Optical cable intrusion detection method and device based on distributed optical fiber vibration sensing

Abstract

The invention provides an optical cable intrusion detection method and device based on distributed optical fiber vibration sensing, belongs to the technical field of optical fiber sensing and security monitoring, and solves the problems that existing optical cable security monitoring is high in false alarm rate, poor in real-time performance and difficult to meet long-distance optical cable security requirements. The method comprises the following steps: acquiring a first vibration signal in a preset amplitude range; performing frequency filtering processing on the first vibration signal to obtain a second vibration signal; performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector; performing multi-scale time sequence classification processing on the high-dimensional feature vector to obtain classification data; and determining the vibration state of the optical cable according to the classification data. According to the scheme, long-distance security monitoring of the optical cable is realized, and the accuracy of a detection result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of optical fiber sensing and security monitoring, and particularly to a method and device for detecting cable intrusion based on distributed optical fiber vibration sensing. Background Art

[0002] With the accelerating urbanization process and the rapid development of the information society, long-distance optical cables, as key infrastructure for information transmission, the importance of their security protection has become increasingly prominent. Traditional security monitoring means mainly rely on manual inspections and simple sensor alarm systems. The manual inspection method has problems such as low efficiency and strong subjectivity, and it is difficult to monitor long-distance optical cables in real time and comprehensively. Although the simple sensor alarm system realizes automated monitoring to a certain extent, in the complex environmental noise background, its accuracy and reliability are poor, and it cannot effectively distinguish real vibration events and environmental interferences, such as noise interferences generated by wind and rain, vehicle driving, animal activities, etc., and is easily misjudged as real intrusion or damage events, with frequent false alarms and missed alarms, resulting in unnecessary on-site investigations by staff, wasting a large amount of manpower and material resources; on the other hand, the real-time performance of the system is poor. In the process of data processing and analysis of the existing system, it often takes a long time, and it is difficult to accurately classify and respond to vibration events in the first time, and it cannot meet the requirements of timeliness for the security protection of long-distance optical cables.

[0003] Therefore, in order to ensure the safe operation of long-distance optical cables, it is urgent to develop an efficient and stable optical fiber vibration event classification system and device to improve the accuracy and real-time performance of monitoring, reduce the false alarm rate, and provide reliable technical support for the security protection of long-distance optical cables. Summary of the Invention

[0004] The present invention provides a method and device for detecting cable intrusion based on distributed optical fiber vibration sensing, which solves the problems of high false alarm rate, poor real-time performance, and difficulty in meeting the security requirements of long-distance optical cables in the existing cable security monitoring.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows:

[0006] An embodiment of the present invention provides a method for detecting cable intrusion based on distributed optical fiber vibration sensing, including:

[0007] Obtaining a first vibration signal within a preset amplitude range;

[0008] Performing frequency filtering processing on the first vibration signal to obtain a second vibration signal;

[0009] Performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector;

[0010] Perform multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data;

[0011] Determine the vibration state of the optical cable according to the classification data.

[0012] Optionally, the obtaining of the first vibration signal within a preset amplitude range includes:

[0013] Obtain the original vibration signal through a distributed optical fiber vibration sensor installed along the subway track at a preset distance;

[0014] Perform amplitude screening processing on the original vibration signal according to a preset dynamic amplitude threshold to obtain the first vibration signal.

[0015] Optionally, the frequency filtering processing of the first vibration signal to obtain the second vibration signal includes:

[0016] Obtain the noise coefficient;

[0017] Perform frequency band decomposition on the first vibration signal according to the noise coefficient to obtain a vibration sequence signal including multiple frequency band vibration signals;

[0018] Perform reconstruction processing on the noise coefficient to obtain the target noise coefficient;

[0019] Perform frequency filtering processing on the vibration sequence signal according to the target noise coefficient to obtain the second vibration signal.

[0020] Optionally, the feature extraction processing of the second vibration signal to obtain the high-dimensional feature vector includes:

[0021] Perform frame addition and windowing processing on the second vibration signal to obtain the short-time frame after windowing;

[0022] Perform time-frequency domain feature extraction processing on the short-time frame after windowing to obtain the time-frequency feature matrix;

[0023] Perform embedding coding processing on the time-frequency feature matrix to obtain the high-dimensional feature vector.

[0024] Optionally, the multi-scale time series classification processing of the high-dimensional feature vector to obtain the classification data includes:

[0025] Input the high-dimensional feature vector into the trained multi-scale time series classification model for processing to obtain the classification data.

[0026] Optionally, the multi-scale time series classification model is trained through the following process:

[0027] Obtain the original vibration sample signal through a distributed optical fiber vibration sensor installed along the subway track at a preset distance;

[0028] Perform amplitude screening processing on the original vibration sample signal to obtain a first vibration sample signal;

[0029] Perform frequency filtering processing on the first vibration sample signal to obtain a second vibration sample signal;

[0030] According to the subway vibration sample type label, the excavation vibration sample type label, and the climbing vibration type label, perform classification processing on the second vibration signal to obtain a vibration event sample library;

[0031] Extract features from the sample data in the vibration event sample library to obtain a time-frequency domain feature matrix;

[0032] According to the time-frequency domain feature matrix, obtain a multi-scale time series classification model.

[0033] Optionally, process the time-frequency domain feature matrix to obtain fused discriminative features, including:

[0034] Perform multi-scale time series feature extraction on the time-frequency domain feature matrix to obtain multi-scale eigenvalues;

[0035] Perform spatio-temporal attention weighting on the multi-scale eigenvalues to obtain attention-weighted features;

[0036] Perform bidirectional time series modeling processing on the attention-weighted features to obtain a prediction probability;

[0037] According to the prediction probability, perform focal loss optimization and parameter update on the classification model with preset parameters to obtain a multi-scale time series classification model.

