Multi-scene distributed optical fiber acoustic sensing vibration data fusion early warning platform

Abstract

The invention discloses a multi-scene distributed optical fiber acoustic sensing vibration data fusion early warning platform, and relates to the technical field of scene vibration early warning. A vibration analysis and sorting module analyzes collected vibration signals by using a multi-source data fusion algorithm, identifies and classifies the characteristics of vibration sources, automatically performs priority sorting on the vibration sources, and performs early warning on the vibration sources; an area with potential threats is selected for secondary analysis and real-time monitoring; and the strategy generation and early warning module automatically predicts the risk level of each region of the subway according to the vibration source identification and analysis result, and formulates a coping strategy based on the priority and the risk level. According to the early warning platform, through multi-sensor data fusion and an intelligent priority ranking mechanism, the precision and timeliness of vibration monitoring are remarkably improved, and real-time dynamic early warning and intelligent resource management are realized, so that efficient and safe operation of a subway system is ensured.

Description

Technical Field

[0001] This invention relates to the field of scene vibration early warning technology, specifically to a multi-scene distributed fiber optic acoustic sensing vibration data fusion early warning platform. Background Technology

[0002] During subway operation, various structures such as tracks, tunnels, trains, and the ground will vibrate. These vibrations have a certain impact on the safe operation of the subway and the surrounding environment. Monitoring subway vibration data helps to grasp the subway's operating status in real time, detect potential equipment failures or changes in the external environment, and issue timely warnings to ensure safe operation. The early warning platform is used to receive and analyze the monitored data in real time, predict potential problems that subway vibration may cause, and issue timely warnings to relevant personnel. By building an early warning platform, an alarm can be automatically issued when the vibration exceeds a certain threshold, helping subway managers to take timely measures to avoid accidents.

[0003] The existing technology has the following drawbacks: Existing early warning platforms lack effective data fusion and prioritization mechanisms when facing different types of vibration sources, making it impossible to quickly identify and classify vibration sources. In particular, they cannot make reasonable distinctions between high-risk and low-risk areas. Therefore, when a fault or anomaly occurs, early warnings are often not issued in a timely and accurate manner, resulting in low response efficiency and even missing the early warning opportunity, which may endanger the safe operation of the subway system.

[0004] Based on this, the present invention proposes a multi-scenario distributed fiber optic acoustic sensor vibration data fusion early warning platform. The early warning platform significantly improves the accuracy and timeliness of vibration monitoring through multi-sensor data fusion and intelligent priority sorting mechanism, realizes real-time dynamic early warning and intelligent resource management, thereby ensuring the efficient and safe operation of the subway system. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform to address the shortcomings in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform, comprising a data acquisition and processing module, a vibration analysis and sorting module, and a strategy generation and early warning module; Data acquisition and processing module: Constructs a comprehensive environmental perception model in the subway environment, integrates data from multiple sensors, and combines external environmental factors to comprehensively monitor the subway's operating status, monitors the subway's vibration signals in real time, and triggers a self-check and analysis mechanism when the preset vibration warning value is reached; Vibration analysis and ranking module: Utilizes multi-source data fusion algorithms to analyze the collected vibration signals, identify and classify the characteristics of vibration sources, automatically prioritize vibration sources, and select areas with potential threats for secondary analysis and real-time monitoring; Strategy generation and early warning module: Based on the vibration source identification and analysis results, it automatically predicts the risk level of each area of ​​the subway, formulates response strategies based on priority and risk level, automatically adjusts the power allocation in the early warning and response process according to the current platform redundancy capacity, and if energy shortage is detected, it pauses the response process, records the currently completed early warning tasks, and continues the remaining early warning work after recovery.

[0007] Preferably, the vibration analysis and sorting module receives pre-processed vibration signals and related environmental data from the data acquisition and processing module; Similar features extracted from different sensors are aligned and stitched together to form a comprehensive feature matrix, which contains complementary information from multiple sensors. Using the fused feature vector as input, the type and cause of vibration sources are identified; The extracted vibration feature vectors are input into a pre-trained classification model. The classification model outputs the probability distribution of the vibration source belonging to a certain category through multi-layer feature mapping and nonlinear transformation. The category with the highest probability is selected as the preliminary identification result of the vibration source.

[0008] Preferably, the vibration analysis and sorting module analyzes the attribute parameters of the vibration source. After completing the identification and classification of the vibration source, it automatically sorts and prioritizes all detected vibration sources in the current monitoring area based on risk assessment indicators and priority rules. A vibration source risk scoring model is constructed for ranking. A comprehensive risk score is calculated for each vibration source, and they are ranked in descending order according to the comprehensive risk score. The higher the score, the greater the potential threat that the vibration source poses to the safety of subway operation, and the higher its priority.

[0009] Preferably, the vibration source risk scoring model includes vibration intensity, whether the vibration frequency is within the sensitive frequency band of the equipment, the spatial location of the vibration source, the persistence and development trend of the vibration source, the type of vibration source, and the environmental background.

[0010] Preferably, the vibration analysis and sorting module calculates the time-domain statistical characteristics of the vibration signal, including peak amplitude, root mean square, peak-to-peak value, vibration duration, signal rise time, and signal zero-crossing rate.

[0011] Preferably, the strategy generation and early warning module determines the area directly affected by the vibration source and the associated area that produces secondary effects based on the spatial location of the vibration source, combined with the subway line topology, equipment layout diagram and spatial index model. Based on a historical vibration propagation case library, the intensity, frequency characteristics, and spatial and structural propagation characteristics of the current vibration source are assessed, and its potential impact on the surrounding area is predicted. Based on vibration propagation analysis, assess the vibration intensity and frequency composition of the area to determine whether it is close to the sensitive range of the structure or the operating frequency of the equipment in the area. After clarifying the risk level of each region, based on the priority ranking of vibration sources and the regional risk level, a strategy matching and generation logic is used to formulate differentiated and executable response strategies for each risk region or vibration source.

