Converter station vector seismic state identification method and system based on distributed optical fibers

Abstract

The invention relates to the field of seismic state recognition, and provides a converter station vector seismic state recognition method and system based on distributed optical fibers, and the method comprises the steps: collecting multi-dimensional vibration signal data of a converter station, and obtaining seismic monitoring data through processing; performing phase demodulation processing, time difference analysis and space vector decomposition algorithm processing on the seismic monitoring data to obtain seismic wave propagation vector characteristic data, and performing characteristic integration and optimization to obtain a seismic wave comprehensive state characteristic data set; identifying and classifying the seismic wave comprehensive state feature data set to obtain a seismic wave form classification result; performing analysis based on the seismic wave form classification result to obtain a converter station seismic state evaluation result; and carrying out comparative analysis on the evaluation result and a threshold value, generating earthquake early warning information of the converter station, and providing a corresponding emergency response suggestion according to the earthquake early warning information. According to the invention, accurate identification and classification of seismic waves are realized, and the earthquake safety early warning capability of the converter station is improved.

Description

Technical Field

[0001] The present invention relates to the field of earthquake state identification, and in particular to a method and system for identifying vector earthquake states of converter stations based on distributed optical fibers. Background Art

[0002] Converter station vector seismic state recognition is a technology that uses distributed sensing technology to monitor and warn of earthquakes in converter stations and their surrounding areas. A distributed fiber-optic sensing system deployed around the converter station collects ground vibration signals in real time and converts these multi-dimensional vibration signals into vector features of seismic wave propagation, thereby accurately identifying the direction, intensity, and type of seismic wave propagation. The basic principle is to utilize the sensitivity of fiber Bragg gratings to mechanical strain. Through phase modulation and demodulation techniques, the surface vibrations caused by seismic waves are converted into measurable optical signal changes. Signal processing and feature extraction algorithms are then used to achieve real-time monitoring and assessment of seismic conditions. Converter station seismic state recognition methods have limited ability to suppress background noise and are particularly susceptible to equipment operating noise and external interference in complex environments, resulting in reduced recognition accuracy. The lack of systematic extraction of seismic wave vector features makes it impossible to effectively describe the propagation direction, amplitude changes, and spatial distribution characteristics of seismic waves.

[0003] The existing technology adopts the principle of on-site and regional monitoring, deploys on-site stations at the center, identifies P-wave events through waveform analysis, and uses a state recognition method to trigger an alarm mechanism through a single station / dual station. However, this method mainly relies on vibration data from a single point or a small number of discrete points, which makes it difficult to accurately capture the changes in the vector characteristics of seismic waves during spatial propagation. As a result, the accuracy of earthquake state recognition is limited and the anti-interference ability is weak. Summary of the Invention

[0004] In view of this, the present invention proposes a method and system for vector earthquake state identification of converter stations based on distributed optical fiber, which solves the problem that the existing technology mainly relies on vibration data of a single point or a small number of discrete points, making it difficult to accurately capture the vector characteristic changes of earthquake waves during spatial propagation, resulting in limited earthquake state identification accuracy and weak anti-interference ability.

[0005] The technical solution of the present invention is implemented as follows: In a first aspect, the present invention provides a method for identifying vector earthquake states of converter stations based on distributed optical fibers, comprising the following steps:

[0006] Collecting multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and performing adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data;

[0007] Performing phase demodulation processing and time difference analysis on the seismic monitoring data, and processing the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data;

[0008] The seismic wave propagation vector feature data is integrated and optimized by a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set;

[0009] Based on the deep learning model, seismic wave identification and classification are performed on the seismic wave comprehensive state feature dataset to obtain the seismic wave morphology classification results;

[0010] Based on the seismic wave morphology classification results and seismic wave propagation vector characteristic data, combined with the earthquake intensity model, the seismic status assessment results of the converter station are obtained;

[0011] Compare and analyze the converter station earthquake status assessment result with the earthquake early warning threshold to generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

[0012] On the basis of the above technical solution, preferably, the multi-dimensional vibration signal data of the converter station is collected by a distributed optical fiber sensor, and the multi-dimensional vibration signal data is subjected to adaptive filtering and environmental noise suppression processing to obtain earthquake monitoring data, which specifically includes:

[0013] Deploying a distributed optical fiber sensor network around the converter station according to a preset spatial distribution scheme, wherein the distributed optical fiber sensor network includes optical fiber sensors deployed along three orthogonal directions of the X-axis, the Y-axis, and the Z-axis, and collecting multi-dimensional vibration signal data through the distributed optical fiber sensor network;

[0014] Adaptive filtering is performed on the multidimensional vibration signal data to obtain a multidimensional filtered vibration signal, and an environmental noise suppression algorithm is used to eliminate environmental noise interference in the multidimensional filtered vibration signal to obtain earthquake monitoring data.

[0015] On the basis of the above technical solution, preferably, it further includes:

[0016] A key equipment monitoring system is constructed by deploying dedicated distributed optical fiber sensor arrays on key equipment in a converter station, wherein the key equipment includes transformers, converter valves, filter devices, and connection equipment; the dedicated distributed optical fiber sensor arrays are densely deployed in a grid or ring manner on the core structural parts of the key equipment;

[0017] The key equipment monitoring system is specifically used for:

[0018] Collect vibration characteristic data of key equipment under normal operating conditions and establish a baseline model of equipment vibration characteristics;

[0019] Real-time monitoring of high-frequency vibration signals of key equipment, using higher sampling frequency and spatial resolution than the main system to capture tiny vibration changes;

[0020] Combine time-frequency analysis methods to extract the deformation characteristics of key equipment under earthquake action and distinguish the natural vibration of equipment from abnormal vibration caused by earthquake;

[0021] Establish digital twin models of key equipment based on finite element analysis, match measured vibration data with theoretical models in real time, and calculate stress distribution and potential damage levels of key equipment;

[0022] Correlate and analyze key equipment monitoring data with seismic wave propagation vector feature data acquired by the main system to determine the impact path and damage mode of seismic waves on key equipment;

[0023] Provide key equipment status information to the status assessment module as a decision basis for generating earthquake status assessment results and equipment-level emergency response recommendations.