[0038] An embodiment of the present invention further provides an optical cable intrusion detection device based on distributed optical fiber vibration sensing, including:

[0039] An acquisition module for acquiring a first vibration signal within a preset amplitude range;

[0040] A processing module for performing frequency filtering processing on the first vibration signal to obtain a second vibration signal; performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector; performing multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data;

[0041] A determination module for determining the vibration state of the optical cable according to the classification data.

[0042] An embodiment of the present invention further provides a computing device, including: a processor and a memory storing a computer program. When the computer program is run by the processor, the above method is executed.

[0043] An embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when run on a computer, cause the computer to execute the above method.

[0044] The technical solution of the present invention at least includes the following effects:

[0045] The above solution of the present invention obtains a first vibration signal within a preset amplitude range; performs frequency filtering processing on the first vibration signal to obtain a second vibration signal; performs feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector; performs multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data; determines the vibration state of the optical cable according to the classification data; by adopting a spatio-temporal joint modeling method, it realizes stable monitoring performance under complex environmental noise and improves the intelligent level of long-distance optical cable security monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flowchart of an optical cable intrusion detection method based on distributed fiber optic vibration sensing provided by an embodiment of the present invention;

[0047] Figure 2 is a flowchart of an intrusion monitoring process of an optical cable intrusion detection method based on distributed fiber optic vibration sensing provided by an embodiment of the present invention;

[0048] Figure 3 is a structural diagram of an optical cable intrusion detection device based on distributed fiber optic vibration sensing provided by an embodiment of the present invention;

[0049] Figure 4 is a schematic structural diagram of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] Hereinafter, exemplary embodiments of the present invention will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be completely conveyed to those skilled in the art.

[0051] As Figure 1 and Figure 2 shown, an embodiment of the present invention proposes an optical cable intrusion detection method based on distributed fiber optic vibration sensing, including:

[0052] Step 11, obtaining a first vibration signal within a preset amplitude range;

[0053] Step 12, performing frequency filtering processing on the first vibration signal to obtain a second vibration signal;

[0054] Step 13: Perform feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector;

[0055] Step 14: Perform multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data;

[0056] Step 15: Determine the vibration state of the optical cable according to the classification data.

[0057] In this embodiment, first, the original vibration signal is obtained through optical fiber vibration sensors distributed along the optical cable, and the amplitude of the original vibration signal is processed to obtain the first vibration signal. These sensors can be fiber Bragg grating sensors, acceleration sensors, etc. When a vibration event (such as excavation, climbing, knocking, etc.) occurs around the optical cable, the sensor can sense and convert these physical vibrations into electrical signals or optical signals. To ensure that various possible vibration situations can be accurately captured, the signal acquisition frequency needs to be set according to actual requirements. Generally speaking, for high-frequency vibration events, such as rapid excavation actions, a higher acquisition frequency is required, which can reach thousands of times per second or even higher; while for low-frequency vibration events, such as slow human walking, the acquisition frequency can be appropriately reduced, but it is also necessary to ensure that the vibration characteristics can be accurately recorded.

[0058] Since the original vibration signal collected often contains a large amount of noise, such as environmental noise, sensor self-noise, etc.; denoising processing can adopt filtering techniques; for example, high-frequency noise can be removed through low-pass filtering, and the low-frequency useful components in the vibration signal can be retained; band-pass filtering can screen the vibration signal within a specific frequency range and remove the noise of other irrelevant frequencies. To improve the quality of the vibration signal and facilitate subsequent feature extraction, the signal also needs to be enhanced. For example, signal amplification technology is used to amplify the weak vibration signal to an appropriate amplitude; or signal smoothing technology is used to make the waveform of the signal smoother, reduce the burrs and mutations, and improve the stability of the signal; through the above processing, the second vibration signal is obtained.

[0059] Analyze the characteristics of the second vibration signal from the time domain perspective. Common time domain characteristics include mean value, variance, peak value, peak-to-peak value, rise time, fall time, etc.

[0060] Convert the second vibration signal from the time domain to the frequency domain through methods such as Fourier transform to extract frequency domain characteristics.

[0061] Frequency domain features mainly include frequency components, power spectral density, etc. For example, different vibration events exhibit different frequency distributions in the frequency domain. By analyzing these frequency components, different vibration types can be distinguished. Combining time domain and frequency domain information, methods such as wavelet transform are used for time-frequency domain feature extraction. Wavelet transform can provide local features of the signal at different times and frequencies, more comprehensively reflecting the characteristics of the vibration signal. The extracted time domain, frequency domain, and time-frequency domain features are combined together to form a high-dimensional feature vector. This feature vector can comprehensively and meticulously describe the characteristics of the vibration signal.

[0062] After obtaining the high-dimensional feature vector, it is necessary to select an appropriate classification model to classify the high-dimensional feature vector. Different classification models have different characteristics and applicable scenarios, and need to be selected according to the actual situation. The classification model can be trained using sample data with known categories. The training data should include various different types of vibration events and their corresponding high-dimensional feature vectors and category labels.

[0063] Through training, the classification model learns the characteristic patterns of different types of vibration events, so as to be able to accurately classify new high-dimensional feature vectors. The high-dimensional feature vector to be classified is input into the trained classification model, and the model makes a judgment based on the learned characteristic patterns and outputs the classification result. The classification result can be different types of vibration events, such as excavation, climbing, knocking, etc., or an assessment of the risk level of the vibration event, such as normal, suspicious, dangerous, etc.

[0064] According to the classification result, the vibration state of the optical cable is determined and corresponding alarm rules are set. For example, when the classification result is a "dangerous" vibration event, a high-level alarm is triggered; when the classification result is a "suspicious" vibration event, a low-level alarm is triggered; when the classification result is "normal", no alarm is triggered. The operation state of the security monitoring device is controlled according to the alarm rules. The security monitoring device can be an audible and visual alarm, a text message alarm module, an email alarm system, etc. When the alarm is triggered, the alarm device will emit corresponding alarm signals, such as emitting a loud alarm sound, flashing a warning light, sending an alarm text message or email, etc., to promptly notify relevant personnel for handling.