[0012] Preferably, the strategy generation and early warning module assesses the intensity, frequency characteristics, and spatial and structural propagation characteristics of the current vibration source, and predicts its potential impact on the surrounding area, including the following steps: Construct a regional risk index model and calculate multiple risk factors; Based on multiple risk factors, a comprehensive risk index is calculated for each subway area, and the area is classified into the corresponding risk level category according to the preset risk classification standards.

[0013] Preferably, the strategy generation and early warning module formulates differentiated and executable response strategies for each risk area or vibration source, including the following steps: For high-risk areas, strategies include immediately initiating on-site inspections, suspending operation of relevant lines, notifying maintenance teams to be on standby, and switching to backup equipment. For medium-risk areas, strategies include increasing monitoring frequency, recording vibration evolution trends, and sending warning notices to construction units; For low-risk or routine vibrations, the strategy is continuous monitoring or data archiving.

[0014] Preferably, the vibration analysis and sorting module identifies and classifies the characteristics of vibration sources, analyzes the frequency, waveform and intensity of vibration through fiber optic sensing data and multidimensional data from accelerometers, and then determines whether the vibration is caused by equipment failure, external factors or human factors.

[0015] Preferably, the data acquisition and processing module integrates data from multiple sensors, including a fiber optic acoustic sensor, an accelerometer, and a temperature and humidity sensor. A network of fiber optic acoustic sensors distributed in subway tunnels, along the track, in station structures, and in equipment areas captures sound wave / vibration signals propagating along the optical fiber in real time. Accelerometers are used to acquire local high-frequency vibration responses, while temperature and humidity sensors are used to simultaneously collect ambient temperature and relative humidity data.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention overcomes the limitations of traditional technologies in subway vibration monitoring by employing multi-scenario distributed fiber optic acoustic sensing technology and data fusion algorithms. First, by integrating data from multiple sensors and considering the influence of external environmental factors, a comprehensive environmental perception model is constructed, enabling the system to monitor the subway's operational status in real time and avoiding the impact of external interference on vibration data analysis. Second, utilizing multi-source data fusion technology and a vibration source priority ranking mechanism, the system can effectively identify, classify, and prioritize potential threat areas from vibration signals from different sources, thereby improving the accuracy and timeliness of vibration source analysis. Furthermore, the strategy generation and early warning module automatically adjusts risk level predictions based on vibration source identification results and rationally allocates redundant energy to avoid excessive resource consumption. If energy shortage is detected, the system can pause its response and record the current progress, resuming the early warning task once energy is restored, ensuring system stability and efficiency. In summary, this solution effectively improves the response speed, accuracy, and resource utilization efficiency of the subway vibration monitoring system, ensuring the safe operation of the subway system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a system architecture diagram of the early warning platform of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: Please see Figure 1As shown, this embodiment provides a multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform, including a data acquisition and processing module, a vibration analysis and sorting module, and a strategy generation and early warning module; Data acquisition and processing module: In the subway environment, a comprehensive environmental perception model is constructed, integrating data from multiple sensors (such as fiber optic acoustic sensors, accelerometers, temperature and humidity sensors, etc.), and combining external environmental factors (such as climate, earthquakes, etc.) to comprehensively monitor the subway's operating status, monitor the subway's vibration signals in real time, and trigger self-checking and analysis mechanisms when the preset vibration warning value is reached to ensure the accuracy and timeliness of the data. The vibration signal is then sent to the vibration analysis and sorting module.

[0021] Vibration Analysis and Ranking Module: Utilizing multi-source data fusion algorithms, this module performs detailed analysis of the acquired vibration signals, identifying and classifying the characteristics of vibration sources. For example, by analyzing multi-dimensional data from fiber optic sensors and accelerometers, it analyzes the frequency, waveform, and intensity of vibrations to determine whether the vibration is caused by equipment failure, external factors (such as weather changes, earthquakes, etc.), or human error. It automatically prioritizes vibration sources, selecting areas with potential threats for secondary analysis and real-time monitoring to ensure the accuracy and timeliness of vibration source analysis. The vibration source identification and analysis results are then sent to the strategy generation and early warning module.

[0022] Strategy Generation and Early Warning Module: Based on vibration source identification and analysis results, this module automatically predicts the risk level of each area of ​​the subway and formulates response strategies based on priority and risk level. For high-risk areas, it issues real-time warnings and sends detailed vibration data analysis reports to subway management personnel, providing a basis for decision-making. Simultaneously, to avoid excessive energy and resource consumption, it automatically adjusts power allocation during the early warning and response processes based on the platform's current redundancy capacity, ensuring that the power consumption of each response round does not exceed the preset energy capacity. If energy shortage is detected, the response process is paused, the completed early warning tasks are recorded, and the remaining early warning work continues after resumption, ensuring stability and efficiency.

[0023] This application addresses the limitations of traditional technologies in subway vibration monitoring by employing multi-scenario distributed fiber optic acoustic sensing technology and data fusion algorithms. Firstly, by integrating data from multiple sensors and considering the influence of external environmental factors, a comprehensive environmental perception model is constructed, enabling the system to monitor the subway's operational status in real time and avoiding the impact of external interference on vibration data analysis. Secondly, utilizing multi-source data fusion technology and a vibration source priority ranking mechanism, the system can effectively identify, classify, and prioritize potential threat areas from vibration signals from different sources, thereby improving the accuracy and timeliness of vibration source analysis. Furthermore, the strategy generation and early warning module automatically adjusts risk level predictions based on vibration source identification results and rationally allocates redundant energy, avoiding excessive resource consumption. If energy shortage is detected, the system can pause its response and record the current progress, resuming the early warning task once energy is restored, ensuring system stability and efficiency. In summary, this solution effectively improves the response speed, accuracy, and resource utilization efficiency of the subway vibration monitoring system, ensuring the safe operation of the subway system.

