A micro-motion optical fiber fusion detection method and system

Abstract

The present invention discloses a micro-motion fiber fusion detection method and system, which includes constructing a micro-motion fiber fusion perception network to obtain historical perception data and position data, performing edge calibration to obtain a temperature-compensated electrical signal and performing edge screening to obtain a vibration electrical signal, performing edge matching and determining the monitoring scale characteristics, respectively inputting the monitoring scale characteristics and fiber process parameters into an influence function to obtain a scale index and a process index, performing data inversion on the node vibration wave to obtain the inversion vibration characteristics, performing edge processing on the vibration electrical signal to obtain fiber micro-motion characteristics, constructing a micro-motion fiber fusion detection model according to the fiber micro-motion characteristics, and inputting the to-be-detected perception data into the micro-motion fiber fusion detection model to obtain a detection result. This method can not only improve the efficiency and accuracy of micro-motion fiber fusion detection, but also has good interpretability and can be directly applied to the micro-motion fiber fusion detection system.

Description

Technical Field

[0001] The present invention relates to the field of micro - motion detection, and particularly to a micro - motion optical fiber fusion detection method and system. Background Technique

[0002] With the rapid development of optical fiber sensing technology, the micro - motion optical fiber detection technology has been widely used in the fields of structural health monitoring, earthquake early warning, oil and gas pipeline monitoring, etc. By precisely capturing and analyzing weak vibration signals, the micro - motion optical fiber can detect potential risks in advance, providing strong support for ensuring people's life and property safety and maintaining the stable operation of infrastructure.

[0003] However, traditional micro - motion optical fiber detection technology can only be linearly arranged. The micro - motion seismographs lack flexibility in layout, and the analog signals of seismographs can better preserve low - frequency signals than optical fiber sensors. In addition, the information obtained by optical fiber sensors is too one - sided to comprehensively and accurately perceive micro - motion conditions in complex environments. Finally, optical fiber sensors are poor in dealing with interference from environmental factors such as temperature, resulting in poor accuracy of detection data. The present invention constructs a micro - motion optical fiber fusion perception network to comprehensively collect multi - source information such as optical fiber electrical signals, node vibration waves, temperature, and historical vibration data, and uses edge calibration, screening, matching, and data inversion for data pre - processing, and constructs a micro - motion optical fiber fusion detection model to predict ground vibration. A micro - motion optical fiber fusion detection method and system are designed, effectively overcoming the deficiencies of traditional technologies. It can not only eliminate temperature interference, deeply excavate vibration characteristics, improve the accuracy and reliability of micro - motion detection, provide an efficient and accurate solution for the monitoring work in related fields, and strongly promote the development of micro - motion detection technology towards intelligence and refinement. Summary of the Invention

[0004] The object of the present invention is to provide a micro - motion optical fiber fusion detection method and system.

[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0006] The present invention includes the following steps:

[0007] Construct a micro - motion optical fiber fusion perception network to obtain historical perception data and location data; the historical perception data includes optical fiber electrical signals, node vibration waves, temperature, and historical vibration data;

[0008] Perform edge calibration according to the temperature and the optical fiber electrical signal to obtain a temperature - compensated electrical signal;

[0009] Perform edge screening on the temperature-compensated electrical signal to obtain a vibration electrical signal, perform edge matching on the vibration electrical signal and the node seismic wave according to the position data, and determine the monitoring scale characteristics. Input the monitoring scale characteristics and the optical fiber process parameters into the influence function respectively to obtain a scale index and a process index. The influence function includes a scale influence function and a process influence function.

[0010] Perform data inversion on the node vibration wave to obtain the inversion vibration characteristics, and perform edge processing on the vibration electrical signal to obtain the optical fiber micro-motion characteristics. The inversion vibration characteristics include inversion vibration intensity, inversion vibration azimuth, and inversion vibration type. The inversion vibration azimuth includes an inversion vibration area and an inversion vibration depth.

[0011] Construct a micro-motion optical fiber fusion detection model according to the optical fiber micro-motion characteristics, and input the data to be detected and sensed into the micro-motion optical fiber fusion detection model to obtain a detection result.

[0012] Further, the methods for obtaining the sensed data and the position data include:

[0013] The micro-motion optical fiber fusion sensing network includes a node instrument module, an optical fiber module, and a temperature sensor module.

[0014] The node instrument module collects node seismic waves through a seismograph.

[0015] The optical fiber module collects the interference light intensity signal of the optical fiber sensor through an optical cable, and converts the interference light intensity signal into an optical fiber electrical signal through an optical fiber demodulator. The optical cable connects multiple optical fiber sensors in series to form an optical fiber sensor array.

[0016] The temperature sensor module is used to collect the ambient temperature of the optical fiber sensor.

[0017] The position data of the micro-motion optical fiber fusion sensing network includes main measurement point position data and auxiliary measurement point position data. The specific obtaining methods include:

[0018] Determine the main measurement line, deploy seismographs on the main measurement line, take the positions of the seismographs as the main measurement points, perform position coding on the main measurement points according to the plane coordinates to obtain the main measurement point position data. Set auxiliary measurement lines parallel to the main measurement line, deploy optical fiber sensors on the auxiliary measurement lines, use an optical cable to connect the optical fiber sensors in series to form an optical fiber sensor array, take each optical fiber sensor as an auxiliary measurement point, and perform position coding on the auxiliary measurement points according to the plane coordinates to obtain the auxiliary measurement point position data. The auxiliary measurement lines include a first auxiliary measurement line and a second auxiliary measurement line. The main measurement points and the auxiliary measurement points jointly form a SPAC array. Temperature sensors are also deployed on the auxiliary measurement points.

[0019] Further, the method for obtaining the temperature-compensated electrical signal through edge calibration includes:

[0020] Obtain the ambient temperature and corresponding optical fiber electrical signals of multiple groups of optical fiber sensors, and establish a temperature-electrical signal deviation model using a multi-layer perceptron model;

[0021] Input the ambient temperature of the optical fiber sensor into the temperature-electrical signal deviation model to obtain the electrical signal deviation, and correct the optical fiber electrical signal according to the electrical signal deviation to obtain the temperature-compensated electrical signal.