[0024] On the basis of the above technical solution, preferably, the phase demodulation processing and time difference analysis of the seismic monitoring data are performed, and the data are processed by a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data, which specifically includes:

[0025] Based on the principle of phase-sensitive optical time-domain reflectometry, the seismic monitoring data is phase-demodulated and processed, and the time difference between the arrival of the seismic waves at each monitoring point is analyzed using the time difference positioning method to obtain preliminary demodulated data;

[0026] The spatial vector decomposition algorithm is used to decompose the preliminary demodulated data into three-dimensional spatial components, extract the propagation characteristics of seismic waves in the X-axis, Y-axis and Z-axis directions, and obtain the seismic wave propagation vector characteristic data.

[0027] On the basis of the above technical solution, preferably, the feature integration and optimization of the seismic wave propagation vector feature data by a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set specifically includes:

[0028] Perform feature extraction and feature selection on the characteristic data of seismic wave propagation vectors, obtain key characteristic parameters of seismic waves based on time-frequency domain analysis methods, and construct initial characteristic vectors based on the key characteristic parameters;

[0029] The vector feature fusion algorithm is used to perform multi-scale feature fusion and optimization processing on the initial feature vector to eliminate the feature redundancy of the initial feature vector and obtain the comprehensive state feature dataset of seismic waves.

[0030] On the basis of the above technical solution, preferably, the seismic wave identification and seismic wave classification are performed on the seismic wave comprehensive state feature dataset based on the deep learning model to obtain the seismic wave morphology classification result, which specifically includes:

[0031] A deep learning model based on convolutional neural networks and long short-term memory networks was constructed to perform multimodal feature fusion on the seismic wave comprehensive state feature dataset to generate preliminary classification features;

[0032] The calculation formula of the multimodal feature fusion is:

[0033]

[0034] Among them, F fused is the preliminary classification feature, F CNN and F LSTM are the feature matrices output by the convolutional neural network and the long short-term memory network, respectively, and α1 is F CNN and F LSTM The balance weight coefficient between them, β1 is the interaction feature weight coefficient, For element-by-element multiplication, Cov(F CNN ,F LSTM ) is F CNN and F LSTM The covariance of Var(F CNN ) is F CNN The variance of Var(F LSTM ) is F LSTM variance;

[0035] The initial classification features are dynamically weighted through the attention mechanism, and the Softmax classifier is used to classify the waveform features of P waves and S waves, and the seismic wave morphology classification results are output;

[0036] The calculation formula for dynamically weighting the preliminary classification features through the attention mechanism is:

[0037]

[0038] Among them, W attention is the attention weight matrix, Q is the query matrix, K is the key matrix, d K is the dimension of the key matrix, V is the value matrix, and σ(·) is the softmax normalization function.

[0039] On the basis of the above technical solution, preferably, the analysis based on the seismic wave morphology classification result and the seismic wave propagation vector characteristic data is combined with the seismic intensity model to obtain the converter station seismic state assessment result, which specifically includes:

[0040] Based on the arrival time difference and amplitude characteristics of the P and S waves in the seismic wave morphology classification results, combined with the basis functions of the earthquake intensity model, the preliminary intensity assessment value of the converter station is calculated;

[0041] The calculation formula for the preliminary intensity assessment value is:

[0042]

[0043] Among them, I base is the preliminary intensity assessment value, Δt is the arrival time difference between the P wave and the S wave, A p is the measured value of the amplitude of the P wave, A s is the measured value of the amplitude of the S wave, k1 is the time difference proportional coefficient, k2 is the amplitude proportional coefficient, E soil is the elastic modulus of the soil layer, E rock is the elastic modulus of bedrock, ε is the reference attenuation coefficient, G shear is the shear modulus of the soil layer, G rock is the shear modulus of bedrock;

[0044] Based on the propagation direction and anisotropy index of the seismic wave propagation vector characteristic data, the preliminary intensity assessment value is dynamically corrected in terms of directional intensity to generate the earthquake state assessment result;

[0045] The calculation formula for the dynamic correction of directional intensity is:

[0046]

[0047] Among them, I final is the revised earthquake intensity assessment value, η1 is the empirical correction factor, is the three-dimensional propagation velocity vector, θ is the angle between the propagation direction and the layout direction of the main equipment of the converter station, D aniso is the seismic wave anisotropy index, D threshold is the seismic wave anisotropy index threshold, σ x , σ y , σ z is the standard deviation of the propagation velocity in three-dimensional orthogonal directions.

[0048] Based on the above technical solution, preferably, the comparing and analyzing the earthquake status assessment result of the converter station with the earthquake early warning threshold to generate earthquake early warning information of the converter station, and proposing corresponding emergency response suggestions based on the earthquake early warning information, specifically includes:

[0049] Compare and analyze the converter station earthquake status assessment results with the preset earthquake warning thresholds in a hierarchical manner to generate an earthquake warning level;

[0050] According to the earthquake warning level, combined with the operating status of the converter station equipment and the emergency response plan, corresponding emergency response recommendations are generated.