[0024] The operation process of the early warning platform is as follows: In the subway environment, a comprehensive environmental perception model is constructed, integrating data from multiple sensors (such as fiber optic acoustic sensors, accelerometers, temperature and humidity sensors, etc.) and combining external environmental factors (such as climate, earthquakes, etc.) to comprehensively monitor the subway's operational status. Through data fusion technology, information from different sensors is integrated to monitor subway vibration in real time. When a preset vibration warning value is reached, the system's self-checking and analysis mechanisms are triggered to ensure data accuracy and timeliness. Multi-source data fusion algorithms are used to analyze the collected vibration signals in detail, identifying and classifying the characteristics of vibration sources. For example, by analyzing multi-dimensional data from fiber optic sensors and accelerometers, the frequency, waveform, and intensity of vibration are analyzed to determine whether the vibration is caused by equipment failure, external factors (such as weather changes, earthquakes, etc.), or human factors. The system automatically prioritizes vibration sources, selecting areas with potential threats for secondary analysis and real-time monitoring to ensure the accuracy and timeliness of vibration source analysis. Based on the vibration source identification and analysis results, the system automatically predicts the risk level of each area of ​​the subway and formulates response strategies based on priority and risk level. For high-risk areas, the system will issue real-time warnings and send detailed vibration data analysis reports to subway management personnel, providing a basis for decision-making. Simultaneously, to avoid excessive energy consumption and system resource depletion, the system will automatically adjust power allocation during the warning and response processes based on the current platform redundancy capacity, ensuring that the power consumption of each response round does not exceed the preset energy capacity. If the system detects energy shortages, the response process will be paused, the completed warning tasks will be recorded, and the remaining warning work will resume after the system recovers, ensuring system stability and efficiency.

[0025] In the subway environment, the data acquisition and processing module constructs a comprehensive environmental perception model, integrating data from multiple sensors (such as fiber optic acoustic sensors, accelerometers, temperature and humidity sensors, etc.), and combining external environmental factors (such as climate, earthquakes, etc.) to comprehensively monitor the subway's operating status, monitor the subway's vibration signals in real time, and trigger self-checking and analysis mechanisms when the preset vibration warning value is reached to ensure the accuracy and timeliness of the data.

[0026] In the subway environment, the data acquisition and processing module plays a core role in building a comprehensive environmental perception model. Its primary task is to efficiently integrate and preprocess data from multi-source heterogeneous sensors, enabling comprehensive and high-precision real-time monitoring of the subway's operational status. This module deploys various types of sensors, including distributed-acoustic sensors (DAS), accelerometers, and temperature and humidity sensors, combined with synchronous sensing of external environmental variables (such as meteorological conditions and seismic activity), to construct a multi-dimensional and highly timely environmental perception system. This provides a solid data foundation for subsequent vibration signal analysis and anomaly detection.

[0027] A network of fiber optic acoustic sensors distributed throughout subway tunnels, along the track, station structures, and key equipment areas captures acoustic / vibration signals propagating along optical fibers in real time. These signals can reflect minute vibration changes in the track, tunnel walls, or surrounding soil with extremely high spatial resolution (typically reaching meter or even sub-meter levels). Simultaneously, accelerometers are deployed on critical equipment (such as track fasteners, turnout mechanisms, and train bogies) and structural nodes to acquire local high-frequency vibration responses, particularly transient vibration information under equipment operation or structural stress. Furthermore, temperature and humidity sensors are used to simultaneously collect ambient temperature and relative humidity data to eliminate or correct false vibration signals that may be caused by thermal expansion and contraction, humidity-induced structural deformation, or changes in material properties.

[0028] All sensors transmit raw data in real time to edge computing units or central data aggregation nodes via standardized communication protocols (such as Modbus, CAN, LoRa, fiber optic communication, or industrial Ethernet). During this process, the system employs a time synchronization mechanism (such as PTP protocol or GPS timing) to ensure that data from different sensors are strictly aligned in the time dimension, providing a time-series consistency guarantee for subsequent multi-source data fusion.

[0029] Because sensors operate in the complex subway environment, the raw signals they collect are often affected by environmental noise, electromagnetic interference, signal attenuation, and sampling errors. Therefore, preprocessing and quality enhancement of the raw data are necessary to ensure its reliability and usability. This processing mainly includes the following sub-steps: Noise denoising: The raw signals from the fiber optic acoustic sensor and accelerometer are processed using an adaptive filtering algorithm based on a sliding window. Background noise components are estimated and removed iteratively while retaining effective vibration characteristics. For example, low-frequency steady-state environmental noise can be removed from the target frequency band by designing a band-stop filter; high-frequency random noise can be smoothed using wavelet transform or moving average methods.

[0030] Missing and outlier detection: Anomaly detection of sensor data is performed using statistical distributions (such as the 3σ principle) to identify abnormal sampling points caused by equipment failure, communication interruption, or sudden interference. Interpolation compensation is then performed based on nearby time series data (such as linear interpolation, spline interpolation, or Kalman filter prediction value filling) to ensure the continuity and integrity of time series data.

[0031] Data normalization and unit unification: Given the significant differences in the data range and physical dimensions of different sensors (e.g., acceleration is in m / s², temperature is in ℃, humidity is in %RH), the system will perform unified normalization processing on all input data (e.g., Min-Max standardization or Z-score standardization) to facilitate subsequent multi-source fusion calculations and feature comparisons.

[0032] After completing the preprocessing of data from each sensor, the system will further perform spatiotemporal alignment and fusion processing of multi-source data, aiming to map heterogeneous data from different physical sensors, different sampling frequencies, and different spatial locations to the same spatiotemporal coordinate system to form consistent comprehensive sensing information.