[0022] Further, the method for obtaining the scale index and the process index includes:

[0023] Use the SVM support vector machine to search for the optimal hyperplane to divide the temperature-compensated electrical signal into normal electrical signals and vibration electrical signals; the vibration electrical signal represents the temperature-compensated electrical signal related to ground vibration;

[0024] Perform edge matching to obtain the first vibration area: Extract the position data of the auxiliary measurement points corresponding to the vibration electrical signal and the position data of the main measurement points corresponding to the node seismic wave. Use the ball tree algorithm to perform a radius search for each main measurement point to find all auxiliary measurement points whose distance from the main measurement point is less than or equal to the radius. Match the vibration electrical signals of the auxiliary measurement points that meet the search conditions with the node seismic waves of the corresponding main measurement points, and determine the first vibration area according to the position data of the matching auxiliary measurement points; the vibration electrical signal can be matched with multiple node seismic waves at the same time;

[0025] Determine the monitoring scale characteristics according to the position data of the auxiliary measurement points corresponding to the vibration electrical signal and the edge matching result, and input the monitoring scale characteristics into the scale influence function to obtain the scale index. The expression is:

[0026] ,

[0027] where is the scale index, is the spatial coverage weight, is the measurement point redundancy weight, is the spatial dispersion weight, is the current monitoring area, is the standard monitoring area for the effective monitoring data of the sensing network this time, is the length of the main measurement line, is the maximum monitoring length of the sensing network, is the number of auxiliary measurement points, is the number of main measurement points, is the number of effective auxiliary measurement points for the monitoring data of the sensing network this time, is the average distance between auxiliary measurement points, is the maximum distance between the main measurement point and the auxiliary measurement point, is the distance ratio coefficient, is the critical distance;

[0028] Extract the optical fiber process parameters, input them into the process influence function, and obtain the process index. The expression is as follows:

[0029] ;

[0030] Where is the process index, is the fusion quality weight, is the production deformation weight, is the fusion overlap length, is the refractive index matching degree, is the core refractive index, is the average core refractive index, is the effective aperture ratio, is the effective transmission aperture, is the theoretical core area ratio, 、 is the core area constant, , is the core diameter deviation, is the etching depth ratio, is the etching depth, is the designed thickness, is the doping diffusion gradient, is the extrusion deformation rate, is the etching constant.

[0031] Furthermore, the method for obtaining the fiber optic micro-vibration characteristics by performing edge processing on the vibration electrical signal includes:

[0032] Introduce an adaptive criterion to perform the first empirical mode decomposition on the vibration electrical signal until the local energy ratio of the residual signal is less than the preset threshold and stop. After decomposition, obtain K modal signals and a residual signal . The expression for the stop determination criterion of the first empirical mode decomposition is:

[0033] ;

[0034] Where is the vibration electrical signal at k time, is the residual signal at k time, The local energy ratio is calculated using a sliding window, t is the calculation time, is the window width, , is the signal sampling rate, ;

[0035] A multi-scale decomposition strategy is used to The absolute value is used for the second empirical mode decomposition, and each get N modal signal and a residual signal , the multi-scale decomposition strategy expression is:

[0036] ,

[0037] in for t Current signal at the moment No. i indivual The second empirical mode decomposition obtained j indivual , For the i indivual The absolute value of For the j scale parameters;

[0038] Use sliding window to calculate all Signal components and vibration electrical signals The mutual information , filter the signal components whose mutual information is greater than the dynamic threshold of mutual information to obtain effective information. The expression of the dynamic threshold of mutual information is:

[0039] ,

[0040] ,

[0041] in is the mutual information dynamic threshold, , , is the adjustment factor, is the original signal entropy, is the average mutual information, is the standard deviation of the mutual information;

[0042] According to the screening results, the real signal is Hilbert transformed to obtain the complex analytical signal of the vibration electrical signal expression:

[0043] ,

[0044] in j is an imaginary unit, Vibration signal The Hilbert transform of , P is the Cauchy principal value at the transform, is the integration variable;

[0045] According to each signal component , residual signal and to determine the signal energy weight, the expression is:

[0046] ,

[0047] ,

[0048] ,

[0049] where is the weight coefficient of the signal component , is the weight coefficient of the residual signal , is the weight coefficient of the residual signal , is the energy of the signal component , is the energy of the residual signal , is the energy of the residual signal ;

[0050] Calculate the instantaneous amplitude and the instantaneous phase of the vibration electrical signal:

[0051] ,

[0052] ,

[0053] For the instantaneous phase of the vibration electrical signal, take the reciprocal of time to obtain the instantaneous frequency of the vibration electrical signal. The fiber optic micro-vibration feature is composed of the instantaneous amplitude , instantaneous phase and instantaneous frequency of the vibration electrical signal.

[0054] Furthermore, the method for obtaining the detection result includes:

[0055] A comprehensive dataset is composed of fiber optic micro - motion features, historical vibration data, the first vibration region, inverse vibration characteristics, scale index, and process index. The comprehensive dataset is divided into a training set and a test set according to a ratio of 7:3 using the random forest algorithm; the historical vibration data includes historical vibration intensity, historical vibration orientation, and historical vibration type; the historical vibration orientation includes historical vibration depth and historical vibration region;

[0056] A micro - motion fiber optic fusion detection model is constructed. The micro - motion fiber optic fusion detection model specifically includes an input layer, a base model layer, a strategy layer, and an output layer; the base model layer includes a BP neural network, a STAN network, and an LSTM network;

[0057] The BP neural network is used to learn the non - linear coupling relationship between the instantaneous amplitude of the vibration electrical signal, the instantaneous frequency of the vibration electrical signal, the scale index, the process index, and the historical vibration intensity, predict the surface vibration intensity, assign dynamic weights to key features using the self - attention mechanism, and evaluate the difference between the predicted value and the true value of the surface vibration intensity using the mean square error;

[0058] The STAN network is used to learn the spatio - temporal relationship between the instantaneous phase of the vibration electrical signal, the first vibration region, and the historical vibration region, predict the surface vibration region, focus on the time mutation points and spatial distribution characteristics of the phase data using the spatio - temporal attention mechanism, and evaluate the difference between the predicted value and the true value of the surface vibration region using the cross - entropy loss;