[0065] This technical solution can accurately sense the vibration situation around the optical cable by acquiring vibration signals and performing a series of processes. From the initial signal acquisition, to preprocessing to remove noise and other interferences, and then to feature extraction to obtain high-dimensional feature vectors, the analysis of vibration signals becomes more meticulous and accurate, enabling the precise identification of different types and intensities of vibrations, which helps to accurately determine whether the optical cable is in a safe state; determining classification data based on the high-dimensional feature vectors, thereby determining the vibration state of the optical cable and controlling the operating state of the alarm device, achieving intelligent judgment and decision-making. The system can automatically decide whether to issue an alarm according to the analysis result without manual intervention, improving the timeliness and accuracy of monitoring and reducing the influence of human factors; the entire process can quickly process vibration signals. Once an abnormal situation is detected, it can promptly control the alarm device to issue an alarm, enabling relevant personnel to quickly take measures to avoid damage to the optical cable and ensuring the safe and stable operation of the communication network; the process of preprocessing and feature extraction of vibration signals helps to improve the adaptability of the system to different environments and different types of vibrations. Regardless of how the external environment changes, it can analyze vibration signals more accurately, ensuring the reliability of monitoring, and can be widely applied to various complex environments or different types of optical cable lines.

[0066] In an optional embodiment of the present invention, in step 11, the obtaining of the first vibration signal within a preset amplitude range may include:

[0067] Step 111, obtaining the original vibration signal through a distributed fiber optic vibration sensor installed along the subway track at a preset distance;

[0068] Step 112, performing amplitude screening processing on the original vibration signal according to a preset dynamic amplitude threshold to obtain the first vibration signal.

[0069] In this embodiment, first, hardware deployment is carried out, that is, a distributed fiber optic sensor is deployed every 200m along the subway track, covering a 10km optical cable section, and the sampling rate is set to 2kHz; the sensor collects vibration signals in real time, and the original signal contains the following components:

[0070] (1) Subway train vibration: The frequency is concentrated in 20~200Hz, showing a periodic continuous waveform;

[0071] (2) Environmental noise: such as wind noise (low frequency), vehicle passing (medium frequency), etc.;

[0072] (3) Potential intrusion signals: such as impact-type vibration (excavation behavior, frequency band 50~300Hz) or intermittent vibration (climbing behavior, frequency band 10~100Hz);

[0073] The collected original vibration signal sequence can be expressed as: x raw (t), where t is the time index.

[0074] After the original vibration signal x raw (t) is collected, the dynamic amplitude threshold method is used for screening, and the original vibration signal is screened to remove instantaneous pulse noises such as lightning strikes and equipment interferences, so as to obtain a target vibration signal; specifically,

[0075] Based on the amplitudes of the past N frames (N is taken as 1000 in this example), the mean value μ of the historical signal amplitude and the standard deviation σ of the historical signal amplitude are calculated:

[0076] ;

[0077] ;

[0078] wherein, x raw (t i ) is the i-th signal in the original vibration signal sequence.

[0079] According to the mean value μ of the historical signal amplitude and the standard deviation σ of the historical signal amplitude, a threshold U is set:

[0080] U = μ ± 3σ;

[0081] By setting the threshold U, only the signal segments with amplitudes within this range are retained. The screening process can be expressed as:

[0082] ;

[0083] wherein, x filtered (t) is the first vibration signal.

[0084] In this embodiment, by deploying distributed optical fiber sensors along the subway track and combining the dynamic amplitude threshold screening technology, accurate extraction of effective vibration signals and suppression of instantaneous interferences are realized, ensuring full coverage of the monitoring area and high-precision data acquisition. And based on the mean value and standard deviation of the historical signals, the threshold is dynamically set to filter out amplitude abnormal noises such as lightning strikes and equipment interferences, while retaining subway vibrations and potential intrusion signals, providing low-noise and high-signal-to-noise input data for subsequent processing, and significantly improving the anti-interference ability and reliability of the detection system.

[0085] In an alternative embodiment of the present invention, in step 12, frequency filtering processing is performed on the first vibration signal to obtain a second vibration signal, which may include:

[0086] Step 121, obtaining a noise coefficient;

[0087] Step 122, according to the noise coefficient, performing frequency band decomposition on the first vibration signal to obtain a vibration sequence signal including multiple frequency band vibration signals;

[0088] Step 123, reconstructing the noise coefficient to obtain a target noise coefficient;

[0089] Step 124: Perform frequency filtering on the vibration sequence signal according to the target noise coefficient to obtain a second vibration signal.

[0090] In this embodiment, the first vibration signal is first marked, and segments with a duration of not less than 0.5 s are retained to avoid transient noise residue, and a valid vibration signal segment set is obtained. , where each segment corresponds to a time window;

[0091] The wavelet basis is used to perform a 5-layer decomposition, remove high-frequency detail coefficients (>500Hz), and reconstruct low-frequency signals to retain the effective components of subway vibration and intrusion events. The decomposition process can be expressed as:

[0092] ;

[0093] Among them, c A5 (t) is the fifth-layer approximation coefficient, representing the low-frequency effective signal (such as subway vibration); c Dj (t) is the detail coefficient of the jth layer (j=1,2,...,5), which represents high-frequency noise (such as wind noise and equipment interference).

[0094] For detail coefficient c Dj (t) (j=1,2,...,5) performs hard threshold processing to remove components with frequency >500Hz; only the low-frequency approximate coefficient c is retained A5 (t), reconstructed denoised signal;

[0095] Signal frequency filtering can be expressed as:

[0096] ;

[0097] Among them, ϕ 5,m (t) is the translation version of the fifth layer wavelet basis function; c A5 (t m ) is the translation version of the 5th layer approximation coefficient; m is the translation index; is the reconstructed noise-reduced signal, that is, the second vibration signal after preprocessing.