[0033] Using a unified timestamp as a benchmark, all sensor data are timestamped with millisecond or microsecond precision. For data sequences with time offsets, interpolation or resampling methods (such as linear interpolation or spline interpolation) are used to synchronize them to a standard time grid. Based on the physical location information of the sensors (such as geographic coordinates, tunnel mileage markers, equipment installation points, etc.), a spatial index structure (such as KD-tree or R-tree) is constructed to associate data from different sensors with corresponding subway sections, track segments, or structural units, thereby achieving precise spatial positioning and area division. After completing time and spatial alignment, the system extracts key feature parameters from each sensor (such as vibration amplitude and frequency components of fiber optic acoustic sensors; peak acceleration and vibration energy of accelerometers; real-time temperature and humidity values ​​of temperature and humidity sensors, etc.), integrating these heterogeneous features into a unified comprehensive environmental state characterization vector for subsequent anomaly detection and risk assessment.

[0034] The system will focus on vibration signals captured by fiber optic acoustic sensors and accelerometers, and calculate their key dynamic indicators in real time, including but not limited to: Vibration amplitude (PeakAmplitude); Vibration frequency (Dominant Frequency / Frequency Spectrum); Vibration duration; Vibrational energy (Energy Level); Spectral entropy, or a signal complexity metric; Based on this, the system incorporates a dynamic threshold determination mechanism. This mechanism, based on historical operating data, equipment operating parameters, and an environmental baseline model, sets an adaptively adjustable vibration alert threshold for each monitoring area. When the real-time calculated index of one or more vibration signals exceeds this threshold, the system determines it as a "potential abnormal event" and immediately triggers the following response logic: The system automatically performs self-checks on sensor channels that trigger anomalies, including signal integrity verification, communication link diagnosis, and sensor drift detection, to eliminate false alarms caused by equipment failure or data transmission errors. It retrieves multi-source data from several time windows prior to the anomaly (e.g., 5 to 30 seconds before the anomaly occurs) for retrospective analysis to identify vibration signal trends, spectral shifts, and correlations with other environmental variables (e.g., sudden temperature and humidity changes, external weather events), thereby enhancing the accuracy of anomaly detection. The system uniquely identifies the vibration source that triggers the alert (e.g., event ID, timestamp, area number), and caches relevant data fragments (raw signal, characteristic parameters, environmental context, etc.) in a high-priority data buffer.

[0035] To ensure the entire data processing flow meets the stringent real-time and accuracy requirements for subway operation safety, this module employs an Apache-Kafka-based streaming data processing framework to achieve low-latency access and real-time computation of sensor data, ensuring end-to-end latency from data acquisition to alarm triggering is controlled within milliseconds to seconds. Dual-channel redundancy deployment and cross-validation mechanisms are used for critical sensor data. When significant inconsistencies occur between primary and backup sensor data, the system will initiate data reliability assessment logic, prioritizing analysis of data from highly reliable sources.

[0036] The vibration analysis and ranking module utilizes a multi-source data fusion algorithm to perform detailed analysis of the acquired vibration signals, identifying and classifying the characteristics of vibration sources. For example, by analyzing multi-dimensional data from fiber optic sensors and accelerometers, it analyzes the frequency, waveform, and intensity of vibrations to determine whether the vibrations are caused by equipment failure, external factors (such as weather changes, earthquakes, etc.), or human factors. It automatically prioritizes vibration sources, selecting areas with potential threats for secondary analysis and real-time monitoring, ensuring the accuracy and timeliness of vibration source analysis.

[0037] The vibration analysis and ranking module is the core analysis hub of the entire distributed fiber optic acoustic sensing vibration data fusion and early warning platform. Its main function is to perform in-depth analysis of vibration signals and multi-source environmental data from the data acquisition and processing module, accurately identify the source characteristics, categories, and causes of vibration sources, and prioritize and classify vibration sources based on multi-dimensional features, thereby achieving targeted monitoring and rapid response to potentially high-risk areas. This module integrates fiber optic acoustic sensing data (such as the time series, spatial distribution, and frequency domain characteristics of distributed vibration signals), accelerometer data (such as local high-frequency vibration response, vibration peak value, and energy), and auxiliary environmental information (such as temperature, humidity, meteorological, and seismic data), and uses advanced multi-source data fusion algorithms and pattern recognition technology to construct a high-precision, high-response, and highly robust intelligent vibration source analysis system.

[0038] The system receives pre-processed vibration signals and related environmental data from the data acquisition and processing module. This data is typically in time-series format and includes multi-dimensional observations from different sensors. At this stage, the system performs multi-scale, multi-domain feature extraction operations on each vibration signal (especially fiber optic acoustic sensing and acceleration sensing data) to transform the raw signals into structured feature vectors suitable for pattern recognition and classification.

[0039] The time-domain statistical characteristics of vibration signals are calculated, including peak amplitude, root mean square (RMS), peak-to-peak value, duration, rise time, and zero-crossing rate. These indicators can intuitively reflect the intensity, suddenness, and dynamic behavior of vibration. Using the Fast Fourier Transform (FFT) method, the vibration signal is transformed from the time domain to the frequency domain, and parameters such as energy distribution, dominant frequency, bandwidth, and spectral entropy in key frequency bands are extracted to characterize the frequency composition and complexity of the vibration.

[0040] This system acquires the two-dimensional distribution characteristics of vibration signals in the time-frequency domain, capturing time-varying frequency components in non-stationary signals. It is particularly suitable for analyzing sudden, impact-type, or transient vibration sources. For distributed fiber optic acoustic sensors, the system calculates the spatial location information of the vibration source (such as event location, propagation direction, and wave velocity estimation) based on the vibration amplitude and arrival time difference at each sampling point along the fiber. It also extracts features such as spatial amplitude distribution, vibration propagation delay difference, and spatial gradient along the fiber to determine the spatial distribution characteristics and propagation path of the vibration source. These features are organized into a unified feature vector, with each sensor data point corresponding to one or more feature vectors, along with metadata such as timestamp, spatial location, and sensor type, serving as the basic input for subsequent analysis.