[0059] The LSTM network is used to learn the time - series relationship between the fiber optic micro - motion features, the scale index, the process index, and the historical vibration type, predict the surface vibration type, capture the front - and - back time - series information using the bidirectional LSTM, and evaluate the difference between the predicted value and the true value of the surface vibration type using the cross - entropy loss;

[0060] The strategy layer weighted - fuses the predicted value of the surface vibration intensity of the BP neural network and the inverse vibration intensity based on distance weights, fuses the predicted value of the surface vibration of the STAN network and the inverse vibration orientation using a voting mechanism, weighted - fuses the prediction of the surface vibration type of the LSTM network and the inverse vibration type according to the number of measurement points, and connects to the fully - connected layer of the output layer to output the detection result; the detection result includes vibration detection intensity, vibration detection orientation, and vibration detection type; the distance weight is inversely proportional to the distance between the main measurement point and the auxiliary measurement point;

[0061] The Adam optimizer is used to optimize the network structure, the training set is used to train the model, and the test set is used to evaluate the accuracy of the model;

[0062] The to - be - detected perception data is input into the micro - motion fiber optic fusion detection model to obtain the detection result.

[0063] In a second aspect, a micro - motion fiber optic fusion detection system includes:

[0064] Perception network module: Build a micro-motion optical fiber fusion perception network to obtain historical perception data and position data; the micro-motion optical fiber fusion perception network includes a node instrument module, an optical fiber module, and a temperature sensor module; the optical fiber module includes an optical cable, an optical fiber sensor, and an optical fiber demodulator;

[0065] Edge computing module: Used for edge calibration to obtain a temperature-compensated electrical signal; used for edge screening of the temperature-compensated electrical signal to obtain a vibration electrical signal, edge matching the vibration electrical signal and the node seismic wave according to the position data and determining the monitoring scale feature, and inputting the monitoring scale feature and the optical fiber process parameters into the influence function respectively to obtain a scale index and a process index; used for edge processing of the vibration electrical signal to obtain an optical fiber micro-motion feature;

[0066] Cloud model module: Used to construct a micro-motion optical fiber fusion detection model according to the optical fiber micro-motion feature, and input the perception data to be detected into the micro-motion optical fiber fusion detection model to obtain a detection result;

[0067] Intelligent supervision module: Used to store, view, and manage the optical fiber micro-motion feature, the inverted vibration characteristics, the scale index, the process index, and the detection result, and perform surface vibration analysis according to the detection result.

[0068] The beneficial effects of the present invention are:

[0069] The present invention is a micro-motion optical fiber fusion detection method and system. Compared with the prior art, the present invention has the following technical effects:

[0070] Through steps such as edge calibration, edge screening, edge matching, data inversion, index calculation, and model construction, the present invention can improve the data preprocessing ability and enhance the model adaptability in micro-motion optical fiber fusion detection, thereby improving the efficiency of micro-motion optical fiber fusion detection, optimizing the micro-motion optical fiber fusion detection technology, greatly saving resources, improving work efficiency, realizing the detection of surface vibration, providing technical support for the micro-motion detection technology to develop towards the intelligent and refined direction, having far-reaching significance for ensuring the safe and stable operation of various industries, and being able to adapt to the micro-motion optical fiber fusion detection requirements of different micro-motion optical fiber fusion detection systems and different users, having a certain universality. Description of the drawings

[0071] Figure 1 It is the step flow chart of a micro-motion optical fiber fusion detection method of the present invention;

[0072] Figure 2 It is the SPAC array diagram of a micro-motion optical fiber fusion detection system of the present invention;

[0073] In the figure: 1 - optical fiber demodulator; 2 - auxiliary measuring point; 3 - main measuring point; 4 - main measuring line; 5 - auxiliary measuring line; 6 - optical cable. Specific Embodiment

[0074] The present invention will be further described below through specific embodiments. The illustrative embodiments and explanations of the present invention are used to explain the present invention, but do not limit the present invention.

[0075] A micro-motion optical fiber fusion detection method and system of the present invention includes the following steps:

[0076] As Figure 1 shown, in this embodiment, it includes the following steps:

[0077] Construct a micro-motion optical fiber fusion perception network to obtain historical perception data and position data; the historical perception data includes optical fiber electrical signals, node vibration waves, temperature, and historical vibration data;

[0078] Perform edge calibration according to the temperature and the optical fiber electrical signal to obtain a temperature-compensated electrical signal;

[0079] Perform edge screening on the temperature-compensated electrical signal to obtain a vibration electrical signal, perform edge matching on the vibration electrical signal and the node seismic wave according to the position data, and determine the monitoring scale characteristics. Input the monitoring scale characteristics and optical fiber process parameters into the influence function respectively to obtain a scale index and a process index; the influence function includes a scale influence function and a process influence function;

[0080] Perform data inversion on the node vibration wave to obtain the inverted vibration characteristics, and perform edge processing on the vibration electrical signal to obtain the optical fiber micro-motion characteristics; the inverted vibration characteristics include inverted vibration intensity, inverted vibration azimuth, and inverted vibration type; the inverted vibration azimuth includes an inverted vibration area and an inverted vibration depth;

[0081] Construct a micro-motion optical fiber fusion detection model according to the optical fiber micro-motion characteristics, and input the perception data to be detected into the micro-motion optical fiber fusion detection model to obtain a detection result.