[0098] In this embodiment, the wavelet multi-scale decomposition and hard threshold filtering technology are used to achieve the technical effect of effectively separating low-frequency effective signals and suppressing high-frequency noise. The first vibration signal is divided into multiple frequency bands by 5-layer wavelet decomposition, and the high-frequency detail coefficients with a frequency greater than 500Hz are eliminated, while the low-frequency approximate coefficients are retained. At the same time, hard threshold processing and signal reconstruction are used to eliminate residual transient noise, ensuring that the signal-to-noise ratio of the signal after denoising is significantly improved, providing a pure time-frequency data foundation for subsequent feature extraction.

[0099] In an alternative embodiment of the present invention, in step 13, performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector may include:

[0100] Step 131, performing frame segmentation and windowing processing on the second vibration signal to obtain a short-time frame after windowing;

[0101] Step 132, performing time-frequency domain feature extraction processing on the short-time frame after windowing to obtain a time-frequency feature matrix;

[0102] Step 133, performing embedding coding processing on the time-frequency feature matrix to obtain a high-dimensional feature vector.

[0103] In this embodiment, first, the second vibration signal (long-time sequence signal) is segmented into short-time segments to reduce spectral leakage and facilitate time-frequency analysis. The frame segmentation parameters are: frame length 100 ms, frame shift 50 ms, and Hamming window smoothing;

[0104] In step 131, the frame segmentation and windowing processing process can be expressed as:

[0105] ;

[0106] where, is the short-time frame after windowing; t m is the start time of the m-th frame; N is the number of sampling points corresponding to the frame length; w(n) is the Hamming window function; where,

[0107] ;

[0108] After obtaining the short-time frame after windowing, calculate the Mel spectrum (64 dimensions) and Mel frequency cepstral coefficients (12 dimensions) for each frame of the signal, and fuse them into a 76-dimensional feature vector;

[0109] The specific implementation process of step 132 includes:

[0110] (1) Performing short-time Fourier transform on the short-time frame after windowing:

[0111] ;

[0112] where, f s is the sampling frequency; f is the frequency component; X (m) (f) is the signal after short-time Fourier transform.

[0113] (2) Performing Mel filter bank mapping:

[0114] ;

[0115] where, H b(f) is the frequency response of the b-th Mel filter; is the Mel spectrogram.

[0116] (3) Calculate the Mel-frequency cepstral coefficients. First, take the logarithm of the Mel spectrogram: ; Then perform a discrete cosine transform to extract the Mel-frequency cepstral coefficients:

[0117] ;

[0118] where, MFCC (m) (c) are the Mel-frequency cepstral coefficients;

[0119] (4) Horizontally concatenate the 64-dimensional Mel spectrogram and the 12-dimensional Mel-frequency cepstral coefficients MFCC (m) (c) in the feature dimension to obtain the time-frequency feature matrix F (m) ;

[0120] In step 133, after obtaining the time-frequency feature matrix F (m) perform an embedding encoding process on it, that is, through the cascade of convolution and recurrent neural network, fuse local frequency domain patterns and global temporal dependencies to generate a compact high-dimensional representation. Specifically, the first layer is a 1D convolutional layer, and the 1D convolutional kernel (width 5, stride 2) extracts local frequency domain patterns; the second layer is a bidirectional LSTM layer, and the bidirectional LSTM (64 hidden units) captures temporal dependencies; the output is a 256-dimensional high-dimensional vector representing the spatio-temporal features of the vibration event.

[0121] In this embodiment, through frame division and windowing, time-frequency feature fusion, and deep learning embedding encoding, effective extraction and characterization optimization of high-dimensional spatio-temporal features are achieved. Specifically, it includes: 1) Frame division and windowing reduce spectral leakage and segment short-time signals; 2) Extract 76-dimensional Mel spectrogram (64-dimensional) and MFCC (12-dimensional) to fuse time-frequency characteristics; 3) Cascade encoding of 1D convolution (local frequency domain patterns) and bidirectional LSTM (global temporal dependencies) generates a 256-dimensional high-dimensional vector to comprehensively depict the spatio-temporal features of the vibration event, providing robust and highly discriminative input features for multi-scale classification.

[0122] In an alternative embodiment of the present invention, in step 14, performing multi-scale temporal classification processing on the high-dimensional feature vector to obtain classification data may include:

[0123] Step 141, input the high-dimensional feature vector into a trained multi-scale temporal classification model for processing to obtain classification data.

[0124] In this embodiment, inputting the high-dimensional feature vector into the multi-scale temporal classification model for processing, and finally obtaining classification data. This process includes two stages: feature extraction and weighted processing, and adaptive fusion processing;

[0125] Among them, the multi-scale time series classification model is trained through the following process:

[0126] Step 1, obtain the original vibration sample signal through a distributed fiber optic vibration sensor installed along the subway track at a preset distance;

[0127] Step 2, perform amplitude screening processing on the original vibration sample signal to obtain the first vibration sample signal;

[0128] Step 3, perform frequency filtering processing on the first vibration sample signal to obtain the second vibration sample signal;

[0129] Step 4, classify the second vibration signal according to the subway vibration sample type label, excavation vibration sample type label, and climbing vibration type label to obtain a vibration event sample library;

[0130] Step 5, extract features from the sample data in the vibration event sample library to obtain a time-frequency domain feature matrix;

[0131] Step 6, obtain the multi-scale time series classification model according to the time-frequency domain feature matrix.

[0132] Specifically, the amplitude screening process of the original vibration sample signal in Step 2 includes the following process:

[0133] Dynamic threshold alarm: adaptively adjust the classification confidence threshold according to the environmental noise level (such as the increase in noise during rain and snow weather) (default 0.9, reduced to 0.7 when the noise is strong);

[0134] Alarm rule: trigger a real-time alarm when it is detected as "excavation" for 3 consecutive frames or the confidence of a single frame "climbing">0.95.