[0041] After obtaining the multidimensional feature vectors of each sensor, the module will further perform multi-source data fusion processing, organically integrating information from heterogeneous data sources such as fiber optic sensors, accelerometers, and environmental sensors to achieve a more comprehensive and accurate description and interpretation of the vibration source. The specific fusion logic and processing steps are as follows: Similar features (such as frequency, amplitude, and energy) extracted from different sensors are aligned and stitched together to form a comprehensive feature matrix. This matrix will contain complementary information from multiple sensors, such as the spatial continuous vibration distribution provided by fiber optic sensing and the local high-frequency detail response provided by accelerometer sensing. In some scenarios, the system can first perform independent vibration source detection and preliminary classification on the data from a single sensor (such as based on thresholds or simple models), and then perform weighted voting or confidence fusion on the judgment results of each sensor to improve the robustness of the overall judgment.

[0042] The system employs a Multilayer Perceptron (MLP) model, using the fused feature vectors as input to identify the category and cause of vibration sources. During the model training phase, labeled data (such as samples of vibration sources known to be equipment failure, earthquakes, train operation, or human construction) is used to learn the distribution patterns and distinguishing boundaries of different vibration categories in the feature space. Example of classification logic: The extracted vibration feature vectors are input into a pre-trained classification model. The model uses multi-layer feature mapping and nonlinear transformation to output the probability distribution of the vibration source belonging to a certain category (such as "equipment failure", "earthquake", "train operation", "external construction", "meteorological impact" or "unknown anomaly"). The system selects the category with the highest probability as the preliminary identification result of the vibration source.

[0043] In addition to classifying the vibration source, the system will further analyze the attribute parameters of the vibration source, such as estimating the spatial location of the vibration source (especially for fiber optic sensing), vibration initiation time, duration, propagation speed, energy level, and whether it is periodic or impactful, to provide richer decision-making basis for subsequent prioritization and risk assessment.

[0044] After identifying and classifying vibration sources, the module will automatically sort and prioritize all detected vibration sources within the current monitoring area based on a series of risk assessment indicators and priority rules. The aim is to identify the vibration sources with the greatest potential threat and requiring immediate attention and intervention, thereby optimizing the allocation of early warning resources and the formulation of response strategies. The sorting logic and processing procedure are as follows: Taking into account the following key factors, a vibration source risk scoring model for ranking is constructed: Vibration intensity (e.g., amplitude, energy); Is the vibration frequency within the equipment's sensitive frequency band? Spatial location of the vibration source (whether it is close to critical equipment, track joints, structural weak points, etc.); The persistence and development trend of the vibration source (whether it continues to increase or spread). Vibration source categories (such as earthquakes and equipment failures) generally pose a higher risk than ordinary train vibrations. Environmental context (e.g., whether there are currently high-risk external factors such as severe weather or earthquake warnings); Calculate a comprehensive risk score for each vibration source (e.g., by aggregating the above indicators through a weighted summation method), and sort them in descending order of score. The higher the score, the greater the potential threat the vibration source poses to the safety of subway operation, and the higher its priority.

[0045] After identifying the key indicators mentioned above, the system will assign an evaluation value or status identifier to each detected vibration source based on its real-time characteristics and contextual information. Furthermore, it will assign a weight coefficient to each indicator (which can be determined by expert experience or optimized through training with historical data) to reflect the degree of contribution of the indicator to the overall risk.

[0046] For each indicator (such as vibration intensity, frequency matching degree, spatial importance, etc.), the system first converts its original observation value or classification result into a comparable numerical state. For example: vibration intensity → expressed as a standardized energy value or amplitude level (e.g., low / medium / high → 1 / 2 / 3); whether it is in the sensitive frequency band of the equipment → yes = 1, no = 0, or further refined to a continuous value between 0 and 1 for the matching degree; spatial importance → determined according to the predefined "critical area map", divided into several levels (e.g., non-critical area = 1, general area = 2, critical equipment area = 3, extremely critical area = 4); vibration category → the risk category output by the classification model is mapped to the risk level value (e.g., ordinary train vibration = 1, external construction = 2, equipment failure = 4, earthquake = 5); persistence and trend → judged as "stable / slowly increasing / rapidly increasing / spreading" based on time series analysis, and mapped to the corresponding risk trend value; environmental background → judges whether there are high-risk environmental factors, such as rainstorms, earthquake early warnings, etc., and if so, adds additional risk points.

[0047] Using a weighted summation approach, the numerical values ​​of each of the above indicators are multiplied by their corresponding weights and then summed to obtain the overall risk score (RiskScore) for the vibration source. The weights reflect the importance of each indicator to the overall risk assessment; for example, "vibration intensity" and "vibration source category" may be assigned higher weights, while the weight of "environmental background" may dynamically increase under specific circumstances.

[0048] All detected vibration sources are ranked in descending order based on their calculated comprehensive risk scores. The system automatically categorizes vibration sources into several priority levels according to their scores (e.g., Priority 1 – Extremely High Risk, Priority 2 – High Risk, Priority 3 – Medium Risk, Priority 4 – Low Risk). The highest-ranked (highest-scoring) vibration source is considered the most urgent potential threat requiring immediate attention and intervention; the system will initiate focused monitoring, secondary analysis, real-time early warning, and emergency response plans for it. Lower-ranked, low-scoring vibration sources may be subject to routine monitoring or data archiving, and will not occupy high-priority response resources.

[0049] Example scenario: Monitoring data of a subway line during off-peak hours at night. Vibration source A: High-frequency impact vibration was detected at a point in the track section. The vibration amplitude was high, the energy was large, and the frequency was close to the sensitive frequency band of the turnout switch machine. The location was right at the connection of the turnout structure. The vibration continued to increase and was classified as "suspected equipment failure". There was no adverse environmental background at present.

[0050] →Overall score: 94 (Very high), Priority: 1 (Emergency response) Vibration source B: Medium amplitude vibration was detected in a tunnel section. The frequency was normal, the location was in the ordinary track area, and the classification was "train passing". There was no continuous increase, and the environmental background was normal.