[0082] In this embodiment, the method for obtaining the perception data and the position data includes:

[0083] The motor vehicle perception network includes an OBD terminal, a V2X communication module, and an in-vehicle sensor system;

[0084] The micro-motion optical fiber fusion perception network includes a node instrument module, an optical fiber module, and a temperature sensor module;

[0085] The node instrument module collects node seismic waves through a seismograph;

[0086] The optical fiber module collects the interference light intensity signal of the fiber optic sensor through an optical cable, and converts the interference light intensity signal into an optical fiber electrical signal through an optical fiber demodulator; the optical cable connects multiple fiber optic sensors in series to form a fiber optic sensor array;

[0087] The temperature sensor module is used to collect the ambient temperature of the fiber optic sensor;

[0088] The position data of the micro-vibration fiber fusion sensing network includes the position data of the main measurement points and the position data of the auxiliary measurement points. The specific acquisition method includes:

[0089] Determine the main measurement line, deploy seismographs on the main measurement line, take the positions of the seismographs as the main measurement points, perform position coding on the main measurement points according to the plane coordinates to obtain the position data of the main measurement points, set auxiliary measurement lines parallel to the main measurement line, deploy fiber optic sensors on the auxiliary measurement lines, use an optical cable to connect the fiber optic sensors in series to form a fiber optic sensor array, take each fiber optic sensor as an auxiliary measurement point, and perform position coding on the auxiliary measurement points according to the plane coordinates to obtain the position data of the auxiliary measurement points; the auxiliary measurement lines include a first auxiliary measurement line and a second auxiliary measurement line; the main measurement points and the auxiliary measurement points jointly form a SPAC array; temperature sensors are also deployed on the auxiliary measurement points;

[0090] In the actual evaluation, the surface vibration detection system SPAC array is as Figure 2 shown.

[0091] In this embodiment, the method for obtaining the temperature-compensated electrical signal through edge calibration includes:

[0092] Obtain the ambient temperature of multiple groups of fiber optic sensors and the corresponding fiber optic electrical signals, and use a multi-layer perceptron model to establish a temperature-electrical signal deviation model;

[0093] Input the ambient temperature of the fiber optic sensor into the temperature-electrical signal deviation model to obtain the electrical signal deviation, and correct the fiber optic electrical signal according to the electrical signal deviation to obtain the temperature-compensated electrical signal;

[0094] In the actual evaluation, the control variable method is used to ensure that the optical fiber is in a stress-free state and a constant vibration state, obtain the fiber optic electrical signals of the fiber optic sensor at different ambient temperatures, and use a multi-layer perceptron model to automatically learn the relationship between the fiber optic electrical signal and the temperature to establish a temperature-electrical signal deviation model; the multi-layer perceptron model includes a polynomial regression function, a mean square error loss function, and a gradient descent algorithm;

[0095] Taking the edge calibration of the optical fiber electrical signal of the auxiliary measurement point located at (10√3, 10) as an example, the current eigenvector of the optical fiber electrical signal within 5 specific acquisition time points is [1.2, 1.5, 1.8, 2.1, 2.4] (unit: mA), the voltage eigenvector is [0.5, 0.6, 0.7, 0.8, 0.9] (unit: V), and the corresponding temperature eigenvector is [20.2, 20.5, 20.8, 20.7, 20.4] (unit: °C). Input the corresponding data into the temperature-electric signal deviation model established by the multi-layer perceptron model:

[0096] ,

[0097] ,

[0098] where is the temperature compensation current signal, is the temperature compensation voltage signal, is the optical fiber current signal, is the optical fiber voltage signal, T is the temperature at the acquisition point;

[0099] Obtain the temperature compensation current signal eigenvector [1.179, 1.479, 1.779, 2.079, 2.379] (unit: mA) and voltage eigenvector [0.489, 0.589, 0.689, 0.789, 0.889] (unit: V) of the auxiliary measurement point at the position of (10√3, 10).

[0100] In this embodiment, the method for obtaining the scale index and the process index includes:

[0101] Use the SVM support vector machine to search for the optimal hyperplane to divide the temperature compensation electrical signal into normal electrical signals and vibration electrical signals; the vibration electrical signal represents the temperature compensation electrical signal related to ground vibration;

[0102] Perform edge matching to obtain the first vibration region: Extract the position data of the auxiliary measurement point corresponding to the vibration electrical signal and the position data of the main measurement point corresponding to the node seismic wave. Use the ball tree algorithm to perform radius search for each main measurement point, find all auxiliary measurement points whose distance from the main measurement point is less than or equal to the radius, match the vibration electrical signal of the auxiliary measurement point that meets the search conditions with the node seismic wave of the corresponding main measurement point, and determine the first vibration region according to the position data of the matching auxiliary measurement point; the vibration electrical signal can be matched with multiple node seismic waves at the same time;

[0103] Determine the monitoring scale characteristics according to the position data of the auxiliary measurement point corresponding to the vibration electrical signal and the edge matching result, and input the monitoring scale characteristics into the scale influence function to obtain the scale index. The expression is:

[0104] ,

[0105] wherein is the scale index, is the spatial coverage weight, is the measurement point redundancy weight, is the spatial dispersion weight, is the current monitoring area, is the standard monitoring area for which the monitoring data of the sensing network is valid this time, is the length of the main measurement line, is the maximum monitoring length of the sensing network, is the number of auxiliary measurement points, is the number of main measurement points, is the number of auxiliary measurement points for which the monitoring data of the sensing network is valid this time, is the average distance between auxiliary measurement points, is the maximum distance between the main measurement point and the auxiliary measurement points, is the distance ratio coefficient, is the critical distance;

[0106] The process parameters of the optical fiber are extracted and input into the process influence function to obtain the process index. The expression is:

[0107] ;

[0108] wherein is the process index, is the fusion quality weight, is the production deformation weight, is the fusion overlap length, is the refractive index matching degree, is the core refractive index, is the average core refractive index, is the effective aperture ratio, is the effective transmission aperture, is the theoretical core area ratio, , is the core area constant, , is the core diameter deviation, is the etching depth ratio, is the etching depth, is the designed thickness, is the doping diffusion gradient, is the extrusion deformation rate, is the etching constant;

[0109] In the actual evaluation, the monitoring scale features include the main survey line length, monitoring area, number of main measurement points, number of auxiliary measurement points, maximum distance between the main measurement point and the auxiliary measurement points, and average distance between the auxiliary measurement points; the main survey line length is equal to the sum of the distance between the two edge main measurement points and the monitoring diameter of the main measurement point; the monitoring area is equal to the area of the closed figure formed by the edge auxiliary measurement points; the process parameters include the optical fiber fusion length, extrusion degree of the optical fiber production, etching depth, refractive index, aperture, and core diameter;