[0051] In a second aspect, the present invention further provides a vector earthquake state identification system for a converter station based on distributed optical fiber, the system comprising:

[0052] A data monitoring module is used to collect multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and perform adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data;

[0053] a processing and analysis module, configured to perform phase demodulation processing and time difference analysis on the seismic monitoring data, and process the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data;

[0054] A feature integration module is used to integrate and optimize the seismic wave propagation vector feature data using a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set;

[0055] The morphological classification module is used to identify and classify seismic waves based on the comprehensive state feature dataset of seismic waves based on the deep learning model to obtain the seismic wave morphological classification results;

[0056] The status assessment module is used to analyze the seismic wave morphology classification results and seismic wave propagation vector characteristic data in combination with the earthquake intensity model to obtain the seismic status assessment results of the converter station;

[0057] The early warning response module is used to compare and analyze the earthquake status assessment result of the converter station with the earthquake early warning threshold, generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

[0058] Based on the above technical solution, preferably, the system further includes a converter station topology monitoring module, which is used to monitor the vibration state of the main structural components inside the converter station using distributed optical fiber sensors. The main structural components include support towers, current-carrying circuits and connection equipment;

[0059] The converter station topology monitoring module is specifically used for:

[0060] Collect and record the vibration characteristic data of main structural components under normal operation and earthquake conditions;

[0061] Establish a topological relationship model of the main structural components and analyze the vibration energy transmission path and mutual influence relationship between the structural components;

[0062] Combined with the characteristic data of seismic wave propagation vectors, the direction and intensity of the force exerted by seismic waves on major structural components are calculated in real time;

[0063] Predict the stress and deformation degree and potential cascading failure paths of major structural components based on deep learning models;

[0064] Compare the vibration status assessment results of major structural components with safety thresholds to identify the key structural components that are first affected by the earthquake and predict the chain reactions that may be triggered by the earthquake;

[0065] Provide structural component vibration status information to the status assessment module and early warning response module as a supplementary decision-making basis for earthquake status assessment and emergency response recommendations.

[0066] The method and system for identifying vector earthquake states in converter stations based on distributed optical fibers of the present invention have the following beneficial effects compared to the prior art:

[0067] (1) By collecting multi-dimensional vibration signal data through distributed fiber optic sensors and combining adaptive filtering, space vector decomposition, vector feature fusion and deep learning models, accurate identification and classification of seismic waves are achieved. At the same time, combined with the earthquake intensity model and dynamic early warning mechanism, earthquake status assessment and early warning information for the converter station are generated, providing the converter station with a high-precision, real-time and excellent anti-interference earthquake monitoring and emergency response solution, thereby improving the earthquake safety protection capability of the converter station;

[0068] (2) By taking the seismic wave morphology classification results and seismic wave propagation vector characteristic data as input parameters and combining them with the earthquake intensity model for comprehensive analysis, not only the type characteristics of seismic waves are considered, but also the vector characteristics of seismic waves in the process of spatial propagation are integrated, thereby achieving a multi-dimensional and high-precision assessment of the seismic state of the converter station, avoiding the limitations of the traditional single parameter evaluation method, and improving the accuracy of seismic state assessment;

[0069] (3) By dynamically comparing and analyzing the earthquake status assessment results of the converter station with the preset earthquake warning threshold, a multi-level warning and differentiated response mechanism was established. According to different levels of warning status, corresponding warning information was automatically generated, and targeted emergency response suggestions were provided. This not only improved the timeliness of the warning, but also provided reasonable emergency response plans based on actual conditions, reduced the safety risks of the converter station in earthquake events, and improved the operability of the emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0071] Figure 1 This is a flow chart of a method for identifying vector earthquake states in a converter station based on distributed optical fiber according to the present invention;

[0072] Figure 2 This is a structural diagram of a converter station vector earthquake state identification system based on distributed optical fiber according to the present invention. DETAILED DESCRIPTION

[0073] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] See also Figure 1 The present invention provides a method for identifying the vector earthquake state of a converter station based on distributed optical fiber, comprising the following steps:

[0075] Collecting multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and performing adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data;

[0076] Performing phase demodulation processing and time difference analysis on the seismic monitoring data, and processing the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data;

[0077] The seismic wave propagation vector feature data is integrated and optimized by a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set;

[0078] Based on the deep learning model, seismic wave identification and classification are performed on the seismic wave comprehensive state feature dataset to obtain the seismic wave morphology classification results;

[0079] Based on the seismic wave morphology classification results and seismic wave propagation vector characteristic data, combined with the earthquake intensity model, the seismic status assessment results of the converter station are obtained;

[0080] Compare and analyze the converter station earthquake status assessment result with the earthquake early warning threshold to generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

[0081] Specifically, this embodiment collects multi-dimensional vibration signal data through distributed fiber optic sensors, and combines adaptive filtering, spatial vector decomposition, vector feature fusion and deep learning models to achieve accurate identification and classification of seismic waves. At the same time, combined with the earthquake intensity model and dynamic early warning mechanism, it generates earthquake status assessment and early warning information for the converter station, providing the converter station with a high-precision, real-time and excellent anti-interference earthquake monitoring and emergency response solution, thereby improving the earthquake safety protection capability of the converter station.

[0082] The method includes collecting multi-dimensional vibration signal data of the converter station by using a distributed optical fiber sensor, and performing adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data, specifically comprising:

[0083] Deploying a distributed optical fiber sensor network around the converter station according to a preset spatial distribution scheme, wherein the distributed optical fiber sensor network includes optical fiber sensors deployed along three orthogonal directions of the X-axis, the Y-axis, and the Z-axis, and collecting multi-dimensional vibration signal data through the distributed optical fiber sensor network;

[0084] In a specific embodiment, the preset spatial distribution scheme includes: distributing distributed optical fiber sensors in a grid pattern along the peripheral area of the converter station, with a grid spacing of 0.5-2 meters and a sampling frequency of not less than 1000 Hz, to form a closed-loop monitoring network structure;

[0085] In a specific embodiment, the distributed optical fiber sensor is a phase-sensitive distributed optical fiber sensor.