[0051] →Overall score: 32 (low), priority: 4 (routine monitoring) Vibration source C: Low amplitude vibrations with frequencies close to the sensitive frequency band of power supply equipment were detected near a station. The location is adjacent to a substation. The vibrations are continuous and slowly increasing. The environmental background is "Weather warning: strong wind".

[0052] →Overall score: 77 (high), priority: 2 (high attention, early screening).

[0053] For high-priority vibration sources that rank highly, the system will automatically mark their areas as "key monitoring areas" and trigger secondary analysis processes (such as calling higher-precision sensors, extending the observation window, and combining video or image data for multimodal verification). At the same time, these areas will be included in the real-time tracking list to continuously monitor their vibration evolution trends, ensuring the accuracy of the analysis and the timeliness of the response.

[0054] Finally, the vibration analysis and sequencing module will output the following key information to the strategy generation and early warning module: The results of the vibration source classification (such as equipment failure, external earthquake, human construction, etc.); Spatial location and influence range of the vibration source; Risk assessment and prioritization of vibration sources; The relevant set of characteristic parameters (such as frequency, amplitude, energy, duration, etc.); Does the current area require secondary analysis or real-time intervention? This information will be encapsulated and transmitted in structured data formats (such as JSON, XML, or custom data protocols) to ensure that downstream modules can quickly parse it and use it to formulate accurate response strategies and early warning plans.

[0055] Based on vibration source identification and analysis, the strategy generation and early warning module automatically predicts the risk level of each area of ​​the subway and formulates response strategies based on priority and risk level. For high-risk areas, it issues real-time warnings and sends detailed vibration data analysis reports to subway management personnel, providing a basis for decision-making. Simultaneously, to avoid excessive energy and resource consumption, it automatically adjusts power allocation during the early warning and response process according to the platform's current redundancy capacity, ensuring that the power consumption of each response round does not exceed the preset energy capacity. If energy shortage is detected, the response process is paused, the completed early warning tasks are recorded, and the remaining early warning work continues after resumption, ensuring stability and efficiency.

[0056] As the decision-making center and execution front-end of the multi-scenario distributed fiber optic acoustic sensor vibration data fusion early warning platform, the strategy generation and early warning module undertakes the core function of transforming vibration source identification and analysis results into actionable risk management strategies and precise early warning commands. Based on the vibration source category, spatial location, risk score, priority ranking, and environmental context information provided by the data acquisition and processing module and the vibration analysis and ranking module, this module constructs an intelligent risk prediction model, a dynamic strategy generation mechanism, and resource optimization scheduling logic. This enables precise classification, targeted early warning, resource adaptation, and process control of vibration risks in different areas of the subway, ensuring efficient, stable, and sustainable system response while guaranteeing subway operation safety.

[0057] After receiving information such as vibration source category, risk score, priority, and spatial location from the vibration analysis and ranking module, the strategy generation and early warning module first initiates a regional-level risk assessment process. The purpose is to quantify and classify the overall risk of various functional areas of the subway (such as track sections, stations, equipment rooms, tunnel sections, and control centers) affected or potentially related to the vibration sources, providing a global perspective for subsequent strategy formulation. The specific processing logic is as follows: Based on the spatial location of the vibration source (e.g., the vibration origin point located by fiber optic sensing, or a localized area monitored by acceleration sensing), and combined with the subway line topology, equipment layout diagram, and spatial indexing model (e.g., GIS or BIM), the area directly affected by the vibration source and related areas that may have secondary impacts (e.g., adjacent track sections, shared structures, and key electromechanical equipment areas) are determined. Based on a historical vibration propagation case library, the intensity, frequency characteristics, and spatial and structural propagation characteristics of the current vibration source (e.g., vibration attenuation rate, structural resonance probability) are assessed to predict its potential impact on the surrounding area.

[0058] Calculation logic for multiple risk factors: Taking into account the following key factors, a regional risk index model is constructed: The risk level of the vibration source itself (e.g., high-risk category: equipment failure, earthquake; medium- and low-risk category: regular train operation, external construction, etc.). The functional importance of the affected area (such as core facilities like control centers, substations, and signal equipment rooms, which have higher weight); The expected intensity and frequency characteristics of vibration propagation to this region; Does the current area have any existing hidden dangers, aging equipment, or high operating load, among other overlapping risk factors? Based on the above factors, a comprehensive risk index (RiskIndex) is calculated for each subway area, and the area is classified into the corresponding risk level category according to the preset risk classification standards (such as low risk, medium risk, high risk, and emergency risk).

[0059] Key risk factors to be considered include: Based on the previous vibration source classification results, determine which risk category this vibration belongs to: High-risk categories include: equipment failure (track breakage, switch jamming), earthquakes, and precursors to structural collapse. Medium-risk categories include: external construction, heavy vehicle traffic, and localized track wear. Low-risk categories include: regular train operation, environmental vibration, and minor airflow disturbances.

[0060] If the vibration source is determined to be "equipment failure (cracked rail weld)," its basic risk level is "high"; if it is "passing ordinary train," it is "low."

[0061] Different subway zones have different functions and safety sensitivities during operation, and their risk weights vary greatly. High-weight areas, such as control centers, substations, signal equipment rooms, communication equipment rooms, and emergency power supply rooms, could be damaged, potentially leading to systemic paralysis or major safety incidents. Medium-weighted areas: such as ordinary track sections, non-critical equipment areas, and ordinary station halls; Low-weight areas: such as idle equipment rooms during non-operational periods, auxiliary rooms far from core functions, etc.

[0062] Vibration affects the "traction substation" area, whose functional importance is far greater than that of the "ordinary track section". Therefore, even if the vibration intensity is similar, its risk index should be higher.

[0063] Based on the aforementioned vibration propagation analysis, assess the actual vibration intensity (such as amplitude and energy) and frequency composition that the area may experience, and determine whether it is close to the sensitive range of the structure or the operating frequency of the equipment in the area.