[0110] The SVM is used to screen the temperature compensation electrical signals of the optical fiber sensors on the optical fiber sensing matrix to obtain the vibration electrical signals. Taking the main measurement points at the positions of (0,0) and (140√3,0) as examples, the corresponding auxiliary measurement points are matched by the ball tree algorithm as [(-10√3,10),(10√3,10),(0,-10),(0,-20),(-20√3,-20),(20√3,-20),(0,40)] and [(130√3,10),(150√3,10),(140√3,-10),(140√3,-20),(120√3,-20),(160√3,-20),(140√3,40)]. Among them, the temperature compensation current signals at the corresponding positions of the following auxiliary measurement points [(-10√3,10),(10√3,10),(0,-20),(-20√3,-20),(0,40)] and [(150√3,10),(140√3,-10),(120√3,-20),(160√3,-20),(140√3,40)] are vibration electrical signals. Therefore, the auxiliary measurement points matched to the edge of the main measurement point at the position of (0,0) are [(-10√3,10),(10√3,10),(0,-20),(-20√3,-20),(0,40)], and the auxiliary measurement points matched to the edge of the main measurement point at the position of (140√3,0) are [(150√3,10),(140√3,-10),(120√3,-20),(160√3,-20),(140√3,40)];

[0111] The monitoring scale features are statistically obtained according to the edge matching results: the current monitoring area 、the main survey line length 、the number of auxiliary measurement points 、the number of main measurement points 、the average distance between the auxiliary measurement points 、the maximum distance between the main measurement point and the auxiliary measurement points ; Take the spatial coverage weight = 0.4, the measurement point redundancy weight = 0.3, the spatial dispersion weight = 0.3, the standard monitoring area 、the maximum monitoring length 、the number of effective auxiliary measurement points , distance ratio coefficient , critical distance , the scale index is calculated ;

[0112] Obtain the optical fiber process parameters: fusion overlap length , refractive index matching degree , effective aperture ratio , core diameter deviation (unit: μm), etching depth ratio , doping diffusion gradient (unit: wtz% / μm), extrusion deformation rate ; Take the fusion quality weight = 0.6, manufacturing deformation weight = 0.4, core area constant , , etching constant , the process index is calculated .

[0113] In this embodiment, the method for obtaining the micro-vibration characteristics of the optical fiber by performing edge processing on the vibration electrical signal includes:

[0114] Introduce an adaptive criterion to perform the first empirical mode decomposition on the vibration electrical signal until the local energy ratio of the residual signal is less than a preset threshold , and stop when K modal signals and a residual signal are obtained. The stop determination criterion expression for the first empirical mode decomposition is:

[0115] ;

[0116] where is the vibration electrical signal at k time, is the residual signal at k time, The local energy ratio is calculated using a sliding window, t is the calculation time, is the window width, , is the signal sampling rate, ;

[0117] Adopt a multi-scale decomposition strategy to perform the second empirical mode decomposition on the absolute value of each , and each obtains N modal signals and a residual signal , the multi-scale decomposition strategy expression is:

[0118] ,

[0119] where is t the current signal at time the i th obtained by the second empirical mode decomposition of the j th , is the absolute value of the i th , is the j th scale parameter;

[0120] The sliding window is used to calculate the mutual information of all signal components and the vibration electrical signal , and the signal components with mutual information greater than the dynamic threshold of mutual information are selected to obtain the effective information. The expression of the dynamic threshold of mutual information is:

[0121] ,

[0122] ,

[0123] where is the dynamic threshold of mutual information, , , are adjustment coefficients, is the entropy of the original signal, is the average mutual information, is the standard deviation of the mutual information;

[0124] According to the screening results, the real signal is subjected to Hilbert transform to obtain the complex analytical signal of the vibration electrical signal expression:

[0125] ,

[0126] where j is the imaginary unit, is the Hilbert transform of the vibration electrical signal , P is the Cauchy principal value at this transform, is the integration variable;

[0127] According to each signal component , residual signal and Determine the energy weight of the energy determination signal, and the expression is:

[0128] ,

[0129] ,

[0130] ,

[0131] where is the weight coefficient of the signal component , is the weight coefficient of the residual signal , is the weight coefficient of the residual signal , is the energy of the signal component , is the energy of the residual signal , is the energy of the residual signal ;

[0132] Calculate the instantaneous amplitude of the vibration electrical signal and the instantaneous phase of the vibration electrical signal:

[0133] ,

[0134] ,

[0135] For the instantaneous phase of the vibration electrical signal, obtain the instantaneous frequency of the vibration electrical signal by taking the reciprocal of time. The fiber optic micro-vibration characteristics are composed of the instantaneous amplitude , instantaneous phase and instantaneous frequency of the vibration electrical signal;

[0136] In actual evaluation, taking the data inversion of the main measurement point node vibration wave at the positions of (0,0) and (140√3,0) as an example, the inversion vibration characteristics are obtained: the corresponding vibration intensities are 1.5 m / s 2 and 1.8 m / s 2 , the vibration area is (-40,-40)-(-40,40)-(140√3 + 40,40)-(140√3 + 40,-40), the vibration depth is 200 m, and the vibration type is shallow surface vibration);

[0137] Taking the edge processing of the vibration electrical signal of the auxiliary measurement point with the warm-up current signal feature vector [1.179, 1.479, 1.779, 2.079, 2.379] (unit: mA) and the voltage feature vector [0.489, 0.589, 0.689, 0.789, 0.889] (unit: V) at (10√3, 10) as an example, when performing the first empirical mode decomposition, the signal sampling rate is taken as 10000 to calculate the preset threshold , the window width is taken as 20, and the vibration electrical signal is decomposed to obtain K = 3 modal signals and a residual signal , when performing the second empirical mode decomposition, each is decomposed to obtain modal signals and a residual signal , taking the adjustment coefficients , , , the average mutual information , the standard deviation of the mutual information , calculate the dynamic threshold of the mutual information , screen out the signal components with mutual information greater than 0.26 as effective information, perform Hilbert transform on the real signal according to the screening result to obtain the vibration electrical signal complex analytic signal, and calculate the instantaneous amplitude of the vibration electrical signal at t = 0.1 s , instantaneous phase and instantaneous frequency ;

[0138] Perform edge processing on the vibration electrical signals of all the above-mentioned edge-matched auxiliary measurement points to obtain the corresponding optical fiber micro-motion characteristics.