[0086] Performing adaptive filtering on the multidimensional vibration signal data to obtain a multidimensional filtered vibration signal, and using an environmental noise suppression algorithm to eliminate environmental noise interference in the multidimensional filtered vibration signal to obtain earthquake monitoring data;

[0087] In a specific embodiment, the environmental noise suppression algorithm includes a hybrid noise reduction method that combines wavelet transform denoising and adaptive Kalman filtering, and performs layered suppression of environmental noise by setting thresholds for different frequency bands.

[0088] The multi-dimensional vibration signal data includes:

[0089] X-axis vibration data: vibration signals collected by optical fiber sensors arranged along the X-axis of the converter station;

[0090] Y-axis vibration data: vibration signals collected by optical fiber sensors arranged along the Y-axis of the converter station;

[0091] Z-axis vibration data: vibration signals collected by fiber optic sensors arranged along the Z-axis (vertical direction) of the converter station.

[0092] The multidimensional filtered vibration signal includes the multidimensional filtered vibration signal after adaptive filtering and signal data after environmental noise suppression.

[0093] Specifically, this embodiment significantly reduces interference from environmental background noise, such as converter station equipment operation, wind, and rain, on seismic monitoring data by adaptively filtering and suppressing environmental noise on multidimensional vibration signal data, thereby improving the signal-to-noise ratio of seismic monitoring data. This embodiment ensures more accurate data acquisition, provides a reliable foundation for seismic wave propagation vector feature extraction, identification, and status assessment, and enhances the accuracy and robustness of overall seismic status monitoring and identification.

[0094] In a specific embodiment, a key equipment monitoring system is constructed by deploying dedicated distributed optical fiber sensor arrays on key equipment in a converter station, wherein the key equipment includes transformers, converter valves, filter devices, and connection equipment; the dedicated distributed optical fiber sensor arrays are densely deployed in a grid or ring-like manner on the core structural parts of the key equipment;

[0095] The key equipment monitoring system is specifically used for:

[0096] Collect vibration characteristic data of key equipment under normal operating conditions and establish a baseline model of equipment vibration characteristics;

[0097] Real-time monitoring of high-frequency vibration signals of key equipment, using higher sampling frequency and spatial resolution than the main system to capture tiny vibration changes;

[0098] Combine time-frequency analysis methods to extract the deformation characteristics of key equipment under earthquake action and distinguish the natural vibration of equipment from abnormal vibration caused by earthquake;

[0099] Establish digital twin models of key equipment based on finite element analysis, match measured vibration data with theoretical models in real time, and calculate stress distribution and potential damage levels of key equipment;

[0100] Correlate and analyze key equipment monitoring data with seismic wave propagation vector feature data acquired by the main system to determine the impact path and damage mode of seismic waves on key equipment;

[0101] Provide key equipment status information to the status assessment module as a decision basis for generating earthquake status assessment results and equipment-level emergency response recommendations.

[0102] The phase demodulation processing and time difference analysis are performed on the seismic monitoring data, and the processing is performed using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data, specifically including:

[0103] Based on the principle of phase-sensitive optical time-domain reflectometry, the seismic monitoring data is phase-demodulated and processed, and the time difference between the arrival of the seismic waves at each monitoring point is analyzed using the time difference positioning method to obtain preliminary demodulated data;

[0104] The preliminary demodulation data includes demodulation data based on the principle of phase-sensitive optical time-domain reflectometry and time difference positioning data (taking into account the time delay between different monitoring points).

[0105] In a specific embodiment, the time difference positioning method uses a cross-correlation algorithm to calculate the time delay of seismic waves between different monitoring points, combines the spatial layout position information of the optical fiber sensors, establishes a set of seismic wave propagation time difference equations, and solves the propagation path and velocity of the seismic waves using the least squares method;

[0106] The spatial vector decomposition algorithm is used to decompose the preliminary demodulated data into three-dimensional spatial components, extract the propagation characteristics of the seismic wave in the X-axis, Y-axis and Z-axis directions, and obtain the seismic wave propagation vector characteristic data;

[0107] In a specific embodiment, the space vector decomposition algorithm adopts a principal component analysis method to obtain the main propagation components of seismic waves in three orthogonal directions through eigenvalue decomposition, and assigns characteristic weights to the components in each direction based on energy contribution rate.

[0108] Specifically, this embodiment performs phase demodulation and time difference analysis on seismic monitoring data, and employs a spatial vector decomposition algorithm to accurately extract the propagation direction, velocity, and vector characteristics of seismic waves in space. This embodiment effectively combines the dynamic characteristics of seismic wave propagation with multidimensional spatial distribution information, providing a high-quality vector feature dataset for feature fusion and state assessment. This significantly improves the system's accuracy in identifying seismic wave morphology and propagation characteristics, achieving spatial globalization and high-resolution feature analysis for seismic monitoring.

[0109] The seismic wave propagation vector characteristic data includes the main propagation components in three orthogonal directions, the energy contribution rate of the components in each direction, and characteristic weight distribution data.