[0135] The method of frequency filtering processing on the first vibration sample signal in Step 3 is the same as the filtering processing method in Step 12;

[0136] In Step 4, classify the preprocessed vibration signal according to the subway vibration sample type label, excavation vibration sample type label, and climbing vibration type label to obtain a vibration event sample library, including: classify and label according to vibration type to construct a standardized sample library. The classification rules are as follows:

[0137] (1) Subway vibration. Its characteristic is a continuous vibration waveform, with a duration of 30 - 60 seconds and a main frequency of 80 Hz; extract the signal segment when the train passes and label it as "subway".

[0138] (2)Excavation vibration. Characterized by impact-type vibration, with a duration of 0.5 - 2 seconds and a frequency band of 50 - 300 Hz; Simulated excavation experiment data, labeled as "excavation".

[0139] (3)Climbing vibration. Characterized by intermittent vibration, lasting for 5 - 10 seconds and having a frequency band of 10 - 100 Hz; Simulated climbing behavior data, labeled as "climbing".

[0140] The labeled vibration events are output as a data set, expressed as:

[0141] y k ∈ {subway vibration, excavation vibration, climbing vibration};

[0142] Construct a standardized sample library, including 3 types of labels (subway vibration, excavation vibration, climbing vibration), with 1000 groups of samples for each type, and divide the training set, validation set, and test set according to 8:1:1.

[0143] In step 5, feature extraction is performed on the sample data in the vibration event sample library to obtain a time-frequency domain feature matrix; Specifically, the vibration event sample library is processed according to the process of step 13 to obtain a time-frequency domain feature matrix;

[0144] In step 6, according to the time-frequency domain feature matrix, a multi-scale time series classification model is obtained; Specifically, first, a classification model with preset parameters is established, where the model structure is a stacked 4-layer TimeMixer block, each block contains dilated convolution (dilation factor 2) and local attention mechanism to extract multi-scale time series features; 8-head self-attention mechanism models long-range dependencies and weights and fuses features of different time steps; Pre-trained based on the UCI-HAR public vibration data set and transferred to the subway scenario for fine-tuning; Training strategy: Focal Loss (α = 0.25, γ = 2) alleviates class imbalance; AdamW (initial learning rate 3e-4, weight decay 1e-5); In the fine-tuning stage, the first 3 layers of the TimeMixer block are frozen, and only the top layer and the classifier are trained.

[0145] In this stage, by constructing a multi-scale time series analysis model, refined feature extraction and classification of vibration signals are realized, and discriminative features are fused. The specific implementation process includes:

[0146] (1)Standardize the input data:

[0147] ;

[0148] where μ is the mean, σ is the standard deviation, is the standard value, and x i is an element in the time-frequency domain feature matrix.

[0149] (2)Multi-scale feature extraction, dilated convolution operation to obtain multi-scale eigenvalues:

[0150] ;

[0151] where d is the dilation factor, J is the convolution kernel size, y[t] is the output feature at time point t after the convolution operation, w[k] is the weight parameter at the k-th position of the convolution kernel, and x[t] is the eigenvalue of the element in the time-frequency domain feature matrix at time point t.

[0152] When the output feature of the convolution operation of the vibration signal at time point t passes through multiple TimeMixer blocks, the original vibration mode (such as the low-frequency component of subway vibration) is retained through residual connection, and at the same time, new local temporal features (such as the impact waveform of excavation vibration) are added; in adversarial domain adaptation, the residual connection ensures that the vibration feature differences between the source domain (concrete track of Line A) and the target domain (steel rail of Line B) are effectively modeled, avoiding the loss of key information during the domain alignment process; by fusing shallow details (such as the signal after high-frequency noise suppression) and deep abstract features (such as spatio-temporal dependence), the discriminative ability of the model for intrusion events such as "climbing" and "excavation" is enhanced. The residual connection formula is:

[0153] ;

[0154] where y l is the output of the current convolution layer, y l-1 is the output of the previous convolution layer (the input of the current convolution layer), and f(y l-1 ) is the processing result of the current convolution layer on the input.

[0155] (3)Calculate the attention weights to obtain the attention-weighted features; the attention weights are the core components of the attention mechanism, which are used to dynamically allocate the importance weights of different input features. Its functions include: automatically focusing on the key regions in the input data (such as specific time points or frequency bands of vibration signals) according to the task requirements, suppressing irrelevant noises; capturing the correlations between long-distance features (such as the continuous waveform of subway vibration and the intermittent pattern of climbing vibration); extracting complementary features from different subspaces (such as time domain, frequency domain, energy domain) through the multi-head attention mechanism to enhance the expression ability of the model; the calculation formula is:

[0156] ;

[0157] where , T is the parameter for adjusting the weight distribution, α ij is the attention weight of the i-th query to the j-th key, d k is the dimension of the key vector, M j is the key vector at the j-th time step, is the transpose of the query vector at the i-th time step, e ik is the set of similarities between the query and all keys.

[0158] (4) Perform bidirectional temporal modeling on the attention-weighted features to obtain the prediction probability; where the gating mechanism:

[0159] ;

[0160] ;

[0161] where p t is the output of the forget gate, and its meaning is the prediction probability; i t is the output of the input gate; σ is the activation function, preferably the sigmoid function; W f 、W i are weight matrices, corresponding to the weight parameters of the forget gate and the input gate respectively; h t-1 is the hidden state at the previous moment; x t is the eigenvalue of the element in the time-frequency domain feature matrix at time point t; b f 、b i are bias terms, corresponding to the bias parameters of the forget gate and the input gate respectively, and are used to adjust the activation threshold of the gate.