[0139] In this embodiment, the method for obtaining the detection result includes:

[0140] Form a comprehensive data set with the optical fiber micro-motion characteristics, historical vibration data, the first vibration area, the inverted vibration characteristics, the scale index, and the process index, and use the random forest algorithm to divide the comprehensive data set into a training set and a test set according to 7:3; the historical vibration data includes historical vibration intensity, historical vibration orientation, and historical vibration type; the historical vibration orientation includes historical vibration depth and historical vibration area;

[0141] Construct a micro-motion optical fiber fusion detection model, and the micro-motion optical fiber fusion detection model specifically includes an input layer, a base model layer, a strategy layer, and an output layer; the base model layer includes a BP neural network, a STAN network, and an LSTM network;

[0142] The BP neural network is used to learn the non - linear coupling relationship between the instantaneous amplitude of the vibration electrical signal, the instantaneous frequency of the vibration electrical signal, the scale index, the process index, and the historical vibration intensity, predict the surface vibration intensity, use the self - attention mechanism to assign dynamic weights to key features, and use the mean square error to evaluate the difference between the predicted value and the true value of the surface vibration intensity;

[0143] The STAN network is used to learn the spatio - temporal relationship between the instantaneous phase of the vibration electrical signal, the first vibration region, and the historical vibration region, predict the surface vibration region, use the spatio - temporal attention mechanism to focus on the time mutation points and spatial distribution characteristics of the phase data, and use the cross - entropy loss to evaluate the difference between the predicted value and the true value of the surface vibration region;

[0144] The LSTM network is used to learn the temporal relationship between the fine - movement characteristics, the scale index, the process index, and the historical vibration type, predict the surface vibration type, use the bidirectional LSTM to capture the temporal information before and after, and use the cross - entropy loss to evaluate the difference between the predicted value and the true value of the surface vibration type;

[0145] The strategy layer weighted - fuses the predicted value of the surface vibration intensity of the BP neural network and the inverted vibration intensity based on the distance weight, uses the voting mechanism to fuse the predicted value of the surface vibration of the STAN network and the inverted vibration azimuth, weighted - fuses the prediction of the surface vibration type of the LSTM network and the inverted vibration type according to the number of measuring points, and connects to the fully - connected layer of the output layer to output the detection result; the detection result includes the vibration detection intensity, the vibration detection azimuth, and the vibration detection type; the distance weight is inversely proportional to the distance between the main measuring point and the auxiliary measuring point;

[0146] The Adam optimizer is used to optimize the network structure, the training set is used to train the model, and the test set is used to evaluate the accuracy of the model;

[0147] The perception data to be detected is input into the micro - motion optical fiber fusion detection model to obtain the detection result;

[0148] In the actual evaluation, the perception data of 7 main measuring points and the matching auxiliary measuring points are input into the micro - motion optical fiber fusion detection model: taking the main measuring point at the position (0,0) as an example, the predicted values of the vibration intensity output by the BP neural network for the auxiliary measuring points at the positions (- 10√3,10), (10√3,10), (0, - 20), (- 20√3, - 20), (0,40) are 1.2m / s 2 、1.4m / s 2 、1.6m / s 2 、1.7m / s 2 、1.5m / s 2, according to the distance weight = the distance between the auxiliary measurement point and the main measurement point / the sum of the distances between all auxiliary measurement points and the main measurement point, calculate the distance weights to be 0.1429, 0.1429, 0.1429, 0.2857, 0.2857 respectively, and calculate the comprehensive vibration intensity of the auxiliary measurement point to be 1.5144 m / s according to the distance weight. 2 , the vibration detection intensity of the main measurement point at the (0,0) position is obtained by averaging and weighting the comprehensive vibration intensity and the inversion vibration intensity to be 1.5072 m / s. 2 , thus, the vibration intensities of the 7 main measurement points on the main measurement line are 1.5072 m / s 2 、1.5526 m / s 2 、1.6122 m / s 2 、1.6598 m / s 2 、1.7055 m / s 2 、1.7614 m / s 2 、1.8033 m / s 2；

[0149] The predicted vibration area of the 25 auxiliary measurement points through the STAN network is (-20√3, -20)-(0, 40)-(140√3, 40)-(160√3 + 20, -20). By using the strategy layer voting mechanism to determine the area edge points for the predicted vibration area and the inversion vibration area, the vibration detection area is obtained as (-20, -20√3)-(-20, 40)-(140√3 + 20, 40)-(140√3 + 20, -20), and the corresponding vibration detection depth is 200 mm;

[0150] Through the LSTM network prediction for the 25 auxiliary measurement points in this area, the prediction results are obtained (18 auxiliary measurement points predict shallow surface vibrations caused by geological activities / probability 72%, 4 auxiliary measurement points predict shallow surface vibrations caused by human activities / probability 16%, 3 auxiliary measurement points predict shallow focus earthquakes / probability 12%). The inversion vibration type is shallow surface vibration (the probabilities caused by geological activities and human activities are 82% and 18% respectively). By averaging and weighting the probabilities of the vibration types predicted by the LSTM network and the probabilities of the inversion vibration types, the following are obtained: the probability of shallow surface vibrations caused by geological activities is 77%, the probability of shallow surface vibrations caused by human activities is 17%, and the probability of shallow focus earthquakes is 6%. The output vibration type is shallow surface vibrations caused by geological activities.

[0151] In the second aspect, a micro - motion optical fiber fusion detection system includes:

[0152] A sensing network module: constructing a micro - motion optical fiber fusion sensing network to obtain historical sensing data and position data; the micro - motion optical fiber fusion sensing network includes a node instrument module, an optical fiber module, and a temperature sensor module; the optical fiber module includes an optical cable, an optical fiber sensor, and an optical fiber demodulator;

[0153] Edge computing module: used to perform edge calibration to obtain a temperature-compensated electrical signal; used to perform edge screening on the temperature-compensated electrical signal to obtain a vibration electrical signal, perform edge matching on the vibration electrical signal and the node seismic wave according to position data and determine the monitoring scale characteristics, and input the monitoring scale characteristics and the optical fiber process parameters into the influence function respectively to obtain a scale index and a process index; used to perform edge processing on the vibration electrical signal to obtain the optical fiber micro-motion characteristics;

[0154] Cloud model module: used to construct a micro-motion optical fiber fusion detection model according to the optical fiber micro-motion characteristics, and input the data to be detected and sensed into the micro-motion optical fiber fusion detection model to obtain a detection result;

[0155] Intelligent supervision module: used to store, view and manage the optical fiber micro-motion characteristics, the inverted vibration characteristics, the scale index, the process index and the detection result, and perform surface vibration analysis according to the detection result.