[0110] The vector feature fusion algorithm is used to integrate and optimize the seismic wave propagation vector feature data to obtain a seismic wave comprehensive state feature data set, specifically including:

[0111] Perform feature extraction and feature selection on the characteristic data of seismic wave propagation vectors, obtain key characteristic parameters of seismic waves based on time-frequency domain analysis methods, and construct initial characteristic vectors based on the key characteristic parameters;

[0112] In a specific embodiment, the key characteristic parameters include time-frequency characteristics such as amplitude, frequency, phase and energy distribution of seismic waves, and multi-resolution analysis is performed on seismic wave signals through wavelet transform to extract characteristic coefficients at different scales to form a multi-dimensional feature vector;

[0113] The vector feature fusion algorithm is used to perform multi-scale feature fusion and optimization processing on the initial feature vector to eliminate the feature redundancy of the initial feature vector and obtain the comprehensive state feature data set of seismic waves;

[0114] In a specific embodiment, the vector feature fusion algorithm adopts a feature fusion method based on the attention mechanism, calculates the correlation weights between different features, adaptively adjusts the feature combination method, and realizes the optimal selection and fusion of features.

[0115] Specifically, this embodiment extracts and selects features from seismic wave propagation vector feature data, combined with time-frequency domain analysis methods, to effectively obtain key characteristic parameters of seismic waves (such as amplitude, frequency, phase, and energy distribution). Multi-resolution analysis using wavelet transforms enables the system to extract characteristic coefficients at different scales, forming a more comprehensive multidimensional feature vector. Furthermore, a vector feature fusion algorithm based on an attention mechanism further eliminates feature redundancy and enables optimal feature selection and combination.

[0116] The seismic wave identification and seismic wave classification are performed on the seismic wave comprehensive state feature dataset based on the deep learning model to obtain the seismic wave morphology classification result, specifically including:

[0117] A deep learning model based on convolutional neural networks and long short-term memory networks was constructed to perform multimodal feature fusion on the seismic wave comprehensive state feature dataset to generate preliminary classification features;

[0118] The calculation formula of the multimodal feature fusion is:

[0119]

[0120] Among them, F fused is the preliminary classification feature, F CNN and F LSTM are the feature matrices output by the convolutional neural network and the long short-term memory network, respectively, and α1 is F CNN and F LSTM The balance weight coefficient between them, β1 is the interaction feature weight coefficient, For element-by-element multiplication, Cov(F CNN ,F LSTM ) is F CNN and F LSTM The covariance of Var(F CNN ) is F CNNThe variance of Var(F LSTM ) is F LSTM variance;

[0121] The initial classification features are dynamically weighted through the attention mechanism, and the Softmax classifier is used to classify the waveform features of P waves and S waves, and the seismic wave morphology classification results are output;

[0122] The calculation formula for dynamically weighting the preliminary classification features through the attention mechanism is:

[0123]

[0124] Among them, W attention is the attention weight matrix, Q is the query matrix, K is the key matrix, d K is the dimension of the key matrix, V is the value matrix, and σ(·) is the softmax normalization function.

[0125] Specifically, this example constructs a deep learning model based on convolutional neural networks and long-short-term memory networks. This model generates preliminary classification features by fusing multimodal features of a dataset of comprehensive seismic wave state characteristics. This approach effectively combines the advantages of convolutional neural networks in image feature extraction with the powerful capabilities of long-short-term memory networks in time series data modeling, enabling a more comprehensive capture and representation of seismic wave characteristics in both the time and frequency domains.

[0126] This embodiment effectively integrates the output features from convolutional neural networks and long-short-term memory networks, thereby improving feature representation capabilities. By setting balanced weight coefficients and interactive feature weight coefficients, the model can adaptively adjust the contribution of each feature when processing multiple types of features, improving the ability to recognize complex seismic wave morphologies.

[0127] By implementing the attention mechanism to dynamically weight the preliminary classification features, the weights can be adaptively adjusted according to the importance of different features, which improves the flexibility and accuracy of the model while suppressing noise and redundant information.

[0128] The final seismic wave morphology classification results are obtained through a Softmax classifier, forming a complete classification logic, ensuring the probabilistic classification results and improving the ability to accurately distinguish P and S waves. This not only improves the accuracy of earthquake state identification, but also allows for more accurate and effective responses to different types of seismic waves in practical applications.

[0129] The analysis based on the seismic wave morphology classification results and seismic wave propagation vector characteristic data, combined with the earthquake intensity model, obtains the converter station seismic status assessment results, specifically including:

[0130] Based on the arrival time difference and amplitude characteristics of the P and S waves in the seismic wave morphology classification results, combined with the basis functions of the earthquake intensity model, the preliminary intensity assessment value of the converter station is calculated;

[0131] The calculation formula for the preliminary intensity assessment value is:

[0132]

[0133] Among them, I base is the preliminary intensity assessment value, Δt is the arrival time difference between the P wave and the S wave, A p is the measured value of the amplitude of the P wave, A s is the measured value of the amplitude of the S wave, k1 is the time difference proportional coefficient, k2 is the amplitude proportional coefficient, E soil is the elastic modulus of the soil layer, E rock is the elastic modulus of bedrock, ε is the reference attenuation coefficient, G shear is the shear modulus of the soil layer, G rock is the shear modulus of bedrock;

[0134] Based on the propagation direction and anisotropy index of the seismic wave propagation vector characteristic data, the preliminary intensity assessment value is dynamically corrected in terms of directional intensity to generate the earthquake state assessment result;

[0135] The calculation formula for the dynamic correction of directional intensity is:

[0136]

[0137] Among them, I final is the revised earthquake intensity assessment value, η1 is the empirical correction factor, is the three-dimensional propagation velocity vector, θ is the angle between the propagation direction and the layout direction of the main equipment of the converter station, D aniso is the seismic wave anisotropy index, D threshold is the seismic wave anisotropy index threshold, σ x , σ y , σ z is the standard deviation of the propagation velocity in three-dimensional orthogonal directions.