[0162] (5) During the model training process, due to the significant difference in the sample sizes between subway vibrations (strong persistence, many samples) and climbing / digging vibrations (intermittent, few samples), there is a problem of class imbalance; at the same time, the vibration feature distributions of different subway lines (such as concrete track Line A and steel rail Line B) are different, that is, there is also a cross-domain adaptation problem. Therefore, it is necessary to calculate the difference between the prediction result and the true label through a loss function (such as focal loss), clarify the optimization objective, and convert the loss value into a gradient signal to drive the update of model parameters, gradually improving the classification accuracy, so as to obtain a multi-scale temporal classification model. In the multi-scale temporal classification model, the loss function is used throughout the entire training process, and the loss value calculation formula is:

[0163] ;

[0164] where p t is the prediction probability of the model for the correct class; α t is the loss weight for balancing different classes; γ is the weight for adjusting easy and hard samples; is the final loss value after integrating p t 、α t 、γ, and is used for backpropagation to optimize the model. After obtaining the loss value , the optimizer can be used to update the parameters according to the gradient.

[0165] Regarding the vibration characteristic differences of different subway lines (such as the concrete track of Line A and the steel rail of Line B), adversarial domain adaptation is adopted: Domain discriminator: A fully connected network (2 layers, with 128 nodes in each layer) is used to distinguish the source domain (Line A) from the target domain (Line B); Gradient Reversal Layer: Insert a GRL after the feature extractor to maximize the loss of the domain discriminator to confuse domain features; Joint training: Alternately optimize the classification loss (cross-entropy) and the domain discrimination loss (adversarial loss). Finally, the test accuracy of the target domain is increased by 12.3%. The key technologies in this stage are implemented as follows:

[0166] (1) Maximum Mean Discrepancy (MMD): MMD is used to quantify the difference between two data distributions (such as the source domain and the target domain). By mapping the data to a high-dimensional Reproducing Kernel Hilbert Space (RKHS) and calculating the mean distance, MMD can determine whether the distributions are similar. In optical cable intrusion detection, MMD guides the adversarial domain adaptation training, minimizes the distribution difference of the vibration characteristics between the source domain (the concrete track of Line A) and the target domain (the steel rail of Line B), forces the model to learn domain-independent common features, thereby improving the generalization ability in cross-line scenarios and solving the problem of decreased classification performance caused by environmental differences. The calculation formula is:

[0167] ;

[0168] where MMD is the Maximum Mean Discrepancy, n s is the number of source domain samples, is the kernel space mapping function of the source domain samples, is the kernel space mapping function of the target domain samples.

[0169] (2) Adversarial loss: In the multi-scale time series classification model, the role of calculating the adversarial loss is to measure the domain alignment effect. By maximizing the loss of the domain discriminator to confuse domain features, it forces the model to learn domain-independent common features, thereby reducing the distribution difference of the vibration characteristics between the source domain (such as the concrete track of Line A) and the target domain (such as the steel rail of Line B), improving the generalization ability of the model in different subway line scenarios, and solving the problem of decreased classification performance caused by environmental differences. The calculation formula is:

[0170] ;

[0171] where L adv is the adversarial loss; is the expectation of the target domain samples to ensure that the loss calculation covers all target data; D(G(x)) is the discriminant function, used to estimate the source domain probability of the target domain features and measure the domain alignment effect; G(x) is the feature extraction function.

[0172] (3) Contrastive loss: The core of contrastive loss is to shorten the feature distance of similar samples and push the feature distance of unrelated samples to make the model learn more discriminative feature expressions. In optical cable intrusion detection, it can be used to enhance the feature consistency of similar vibration signals (such as the same intrusion type) and distinguish the feature differences of different types of signals, thereby improving the classification accuracy and robustness of the model for intrusion events. Especially in small sample or noise interference scenarios, it can effectively alleviate the overfitting problem and enhance the interpretability of the feature space. The calculation formula is:

[0173] ;

[0174] Among them, L cont is the contrast loss value; y ij is the sample pair label; d ij is the Euclidean distance of sample pairs in the feature space; m is the margin threshold.

[0175] Channel pruning (removing 20% ​​redundant convolution kernels) and dynamic quantization are performed on the multi-scale time series analysis model, and the model volume is reduced to 35% of the original size; Edge deployment: Equipped with NVIDIA Jetson Xavier NX edge device, integrated inference engine, single inference delay ≤8ms; Dynamic threshold alarm: According to the environmental noise level (such as increased noise in rainy and snowy weather), the classification confidence threshold is adaptively adjusted (default 0.9, reduced to 0.7 when the noise is strong); Alarm rule: trigger a real-time alarm when 3 consecutive frames are detected as "mining" or a single frame "climbing" confidence > 0.95. The key technologies implemented in this stage are as follows:

[0176] (1) Model quantization and linear quantization: Model quantization converts high-precision floating-point parameters (such as 32 bits) into low-precision values ​​(such as 8-bit integers), compressing the model size and improving computing efficiency, making it suitable for mobile terminals or edge devices. Linear quantization is a common method that maps the floating-point range to discrete integer values ​​based on uniform intervals. While reducing storage and computing power consumption, it maintains the numerical distribution characteristics as much as possible, balances the model inference speed and accuracy loss, and thus optimizes deployment performance in resource-constrained scenarios.

[0177] The calculation formula is:

[0178] ;

[0179] in, , Q(x) is the integer value after quantization; round(x / Δ) is the discrete mapping function from floating point to integer; Δ is the quantization step size; x max 、x min is the dynamic range of the floating-point weights, ensuring that quantization covers all data; b is the number of quantization bits;

[0180] (2)Dynamic Threshold Adjustment: Dynamic threshold adjustment adaptively updates the classification / detection threshold by analyzing the data distribution or model output in real time, addressing the false alarm or missed detection issues caused by fixed thresholds due to environmental changes (such as noise fluctuations and signal offsets). In optical cable intrusion detection, it can dynamically optimize the decision boundary based on the characteristics of vibration signals, balancing sensitivity and specificity, enabling the model to maintain stable performance in complex scenarios (such as different lines and weather conditions), and improving the recognition accuracy of intrusion events and the system robustness. The calculation formula is as follows:

[0181] ;

[0182] where τ is the dynamically adjusted threshold; τ0 is the default initial threshold; α is the noise sensitivity coefficient, and the larger α is, the more sensitive the threshold is to noise changes; N is the current environmental noise intensity; N0 is the noise reference threshold.