[0156] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A micro-motion fiber optic fusion detection method, characterized in that, It includes the following steps: S1. Construct a micro-motion optical fiber fusion perception network to obtain historical perception data and position data; the historical perception data includes optical fiber electrical signals, node vibration waves, temperature, and historical vibration data; S2. Perform edge calibration on the temperature and the optical fiber electrical signal to obtain a temperature-compensated electrical signal; S3. Perform edge screening on the temperature-compensated electrical signal to obtain a vibration electrical signal, perform edge matching on the vibration electrical signal and the node seismic wave according to the position data, and determine the monitoring scale feature. Input the monitoring scale feature and the optical fiber process parameters into the influence function respectively to obtain a scale index and a process index; the influence function includes a scale influence function and a process influence function; S4. Perform data inversion on the node vibration wave to obtain the inversion vibration characteristics, and perform edge processing on the vibration electrical signal to obtain the optical fiber micro-motion characteristics; The inversion vibration characteristics include inversion vibration intensity, inversion vibration azimuth, and inversion vibration type; The inversion vibration azimuth includes an inversion vibration area and an inversion vibration depth; S5. Construct a micro-motion optical fiber fusion detection model according to the optical fiber micro-motion characteristics, and input the perception data to be detected into the micro-motion optical fiber fusion detection model to obtain a detection result; The method for obtaining the scale index and the process index includes: Use an SVM support vector machine to search for the optimal hyperplane to divide the temperature-compensated electrical signal into a normal electrical signal and a vibration electrical signal; the vibration electrical signal represents the temperature-compensated electrical signal related to ground vibration; Perform edge matching to obtain the first vibration area: extract the position data of the auxiliary measurement points corresponding to the vibration electrical signal and the position data of the main measurement points corresponding to the node seismic wave. Use the ball tree algorithm to perform radius search on each main measurement point, find all auxiliary measurement points whose distance from the main measurement point is less than or equal to the radius, match the vibration electrical signal of the auxiliary measurement points that meet the search conditions with the node seismic wave of the corresponding main measurement point, and determine the first vibration area according to the position data of the matching auxiliary measurement points; the vibration electrical signal can be matched with multiple node seismic waves at the same time; Determine the monitoring scale feature according to the position data of the auxiliary measurement points corresponding to the vibration electrical signal and the edge matching result, and input the monitoring scale feature into the scale influence function to obtain the scale index. The expression is: ， wherein is the scale index, is the spatial coverage weight, is the measurement point redundancy weight, is the spatial dispersion weight, is the current monitoring area, is the standard monitoring area for the effective monitoring data of the sensing network this time, is the length of the main measurement line, is the maximum monitoring length of the sensing network, is the number of auxiliary measurement points, is the number of main measurement points, is the number of effective auxiliary measurement points for the monitoring data of the sensing network this time, is the average distance between auxiliary measurement points, is the maximum distance between the main measurement point and the auxiliary measurement points, is the distance ratio coefficient, is the critical distance; Extract the optical fiber process parameters and input them into the process influence function to obtain the process index. The expression is: ； wherein is the process index, is the welding quality weight, is the production deformation weight, is the welding overlap length, is the refractive index matching degree, is the core refractive index, is the average core refractive index, is the effective aperture ratio, is the effective transmission aperture, is the theoretical core area ratio, and is the core area constant, , is the core diameter deviation, is the etching depth ratio, is the etching depth, is the designed thickness, is the doping diffusion gradient, is the extrusion deformation rate, is the etching constant.

2. The micro-motion optical fiber fusion detection method according to claim 1, characterized in that The method for obtaining the perception data and the position data includes: The micro-motion optical fiber fusion perception network includes a node instrument module, an optical fiber module, and a temperature sensor module; The node instrument module collects node seismic waves through a seismograph; The optical fiber module collects the interference light intensity signal of the optical fiber sensor through an optical cable, and converts the interference light intensity signal into an optical fiber electrical signal through an optical fiber demodulator; the optical cable is connected in series with multiple optical fiber sensors to form an optical fiber sensor array; The temperature sensor module is used to collect the environmental temperature of the optical fiber sensor; The position data of the micro-motion optical fiber fusion perception network includes the position data of the main measurement points and the position data of the auxiliary measurement points. The specific obtaining method includes: Determine the main survey line, deploy seismographs on the main survey line, take the positions of the seismographs as the main measurement points, obtain the position data of the main measurement points by encoding the positions of the main measurement points according to the plane coordinates, set up auxiliary survey lines parallel to the main survey line, deploy fiber optic sensors on the auxiliary survey lines, and use optical cables to connect the fiber optic sensors in series to form a fiber optic sensor array. Take each fiber optic sensor as an auxiliary measurement point, and obtain the position data of the auxiliary measurement points by encoding the positions of the auxiliary measurement points according to the plane coordinates; the auxiliary survey lines include a first auxiliary survey line and a second auxiliary survey line; the main measurement points and the auxiliary measurement points jointly form a SPAC array; temperature sensors are also deployed on the auxiliary measurement points.

3. The method for detecting micro-vibration fiber fusion according to claim 1, characterized in that The method for obtaining the temperature-compensated electrical signal through edge calibration includes: Obtain the ambient temperatures and corresponding fiber optic electrical signals of multiple groups of fiber optic sensors, and establish a temperature-electrical signal deviation model using a multi-layer perceptron model; Input the ambient temperature of the fiber optic sensor into the temperature-electrical signal deviation model to obtain the electrical signal deviation, and correct the fiber optic electrical signal according to the electrical signal deviation to obtain the temperature-compensated electrical signal.