[0138] Specifically, this embodiment accurately calculates the preliminary intensity assessment value for the converter station by combining the arrival time difference and amplitude characteristics of P and S waves in the seismic wave morphology classification results with the basis functions of the earthquake intensity model. This process takes into account multiple key parameters, such as the actual measured values of the arrival time difference and amplitude, making the preliminary intensity assessment more reliable and accurate. In addition, this embodiment also uses multiple mechanical parameters (such as elastic modulus and shear modulus) to further enhance the understanding of seismic wave propagation characteristics under different geological conditions.

[0139] The formula for the preliminary intensity assessment incorporates the arrival time difference between P and S waves, as well as their respective amplitudes. This improvement more comprehensively reflects the physical properties of seismic waves and improves the accuracy of intensity assessments in complex environments.

[0140] The dynamic correction formula for the preliminary intensity assessment incorporates propagation direction and anisotropy indicators, effectively adjusting the assessment results to varying geological settings and propagation conditions. This correction process comprehensively considers the three-dimensional decomposition of propagation velocity and the relationship between propagation direction and converter station equipment layout, resulting in greater adaptability to seismic wave propagation characteristics.

[0141] Comparing and analyzing the converter station earthquake status assessment result with the earthquake early warning threshold to generate earthquake early warning information for the converter station, and proposing corresponding emergency response suggestions based on the earthquake early warning information specifically include:

[0142] Compare and analyze the converter station earthquake status assessment results with the preset earthquake warning thresholds in a hierarchical manner to generate an earthquake warning level;

[0143] In a specific embodiment, the earthquake early warning levels are graded and compared according to the following rules:

[0144] L1 warning: When the intensity value of the earthquake status assessment result is less than the first warning threshold, it is judged as an L1 warning and only monitoring is required;

[0145] L2 warning: When the intensity value is between the first warning threshold and the second warning threshold, it is judged as L2 warning, and it is necessary to initiate warning notification and check key equipment;

[0146] L3 warning: When the intensity value is greater than the second warning threshold, it is determined to be an L3 warning and a comprehensive emergency response procedure needs to be initiated.

[0147] The first warning threshold and the second warning threshold are set according to the seismic design standards of the converter station and historical earthquake data;

[0148] Generate corresponding emergency response recommendations based on the earthquake warning level, combined with the operating status of converter station equipment and emergency response plans;

[0149] In a specific embodiment, the emergency response suggestion is generated according to the earthquake warning level, specifically including:

[0150] Level 1 warning: Provides real-time monitoring recommendations for converter station equipment and records earthquake status assessment results;

[0151] Level 2 warning: Generates inspection recommendations, including checking the operating status of key equipment in the converter station (such as transformers, converter valves, and cooling systems), and initiates warning notifications to on-duty personnel;

[0152] Level 3 warning: Generate comprehensive emergency response recommendations, including emergency shutdown of non-critical equipment, activation of emergency power supply, evacuation of on-site personnel, and notification to higher-level management departments to activate emergency plans.

[0153] Specifically, this embodiment realizes the graded early warning of earthquake risks at converter stations by constructing a multi-dimensional and multi-level earthquake early warning threshold system and intelligently matching and dynamically comparing it with real-time assessment results. This embodiment forms a complete set of "monitoring-assessment-early warning-response" closed-loop mechanism by setting different levels of early warning thresholds (including but not limited to minor early warning, general early warning, severe early warning, etc.) and combining specific emergency response strategies for each level. This embodiment can not only identify and warn potential earthquake risks in a timely manner, but also automatically trigger corresponding emergency response suggestions according to different early warning levels, thereby improving the emergency response efficiency and safety management level of converter stations in earthquake events, and providing more reliable protection for equipment and personnel safety.

[0154] See also Figure 2 The present invention also provides a vector earthquake state identification system for converter stations based on distributed optical fiber, the system comprising:

[0155] A data monitoring module is used to collect multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and perform adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data;

[0156] a processing and analysis module, configured to perform phase demodulation processing and time difference analysis on the seismic monitoring data, and process the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data;

[0157] A feature integration module is used to integrate and optimize the seismic wave propagation vector feature data using a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set;

[0158] The morphological classification module is used to identify and classify seismic waves based on the comprehensive state feature dataset of seismic waves based on the deep learning model to obtain the seismic wave morphological classification results;

[0159] The status assessment module is used to analyze the seismic wave morphology classification results and seismic wave propagation vector characteristic data in combination with the earthquake intensity model to obtain the seismic status assessment results of the converter station;

[0160] The early warning response module is used to compare and analyze the earthquake status assessment result of the converter station with the earthquake early warning threshold, generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

[0161] In a specific embodiment, the system further includes a converter station topology monitoring module for monitoring the vibration state of major structural components within the converter station using distributed optical fiber sensors, wherein the major structural components include support towers, current-carrying circuits, and connection equipment;

[0162] The converter station topology monitoring module is specifically used for:

[0163] Collect and record the vibration characteristic data of main structural components under normal operation and earthquake conditions;

[0164] Establish a topological relationship model of the main structural components and analyze the vibration energy transmission path and mutual influence relationship between the structural components;

[0165] Combined with the characteristic data of seismic wave propagation vectors, the direction and intensity of the force exerted by seismic waves on major structural components are calculated in real time;

[0166] Predict the stress and deformation degree and potential cascading failure paths of major structural components based on deep learning models;

[0167] Compare the vibration status assessment results of major structural components with safety thresholds to identify the key structural components that are first affected by the earthquake and predict the chain reactions that may be triggered by the earthquake;

[0168] Provide structural component vibration status information to the status assessment module and early warning response module as a supplementary decision-making basis for earthquake status assessment and emergency response recommendations.