[0064] Consider whether the area already has the following problems: aging equipment, fatigue damage, unrepaired potential hazards; excessive operating load (such as power supply overload, continuous operation of equipment without maintenance); environmental factors (such as humidity leading to decreased electrical insulation, high temperature affecting mechanical performance). If an equipment room already has the potential hazard of "poor heat dissipation of the power supply cabinet," and is further affected by external vibration, its overall risk will be significantly higher than that of a similar room without the potential hazard.

[0065] The multi-factor comprehensive calculation logic is as follows: Each of the above factors is assigned a corresponding evaluation level or value (e.g., high=3, medium=2, low=1). Some continuous indicators (e.g., vibration intensity, attenuation rate) can be divided into intervals or normalized measurements can be used directly. Subsequently, the system performs comprehensive weighted calculations or nonlinear aggregations on these factors based on expert rules, historical statistical weights, or lightweight machine learning models to derive the comprehensive risk index (RiskIndex) for the subway area.

[0066] Finally, the system classifies the area into the corresponding risk level category based on the preset risk classification threshold standards (e.g., 0–30 is low risk, 31–60 is medium risk, 61–85 is high risk, and 86 and above is emergency risk), and assigns it a risk label, such as "high risk - immediate response required", "medium risk - enhanced monitoring", "low risk - routine attention", etc.

[0067] Scenario: Localized vibration source in a subway track section at night in a certain city. Vibration source location: The fiber optic sensing system detected that the vibration point was located in a middle track section of Line 3 (near the tunnel exit), and was initially classified as "suspected equipment failure (loose track fasteners)". The vibration intensity was high and the frequency was concentrated in 10-15Hz.

[0068] Directly affected area: the track section where the vibration occurs; Secondary impact areas: adjacent track sections (vibration propagates along the track structure), nearby turnout areas, shared tunnel lining structures, and adjacent traction substations and signal relay equipment rooms.

[0069] Propagation characteristics analysis: Based on similar historical cases (loose track fasteners causing local impacts, vibrations attenuate slowly in the tunnel concrete structure and are easily reflected and amplified at track joints), the system predicts that the vibration may have a moderate to high level of impact on the two adjacent track sections and the substation area, especially since the substation is located on the vibration propagation path and no additional vibration isolation measures have been taken.

[0070] Substation area: High vibration source risk + extremely high functional importance + expected propagation intensity moderate to high + no known hidden dangers → Overall risk index: 88 → Emergency risk Adjacent track section: High vibration source risk + directly adjacent area + expected high vibration intensity + no special superimposed risks → Comprehensive risk index: 76 → High risk Ordinary station hall area (relatively far away): High risk of vibration source + but far away, expected intensity low + general functional importance → Overall risk index: 32 → Medium risk Response strategy orientation: The system will prioritize real-time early warning, targeted monitoring and emergency investigation of substation areas with "emergency risks" and adjacent track sections with "high risks", while maintaining routine monitoring of medium-risk areas.

[0071] After identifying the risk levels of each area, the module will, based on the priority ranking of vibration sources and the regional risk level, employ strategy matching and generation logic to formulate differentiated and executable response strategies for each risk area or vibration source, ensuring precise resource allocation and effective risk control. The strategy formulation process is as follows: The system has a built-in policy rule base, which predefines recommended responses for different risk levels, vibration categories, and regional functional types, such as: For high-risk areas (such as those where equipment failure causes strong vibrations and is located on critical track sections), strategies may include: immediately initiating on-site inspections, suspending operation of the relevant lines, notifying the maintenance team to be on standby, and switching to backup equipment. For medium-risk areas (such as periodic vibrations caused by external construction but not exceeding safety limits), strategies may include: increasing monitoring frequency, recording vibration evolution trends, and sending warning notices to construction units. For low-risk or routine vibrations (such as normal train passage), the strategy may simply be continuous monitoring or data archiving.

[0072] Based on the real-time attributes of the vibration source (such as vibration intensity trends, whether it is continuously increasing, and whether it involves multi-area linkage) and the system's operating status (such as available manpower, equipment resources, and response window), the preset strategies are dynamically adjusted and optimized to generate a final execution strategy package, including the warning method, response level, responsible party, handling time limit, and expected goals. The final generated response strategy is output in the form of structured instructions, including fields such as strategy ID, applicable area, risk level, vibration source category, recommended action, execution priority, and expected effect evaluation, for automatic system execution or for manual decision-making reference by management personnel.

[0073] For vibration sources identified as high-risk areas or requiring emergency response, the module will trigger a real-time early warning mechanism and send immediate warning signals to subway operation and management personnel through multiple channels (such as system pop-ups, SMS, email, mobile app push notifications, and console alarm lights) to ensure that key personnel are informed of the risk information immediately. Simultaneously, the system will automatically generate a detailed vibration data analysis report, including: Basic information about the vibration source (time, location, duration, vibration type, etc.); Results of vibration source identification and classification; Multidimensional characteristic parameters of vibration signals (time domain, frequency domain, time-frequency characteristics, etc.); Analysis of the affected area and risk propagation; Current risk level and recommended response strategies for the region; Comparison of similar historical cases (if any); Data visualization charts (such as vibration waveforms, spectrum diagrams, risk heat maps, etc.); This report serves as decision support material, helping managers quickly understand the nature of an event, assess its scope of impact, and develop scientific response plans, significantly improving the professionalism and efficiency of emergency response.

[0074] To avoid excessive consumption of system energy and computing resources due to frequent warnings and high-intensity response operations, the strategy generation and warning module introduces an intelligent resource scheduling and power management mechanism to ensure the sustainability of the warning and response process and the stability of the system.

[0075] Real-time monitoring of key indicators such as current platform computing resource utilization, communication bandwidth, power supply status, and sensor workload assesses the overall system's redundancy (i.e., unused available resource capacity). Based on the current redundancy and the amount of upcoming early warning tasks (such as the number of early warnings to be sent), the estimated power consumption required for each round of early warning and response is calculated and compared with a preset energy capacity threshold. If the current task load is within acceptable limits, all strategies are executed in priority order; if it exceeds the limit, resource pruning and task scheduling optimization logic is initiated, prioritizing the response needs of high-risk areas and deferring or downgrading low-priority tasks.

