features; The multi-dimensional features are identified using the trained support vector machine model to determine the fault location of the distributed acoustic sensing data.

2. The fault location method according to claim 1, characterized in that: The method of detecting abnormal data in the processed distributed acoustic sensing data by using a dual threshold algorithm includes: Set short-time energy threshold and variance threshold; Calculating the short-term energy value and variance of each frame of the processed distributed acoustic sensing data; Compare the short-time energy value with the short-time energy threshold to obtain a first comparison result; Compare the variance with the variance threshold to obtain a second comparison result; Abnormal data is determined according to the first comparison result and the second comparison result.

3. The fault location method according to claim 1, characterized in that: The performing different domain analyses on the abnormal data to obtain multi-dimensional features includes: Extracting a set number of time-domain features from the abnormal data; Extracting a set number of frequency domain features from the time domain features; Extracting a set number of wavelet domain features from the frequency domain features; The multi-dimensional features are composed according to the time domain features, the frequency domain features and the wavelet domain features.

4. The fault location method according to claim 1, characterized in that: The using the trained support vector machine model to identify the multi-dimensional features and determine the fault location of the distributed acoustic sensor data includes: Matching the multidimensional features with vibration features in the support vector machine model; If the multi-dimensional feature is consistent with the vibration feature, the position corresponding to the multi-dimensional feature is determined as the fault position of the distributed acoustic sensing data.

5. The fault location method according to any one of claims 1 to 4, characterized in that: The method further comprises: pre-training the support vector machine model; The pre-training of the support vector machine model comprises: Constructing a sample dataset based on anomaly data and distributed acoustic sensing data; Calibrate the sample data set to obtain a calibrated sample data set; Extracting multidimensional features of different domains from each data in the calibration sample data set; Constructing a feature sample library according to the multidimensional features; The multidimensional features in the feature sample library are input into a support vector machine model, and the support vector machine model is trained based on a least squares support vector machine algorithm of particle swarm optimization to obtain a trained support vector machine model.

6. A fault location device for a communication optical cable, characterized in that: include: A data preprocessing module, used for preprocessing distributed acoustic sensing data to obtain processed distributed acoustic sensing data; A detection module, used to detect abnormal data in the processed distributed acoustic sensor data using a dual threshold algorithm; A feature extraction module, used to analyze the abnormal data and extract multi-dimensional features of different domains; The identification module is used to identify the multi-dimensional features using the trained support vector machine model to determine the fault location of the distributed acoustic sensing data.

7. The device according to claim 6, characterized in that The detection module comprises: A threshold unit, used to set a short-time energy threshold and a variance threshold; A calculation unit, used to calculate the short-time energy value and variance of each frame of the processed distributed acoustic sensing data; A first comparing unit, configured to compare the short-time energy value with the short-time energy threshold to obtain a first comparison result; A second comparing unit, configured to compare the variance with the variance threshold to obtain a second comparison result; A determining unit is used to determine abnormal data according to the first comparison result and the second comparison result.

8. The device according to claim 6, characterized in that The feature extraction module comprises: A time domain unit, used to extract a set number of time domain features from the abnormal data; A frequency domain unit, used to extract a set number of frequency domain features from the time domain features; A wavelet domain unit, used to extract a set number of wavelet domain features from the frequency domain features; A synthesis unit is used to compose the multi-dimensional feature according to the time domain feature, the frequency domain feature and the wavelet domain feature.

9. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for locating a fault in a communication optical cable as claimed in any one of claims 1 to 5 when executing the computer program.

10. A computer-readable storage medium, characterized in that: When the computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for locating a fault in a communication optical cable according to any one of claims 1 to 5.

Images