[0169] Specifically, a distributed optical fiber-based converter station vector earthquake state identification system in this embodiment integrates a multi-dimensional monitoring unit, an intelligent analysis unit, and an early warning response unit to achieve high-precision earthquake signal acquisition, deep learning-driven feature extraction and fusion, dynamic hierarchical early warning decision-making, and a highly reliable distributed operation architecture. It not only improves the accuracy, real-time nature, and system reliability of earthquake monitoring, but also provides comprehensive safety protection and scalable standardized solutions for converter stations, effectively reducing the potential risks and economic losses caused by earthquake disasters.

[0170] The present invention also discloses an electronic device, comprising: at least one processor, at least one memory communication interface and a bus: wherein the processor, memory and communication interface communicate with each other through the bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement a method for identifying the vector earthquake state of a converter station based on distributed optical fiber.

[0171] The present invention also discloses a computer-readable storage medium storing computer instructions that cause the computer to implement all or part of the steps of the method for identifying vector seismic states of converter stations based on distributed optical fiber, as described in an embodiment of the present invention. The storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0172] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying vector earthquake states of converter stations based on distributed optical fiber, characterized in that: The following steps are involved: Collecting multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and performing adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data; Performing phase demodulation processing and time difference analysis on the seismic monitoring data, and processing the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data; The seismic wave propagation vector feature data is integrated and optimized by a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set; Based on the deep learning model, seismic wave identification and classification are performed on the seismic wave comprehensive state feature dataset to obtain the seismic wave morphology classification results; Based on the seismic wave morphology classification results and seismic wave propagation vector characteristic data, combined with the earthquake intensity model, the seismic status assessment results of the converter station are obtained; Compare and analyze the converter station earthquake status assessment result with the earthquake early warning threshold to generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

2. The method for identifying vector earthquake states of converter stations based on distributed optical fiber according to claim 1, characterized in that: The method includes collecting multi-dimensional vibration signal data of the converter station by using a distributed optical fiber sensor, and performing adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data, specifically comprising: Deploying a distributed optical fiber sensor network around the converter station according to a preset spatial distribution scheme, wherein the distributed optical fiber sensor network includes optical fiber sensors deployed along three orthogonal directions of the X-axis, the Y-axis, and the Z-axis, and collecting multi-dimensional vibration signal data through the distributed optical fiber sensor network; Adaptive filtering is performed on the multidimensional vibration signal data to obtain a multidimensional filtered vibration signal, and an environmental noise suppression algorithm is used to eliminate environmental noise interference in the multidimensional filtered vibration signal to obtain earthquake monitoring data.

3. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 2, characterized in that: Also includes: A key equipment monitoring system is constructed by deploying dedicated distributed optical fiber sensor arrays on key equipment in a converter station, wherein the key equipment includes transformers, converter valves, filter devices, and connection equipment; the dedicated distributed optical fiber sensor arrays are densely deployed in a grid or ring manner on the core structural parts of the key equipment; The key equipment monitoring system is specifically used for: Collect vibration characteristic data of key equipment under normal operating conditions and establish a baseline model of equipment vibration characteristics; Real-time monitoring of high-frequency vibration signals of key equipment, using higher sampling frequency and spatial resolution than the main system to capture tiny vibration changes; Combine time-frequency analysis methods to extract the deformation characteristics of key equipment under earthquake action and distinguish the natural vibration of equipment from abnormal vibration caused by earthquake; Establish digital twin models of key equipment based on finite element analysis, match measured vibration data with theoretical models in real time, and calculate stress distribution and potential damage levels of key equipment; Correlate and analyze key equipment monitoring data with seismic wave propagation vector feature data acquired by the main system to determine the impact path and damage mode of seismic waves on key equipment; Provide key equipment status information to the status assessment module as a decision basis for generating earthquake status assessment results and equipment-level emergency response recommendations.

4. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 1, characterized in that: The phase demodulation processing and time difference analysis are performed on the seismic monitoring data, and the processing is performed using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data, specifically including: Based on the principle of phase-sensitive optical time-domain reflectometry, the seismic monitoring data is phase-demodulated and processed, and the time difference between the arrival of seismic waves at each monitoring point is analyzed using the time difference positioning method to obtain preliminary demodulated data. The spatial vector decomposition algorithm is used to decompose the preliminary demodulated data into three-dimensional spatial components, extract the propagation characteristics of seismic waves in the X-axis, Y-axis and Z-axis directions, and obtain the seismic wave propagation vector characteristic data.

5. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 1, characterized in that: The vector feature fusion algorithm is used to integrate and optimize the seismic wave propagation vector feature data to obtain a seismic wave comprehensive state feature data set, specifically including: Perform feature extraction and feature selection on the characteristic data of seismic wave propagation vectors, obtain key characteristic parameters of seismic waves based on time-frequency domain analysis methods, and construct initial characteristic vectors based on the key characteristic parameters; The vector feature fusion algorithm is used to perform multi-scale feature fusion and optimization processing on the initial feature vector to eliminate the feature redundancy of the initial feature vector and obtain the comprehensive state feature dataset of seismic waves.

6. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 5, characterized in that: The seismic wave identification and seismic wave classification are performed on the seismic wave comprehensive state feature dataset based on the deep learning model to obtain the seismic wave morphology classification result, specifically including: A deep learning model based on convolutional neural networks and long short-term memory networks was constructed to perform multimodal feature fusion on the seismic wave comprehensive state feature dataset to generate preliminary classification features; The calculation formula of the multimodal feature fusion is: Among them, F fused is the preliminary classification feature, F CNN and F LSTM are the feature matrices output by the convolutional neural network and the long short-term memory network, respectively, and α1 is F CNN and F LSTM The balance weight coefficient between them, β1 is the interaction feature weight coefficient, For element-by-element multiplication, Cov(F CNN ,F LSTM ) is F CNN and F LSTM The covariance of Var(F CNN ) is F CNN The variance of Var(F LSTM ) is F LSTM variance; The initial classification features are dynamically weighted through the attention mechanism, and the Softmax classifier is used to classify the waveform features of P waves and S waves, and the seismic wave morphology classification results are output; The calculation formula for dynamically weighting the preliminary classification features through the attention mechanism is: Among them, W attention is the attention weight matrix, Q is the query matrix, K is the key matrix, d K is the dimension of the key matrix, V is the value matrix, and σ(·) is the softmax normalization function.

7. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 1, characterized in that: The analysis based on the seismic wave morphology classification results and seismic wave propagation vector characteristic data, combined with the earthquake intensity model, obtains the converter station seismic status assessment results, specifically including: Based on the arrival time difference and amplitude characteristics of the P and S waves in the seismic wave morphology classification results, combined with the basis functions of the earthquake intensity model, the preliminary intensity assessment value of the converter station is calculated; The calculation formula for the preliminary intensity assessment value is: Among them, I base is the preliminary intensity assessment value, Δt is the arrival time difference between the P wave and the S wave, A p is the measured value of the amplitude of the P wave, A s is the measured value of the amplitude of the S wave, k1 is the time difference proportional coefficient, k2 is the amplitude proportional coefficient, E soil is the elastic modulus of the soil layer, E rock is the elastic modulus of bedrock, ε is the reference attenuation coefficient, G shear is the shear modulus of the soil layer, G rock is the shear modulus of bedrock; Based on the propagation direction and anisotropy index of the seismic wave propagation vector characteristic data, the preliminary intensity assessment value is dynamically corrected in terms of directional intensity to generate the earthquake state assessment result; The calculation formula for the dynamic correction of directional intensity is: Among them, I final is the revised earthquake intensity assessment value, η1 is the empirical correction factor, is the three-dimensional propagation velocity vector, θ is the angle between the propagation direction and the layout direction of the main equipment of the converter station, D aniso is the seismic wave anisotropy index, D threshold is the seismic wave anisotropy index threshold, σ x , σ y , σ z is the standard deviation of the propagation velocity in three-dimensional orthogonal directions.

8. The method for identifying vector earthquake status of a converter station based on distributed optical fiber according to claim 1, characterized in that: Comparing and analyzing the converter station earthquake status assessment result with the earthquake early warning threshold to generate earthquake early warning information for the converter station, and proposing corresponding emergency response suggestions based on the earthquake early warning information specifically include: Compare and analyze the converter station earthquake status assessment results with the preset earthquake warning thresholds in a hierarchical manner to generate an earthquake warning level; According to the earthquake warning level, combined with the operating status of the converter station equipment and the emergency response plan, corresponding emergency response recommendations are generated.

9. A converter station vector earthquake state identification system based on distributed optical fiber, used to execute the converter station vector earthquake state identification method based on distributed optical fiber according to any one of claims 1 to 8, characterized in that: The system comprises: A data monitoring module is used to collect multi-dimensional vibration signal data of the converter station through distributed optical fiber sensors, and perform adaptive filtering and environmental noise suppression on the multi-dimensional vibration signal data to obtain earthquake monitoring data; A processing and analysis module, configured to perform phase demodulation processing and time difference analysis on the seismic monitoring data, and process the data using a space vector decomposition algorithm to obtain seismic wave propagation vector characteristic data; A feature integration module is used to integrate and optimize the seismic wave propagation vector feature data using a vector feature fusion algorithm to obtain a seismic wave comprehensive state feature data set; The morphological classification module is used to identify and classify seismic waves based on the comprehensive state feature dataset of seismic waves based on the deep learning model to obtain the seismic wave morphological classification results; The status assessment module is used to analyze the seismic wave morphology classification results and seismic wave propagation vector characteristic data in combination with the earthquake intensity model to obtain the seismic status assessment results of the converter station; The early warning response module is used to compare and analyze the earthquake status assessment result of the converter station with the earthquake early warning threshold, generate earthquake early warning information of the converter station, and propose corresponding emergency response suggestions based on the earthquake early warning information.

10. The distributed optical fiber-based converter station vector earthquake state identification system according to claim 9, characterized in that: The system also includes a converter station topology monitoring module for monitoring the vibration state of the main structural components inside the converter station using distributed optical fiber sensors. The main structural components include support towers, current-carrying circuits, and connection equipment. The converter station topology monitoring module is specifically used for: Collect and record the vibration characteristic data of main structural components under normal operation and earthquake conditions; Establish a topological relationship model of the main structural components and analyze the vibration energy transmission path and mutual influence relationship between the structural components; Combined with the characteristic data of seismic wave propagation vectors, the direction and intensity of the force exerted by seismic waves on major structural components are calculated in real time; Predict the stress and deformation degree and potential cascading failure paths of major structural components based on deep learning models; Compare the vibration status assessment results of major structural components with safety thresholds to identify the key structural components that are first affected by the earthquake and predict the chain reactions that may be triggered by the earthquake; Provide structural component vibration status information to the status assessment module and early warning response module as a supplementary decision-making basis for earthquake status assessment and emergency response recommendations.

Images