[0076] When the system detects that overall energy resources (such as electricity, computing resources, network bandwidth, etc.) are becoming strained or have reached a critical threshold, the module will automatically trigger a response pause mechanism, suspending the execution of current non-critical tasks, recording the status and intermediate data of completed warning tasks, and generating a task pause log. Once energy supply returns to normal, the system will automatically resume and continue executing the remaining warning and response processes based on the log, ensuring the continuity, stability, and efficiency of overall operations.

[0077] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0078] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform, characterized in that: It includes a data acquisition and processing module, a vibration analysis and sorting module, and a strategy generation and early warning module; Data acquisition and processing module: Constructs a comprehensive environmental perception model in the subway environment, integrates data from multiple sensors, and combines external environmental factors to comprehensively monitor the subway's operating status, monitors the subway's vibration signals in real time, and triggers a self-check and analysis mechanism when the preset vibration warning value is reached; Vibration analysis and ranking module: Utilizes multi-source data fusion algorithms to analyze the collected vibration signals, identify and classify the characteristics of vibration sources, automatically prioritize vibration sources, and select areas with potential threats for secondary analysis and real-time monitoring; Strategy generation and early warning module: Based on the vibration source identification and analysis results, it automatically predicts the risk level of each area of ​​the subway, formulates response strategies based on priority and risk level, automatically adjusts the power allocation in the early warning and response process according to the current platform redundancy capacity, and if energy shortage is detected, it pauses the response process, records the currently completed early warning tasks, and continues the remaining early warning work after recovery.

2. The multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 1, characterized in that: The vibration analysis and sorting module receives pre-processed vibration signals and related environmental data from the data acquisition and processing module; Similar features extracted from different sensors are aligned and stitched together to form a comprehensive feature matrix, which contains complementary information from multiple sensors. Using the fused feature vector as input, the type and cause of vibration sources are identified; The extracted vibration feature vectors are input into a pre-trained classification model. The classification model outputs the probability distribution of the vibration source belonging to a certain category through multi-layer feature mapping and nonlinear transformation. The category with the highest probability is selected as the preliminary identification result of the vibration source.

3. The multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 2, characterized in that: The vibration analysis and sorting module analyzes the attribute parameters of the vibration source. After completing the identification and classification of the vibration source, it automatically sorts and prioritizes all detected vibration sources in the current monitoring area based on risk assessment indicators and priority rules. A vibration source risk scoring model is constructed for ranking. A comprehensive risk score is calculated for each vibration source, and they are ranked in descending order according to the comprehensive risk score. The higher the score, the greater the potential threat that the vibration source poses to the safety of subway operation, and the higher its priority.

4. The multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 3, characterized in that: The vibration source risk scoring model includes vibration intensity, whether the vibration frequency is within the equipment's sensitive frequency band, the spatial location of the vibration source, the persistence and development trend of the vibration source, the type of vibration source, and the environmental background.

5. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 4, characterized in that: The vibration analysis and sorting module calculates the time-domain statistical characteristics of the vibration signal, including peak amplitude, root mean square, peak-to-peak value, vibration duration, signal rise time, and signal zero-crossing rate.

6. The multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 5, characterized in that: The strategy generation and early warning module determines the area directly affected by the vibration source and the associated area that produces secondary effects based on the spatial location of the vibration source, combined with the subway line topology, equipment layout diagram and spatial index model. Based on a historical vibration propagation case library, the intensity, frequency characteristics, and spatial and structural propagation characteristics of the current vibration source are assessed, and its potential impact on the surrounding area is predicted. Based on vibration propagation analysis, assess the vibration intensity and frequency composition of the area to determine whether it is close to the sensitive range of the structure or the operating frequency of the equipment in the area. After clarifying the risk level of each region, based on the priority ranking of vibration sources and the regional risk level, a strategy matching and generation logic is used to formulate differentiated and executable response strategies for each risk region or vibration source.

7. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 6, characterized in that: The strategy generation and early warning module assesses the intensity, frequency characteristics, and spatial and structural propagation characteristics of the current vibration source, and predicts its potential impact on the surrounding area, including the following steps: Construct a regional risk index model and calculate multiple risk factors; Based on multiple risk factors, a comprehensive risk index is calculated for each subway area, and the area is classified into the corresponding risk level category according to the preset risk classification standards.

8. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 7, characterized in that: The strategy generation and early warning module formulates differentiated and executable response strategies for each risk area or vibration source, including the following steps: For high-risk areas, strategies include immediately initiating on-site inspections, suspending operation of relevant lines, notifying maintenance teams to be on standby, and switching to backup equipment. For medium-risk areas, strategies include increasing monitoring frequency, recording vibration evolution trends, and sending warning notices to construction units; For low-risk or routine vibrations, the strategy is continuous monitoring or data archiving.

9. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 8, characterized in that: The vibration analysis and sorting module identifies and classifies the characteristics of vibration sources. Through multi-dimensional data from fiber optic sensing and accelerometers, it analyzes the frequency, waveform, and intensity of vibrations to determine whether the vibration is caused by equipment failure, external factors, or human factors.

10. A multi-scenario distributed fiber optic acoustic sensing vibration data fusion early warning platform according to claim 9, characterized in that: The data acquisition and processing module integrates data from multiple sensors, including fiber optic acoustic sensors, accelerometers, and temperature and humidity sensors. A network of fiber optic acoustic sensors distributed in subway tunnels, along the track, in station structures, and in equipment areas captures sound wave / vibration signals propagating along the optical fiber in real time. Accelerometers are used to acquire local high-frequency vibration responses, while temperature and humidity sensors are used to simultaneously collect ambient temperature and relative humidity data.

Images