[0183] (3)Real-time Processing Delay: Real-time processing delay refers to the time interval from when the system receives data to when it outputs the result, which directly affects the response speed and decision-making timeliness. In scenarios such as optical cable intrusion detection, low delay can quickly identify anomalies and trigger alarms, reducing the risk of missed detection or false alarms of intrusion events and ensuring the real-time nature of the security system. By optimizing the algorithm complexity, hardware acceleration, or data transmission efficiency, the delay can be reduced, avoiding data backlog and decision failure caused by processing lag, thereby improving the system reliability and user experience. The calculation formula is as follows:

[0184] ;

[0185] where t total is the real-time processing delay, t preprocess is the preprocessing time consumption, t feature is the feature extraction time consumption, and t inference is the model inference time consumption.

[0186] Through the above technical solutions of this embodiment, by acquiring vibration signals and performing preprocessing to remove noise and other interferences, and then extracting features to obtain high-dimensional feature vectors, the analysis of vibration signals becomes more detailed and accurate, enabling precise identification of different types and intensities of vibrations; determining classification data based on the high-dimensional feature vectors, thereby determining the vibration state of the optical cable and controlling the operating state of the alarm device, realizing intelligent judgment and decision-making. It can automatically decide whether to issue an alarm based on the analysis results without manual intervention, improving the timeliness and accuracy of monitoring, reducing the influence of human factors; the entire process can quickly process vibration signals, and once an abnormal situation is detected, it can promptly control the alarm device to issue an alarm, enabling relevant personnel to take measures quickly, avoiding damage to the optical cable, and ensuring the safe and stable operation of the communication network.

[0187] In an alternative embodiment of the present invention, in step 15, determining the vibration state of the optical cable according to the classification data may include:

[0188] Step 151, parsing the classification data to obtain a structured classification result;

[0189] Step 152, making a logical judgment on the classification result according to a preset rule to obtain a preliminary vibration state;

[0190] Step 153, performing false alarm suppression processing on the preliminary vibration state according to historical data, and outputting the optimized vibration state.

[0191] In this embodiment, first, the format of the classification data is parsed, the category distribution of each time window is counted, key information is extracted, and a structured classification result is obtained, such as: {category: "intrusion event", confidence: 0.85, time window: t1~t2}.

[0192] If the classification confidence exceeds a preset threshold, such as confidence > 0.8, it is marked as a valid event; check whether the classification results of multiple time windows are continuous to exclude instantaneous noise interference. For example, if three consecutive windows are all marked as "intrusion event"; combine the spatial positioning data of the fiber optic sensor, such as the physical location where the event occurs, to verify whether it matches the known sensitive area, and finally output the preliminary vibration state, such as: {status: "suspected intrusion", location: "pile number K10+200", confidence: 0.9}.

[0193] After obtaining the preliminary vibration state, check whether the current event is similar to the historical false alarm pattern (such as noise during a specific construction period); if the system deploys multiple fiber optic sensors, cross-verify the classification results of other sensors; then dynamically adjust the decision threshold according to the environmental noise baseline, and output the optimized vibration state, such as: {status: "confirmed intrusion", location: "pile number K10+200", confidence: 0.95}. If the status is "intrusion", trigger an alarm and record the event details (time, location, confidence); if the status is "non-intrusion", only record the log for subsequent analysis.

[0194] As Figure 3 shown, an embodiment of the present invention further provides an optical cable intrusion detection device 30 based on distributed fiber optic vibration sensing, including:

[0195] An acquisition module 31, configured to acquire a first vibration signal within a preset amplitude range;

[0196] A processing module 32, configured to perform frequency filtering on the first vibration signal to obtain a second vibration signal; perform feature extraction on the second vibration signal to obtain a high-dimensional feature vector; perform multi-scale time series classification on the high-dimensional feature vector to obtain classification data;

[0197] A determination module 33, configured to determine the vibration state of the optical cable according to the classification data.

[0198] Optionally, the acquisition module 31 is specifically configured to:

[0199] Obtain an original vibration signal through a distributed optical fiber vibration sensor installed along the subway track at a preset distance;

[0200] Perform amplitude screening on the original vibration signal according to a preset dynamic amplitude threshold to obtain a first vibration signal.

[0201] Optionally, the processing module 32 is specifically configured to:

[0202] Obtain a noise coefficient;

[0203] According to the noise coefficient, decompose the first vibration signal into frequency bands to obtain a vibration sequence signal including multiple frequency band vibration signals;

[0204] Perform reconstruction processing on the noise coefficient to obtain a target noise coefficient;

[0205] According to the target noise coefficient, perform frequency filtering on the vibration sequence signal to obtain a second vibration signal.

[0206] Optionally, the processing module 32 is further specifically configured to:

[0207] Perform frame addition and windowing on the second vibration signal to obtain a short-time frame after windowing;

[0208] Perform time-frequency domain feature extraction on the short-time frame after windowing to obtain a time-frequency feature matrix;

[0209] Perform embedding coding on the time-frequency feature matrix to obtain a high-dimensional feature vector.

[0210] Optionally, the processing module 32 is further specifically configured to:

[0211] Input the high-dimensional feature vector into a trained multi-scale time series classification model for processing to obtain classification data.