4. The micro-motion optical fiber fusion detection method according to claim 1, wherein The method for obtaining the fiber optic micro-vibration characteristics by performing edge processing on the vibration electrical signal includes: Introduce an adaptive criterion to the vibration electrical signal Perform the first empirical mode decomposition until the residual signal has a local energy ratio less than a preset threshold and then stop. The decomposition obtains K modal signals and a residual signal . The expression for the stopping criterion of the first empirical mode decomposition is: ； Among them is k the vibration electrical signal at the moment, is k the residual signal at the moment, The local energy ratio is calculated using a sliding window, t is the calculation moment, is the window width, , is the signal sampling rate, ; The multi-scale decomposition strategy is adopted for each absolute value to perform the second empirical mode decomposition, and each obtains N modal signals and a residual signal , and the expression of the multi-scale decomposition strategy is: ， Among them is t the instantaneous current signal of the i th obtained by the second empirical mode decomposition j th , is the absolute value of the i th , and is the j th scale parameter; Calculate all using a sliding window signal components and vibration electrical signals of the mutual information , screen the signal components with mutual information greater than the dynamic mutual information threshold to obtain effective information. The expression for the dynamic mutual information threshold is: ， ， where is the dynamic threshold of mutual information, , , are adjustment coefficients, is the entropy of the original signal, is the average mutual information, is the standard deviation of mutual information; Perform Hilbert transform on the real signal according to the screening result to obtain the complex analytic signal of the vibration electrical signal Expression: ， wherein j is the imaginary unit, is the vibration electrical signal is the Hilbert transform of, P is the Cauchy principal value at this transform, is the integration variable; According to each signal component , residual signal and to determine the signal energy weight, the expression is: ， ， ， wherein is the weight coefficient of the signal component ; is the weight coefficient of the residual signal ; is the weight coefficient of the residual signal ; is the energy of the signal component ; is the energy of the residual signal ; is the energy of the residual signal ; Calculating the instantaneous amplitude of the vibration electrical signal and the instantaneous phase of the vibration electrical signal : ， ， For the instantaneous phase of the vibration electrical signal Obtain the instantaneous frequency of the vibration electrical signal by taking the reciprocal of time , from the instantaneous amplitude of the vibration electrical signal , instantaneous phase and instantaneous frequency constitute the optical fiber micro - motion characteristics.

5. The micro-motion optical fiber fusion detection method according to claim 1, characterized in that, The method for obtaining the detection result includes: Form a comprehensive data set by combining the fiber optic micro-vibration characteristics, historical vibration data, the first vibration region, the inverted vibration characteristics, the scale index, and the process index, and use the random forest algorithm to divide the comprehensive data set into a training set and a test set according to 7:3; the historical vibration data includes historical vibration intensity, historical vibration azimuth, and historical vibration type; the historical vibration azimuth includes historical vibration depth and historical vibration region; Construct a micro-vibration fiber fusion detection model, which specifically includes an input layer, a base model layer, a strategy layer, and an output layer; the base model layer includes a BP neural network, a STAN network, and an LSTM network; The BP neural network is used to learn the non-linear coupling relationship between the instantaneous amplitude of the vibration electrical signal, the instantaneous frequency of the vibration electrical signal, the scale index, the process index, and the historical vibration intensity, predict the surface vibration intensity, assign dynamic weights to key features using the self-attention mechanism, and use the mean square error to evaluate the difference between the predicted value and the true value of the surface vibration intensity; The STAN network is used to learn the spatio-temporal relationship between the instantaneous phase of the vibration electrical signal, the first vibration region, and the historical vibration region, predict the surface vibration region, use the spatio-temporal attention mechanism to focus on the time mutation points and spatial distribution characteristics of the phase data, and use the cross-entropy loss to evaluate the difference between the predicted value and the true value of the surface vibration region; The LSTM network is used to learn the temporal relationship between the fiber optic micro-vibration characteristics, the scale index, the process index, and the historical vibration type, predict the surface vibration type, use bidirectional LSTM to capture the front and back temporal information, and use the cross-entropy loss to evaluate the difference between the predicted value and the true value of the surface vibration type; The strategy layer weighted-fuses the predicted value of the surface vibration intensity and the inverted vibration intensity of the BP neural network based on distance weights, fuses the predicted value of the surface vibration of the STAN network and the inverted vibration azimuth using a voting mechanism, weighted-fuses the prediction of the surface vibration type of the LSTM network and the inverted vibration type according to the number of measurement points, and connects to the fully connected layer of the output layer to output the detection result; the detection result includes the vibration detection intensity, the vibration detection azimuth, and the vibration detection type; the distance weight is inversely proportional to the distance between the main measurement point and the auxiliary measurement point; The Adam optimizer is used to optimize the network structure, the training set is used to train the model, and the test set is used to evaluate the accuracy of the model; The to-be-detected sensing data is input into the micro-vibration optical fiber fusion detection model to obtain the detection result.

6. A micro-motion optical fiber fusion detection system for performing the method according to any one of claims 1-5, characterized in that It includes: Sensing network module: Construct a micro-vibration optical fiber fusion sensing network to obtain historical sensing data and position data; the micro-vibration optical fiber fusion sensing network includes a node instrument module, an optical fiber module, and a temperature sensor module; the optical fiber module includes an optical cable, an optical fiber sensor, and an optical fiber demodulator; Edge computing module: Used to perform edge calibration to obtain a temperature-compensated electrical signal; used to perform edge screening on the temperature-compensated electrical signal to obtain a vibration electrical signal, perform edge matching on the vibration electrical signal and the node seismic wave according to the position data and determine the monitoring scale feature, and input the monitoring scale feature and the optical fiber process parameter into the influence function to obtain the scale index and the process index; Used to perform edge processing on the vibration electrical signal to obtain the micro-vibration optical fiber feature; Cloud model module: Used to construct a micro-vibration optical fiber fusion detection model according to the micro-vibration optical fiber feature, and input the to-be-detected sensing data into the micro-vibration optical fiber fusion detection model to obtain the detection result; Intelligent supervision module: Used to store, view, and manage the micro-vibration optical fiber feature, the inverted vibration feature, the scale index, the process index, and the detection result, and perform surface vibration analysis according to the detection result.

Images