[0212] Optionally, the multi-scale time series classification model is trained through the following process:

[0213] Obtain an original vibration sample signal through a distributed optical fiber vibration sensor installed along the subway track at a preset distance;

[0214] Perform amplitude screening processing on the original vibration sample signal to obtain a first vibration sample signal;

[0215] Perform frequency filtering processing on the first vibration sample signal to obtain a second vibration sample signal;

[0216] Classify the second vibration signal according to the subway vibration sample type label, the excavation vibration sample type label, and the climbing vibration type label to obtain a vibration event sample library;

[0217] Extract features from the sample data in the vibration event sample library to obtain a time-frequency domain feature matrix;

[0218] Obtain a multi-scale time series classification model according to the time-frequency domain feature matrix.

[0219] Optionally, process the time-frequency domain feature matrix to obtain fusion discriminative features, including:

[0220] Perform multi-scale time series feature extraction on the time-frequency domain feature matrix to obtain multi-scale eigenvalues;

[0221] Perform spatio-temporal attention weighting on the multi-scale eigenvalues to obtain attention-weighted features;

[0222] Perform bidirectional time series modeling processing on the attention-weighted features to obtain prediction probabilities;

[0223] According to the prediction probabilities, perform focal loss optimization and parameter update on the classification model with preset parameters to obtain a multi-scale time series classification model.

[0224] It should be noted that this device corresponds to the above method. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0225] As Figure 4 shown, an embodiment of the present invention further provides a computing device 40, including a processor 41, a memory 42, a program or instruction stored on the memory 42 and executable on the processor 41. When the program or instruction is executed by the processor 41, each process of the above embodiment of the optical cable intrusion detection method based on distributed optical fiber vibration sensing is implemented and can achieve the same technical effects. To avoid repetition, it will not be elaborated here. It should be noted that the computing device in the embodiment of the present invention includes the above-mentioned mobile electronic device and non-mobile electronic device.

[0226] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0227] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0228] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings, direct couplings, or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0229] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0230] In addition, the functional units in each embodiment of the present invention can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0231] If the described functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0232] In addition, it should be noted that in the device and method of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed chronologically in the described order, but it is not necessary to be executed chronologically. Certain steps can be executed in parallel or independently of each other. For those of ordinary skill in the art, it is understandable that all or any steps or components of the method and device of the present invention can be implemented in any computing device (including processors, storage media, etc.) or a network of computing devices in the form of hardware, firmware, software, or a combination thereof, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0233] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the object of the present invention can also be achieved only by providing a program product containing program code for implementing the method or device. That is to say, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be noted that in the device and method of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed chronologically in the described order, but it is not necessary to be executed chronologically. Certain steps can be executed in parallel or independently of each other.

[0234] The above are the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting optical cable intrusion based on distributed optical fiber vibration sensing, characterized in that: include: Acquire a first vibration signal within a preset amplitude range; Performing frequency filtering on the first vibration signal to obtain a second vibration signal; Performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector; Performing multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data; The vibration state of the optical cable is determined according to the classification data.

2. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 1 is characterized in that: The obtaining of a first vibration signal within a preset amplitude range comprises: The original vibration signal is obtained by installing distributed optical fiber vibration sensors along the subway track at preset distances; According to a preset dynamic amplitude threshold, the original vibration signal is subjected to amplitude screening processing to obtain a first vibration signal.

3. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 1 is characterized in that: Performing frequency filtering on the first vibration signal to obtain a second vibration signal includes: Get the noise figure; According to the noise coefficient, the first vibration signal is subjected to frequency band decomposition to obtain a vibration sequence signal including vibration signals of multiple frequency bands; Reconstructing the noise coefficient to obtain a target noise coefficient; According to the target noise coefficient, the vibration sequence signal is subjected to frequency filtering processing to obtain a second vibration signal.

4. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 1 is characterized in that: Performing feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector includes: Performing frame-by-frame windowing processing on the second vibration signal to obtain a windowed short-time frame; Performing time-frequency domain feature extraction processing on the windowed short time frame to obtain a time-frequency feature matrix; The time-frequency feature matrix is ​​subjected to embedded coding processing to obtain a high-dimensional feature vector.

5. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 1 is characterized in that: The high-dimensional feature vector is subjected to multi-scale time series classification processing to obtain classification data, including: The high-dimensional feature vector is input into a trained multi-scale time series classification model for processing to obtain classification data.

6. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 5 is characterized in that: The multi-scale time series classification model is trained through the following process: The original vibration sample signal is obtained by installing distributed optical fiber vibration sensors along the subway track at a preset distance; Performing amplitude screening processing on the original vibration sample signal to obtain a first vibration sample signal; Performing frequency filtering on the first vibration sample signal to obtain a second vibration sample signal; According to the subway vibration sample type label, the mining vibration sample type label and the climbing vibration type label, the second vibration signal is classified and processed to obtain a vibration event sample library; Extracting features from sample data in the vibration event sample library to obtain a time-frequency domain feature matrix; According to the time-frequency domain feature matrix, a multi-scale time series classification model is obtained.

7. The optical cable intrusion detection method based on distributed optical fiber vibration sensing according to claim 6 is characterized in that: According to the time-frequency domain feature matrix, a multi-scale time series classification model is obtained, including: Performing multi-scale time series feature extraction on the time-frequency domain feature matrix to obtain multi-scale eigenvalues; Performing spatiotemporal attention weighting on the multi-scale feature values ​​to obtain attention weighted features; Performing bidirectional temporal modeling processing on the attention weighted features to obtain a prediction probability; According to the predicted probability, the classification model with preset parameters is subjected to focal loss optimization and parameter update to obtain a multi-scale time series classification model.

8. An optical cable intrusion detection device based on distributed optical fiber vibration sensing, characterized in that: include: An acquisition module, used to acquire a first vibration signal within a preset amplitude range; a processing module, configured to perform frequency filtering processing on the first vibration signal to obtain a second vibration signal; perform feature extraction processing on the second vibration signal to obtain a high-dimensional feature vector; and perform multi-scale time series classification processing on the high-dimensional feature vector to obtain classification data; The determination module is used to determine the vibration state of the optical cable according to the classification data.

9. A computing device, characterized in that include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is performed.

10. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images