GIL pipe gallery expansion joint online monitoring system and method based on sensor

Abstract

The invention discloses a sensor-based GIL pipe gallery expansion joint on-line monitoring system and method, and relates to the technical field of power equipment monitoring, and the method comprises the steps: collecting key state information and environmental parameters of an expansion joint through the deployment of an expansion joint sensor and an environmental sensor, and carrying out the preprocessing; performing deep fusion on expansion joint sensor data and environment sensor data, constructing a comprehensive state evaluation and prediction model, and analyzing the influence of different factors including environment, mechanical faults and material aging on the expansion joint state; correcting state parameters for environmental factors; for mechanical faults, fault types, positions and severity are identified, and corresponding early warning is triggered; aiming at material aging, monitoring and predicting performance degradation and residual life in real time, and formulating a maintenance strategy; the model performance is evaluated and updated regularly, an early warning threshold value is adjusted dynamically, and self-adaptive early warning is achieved; a data integration and visualization platform is developed, and visual display and interactive analysis of monitoring data are achieved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of power equipment monitoring, and particularly relates to a GIL pipe gallery expansion joint online monitoring system and method based on sensors. BACKGROUND

[0002] The GIL (gas insulated metal enclosed transmission line) is a kind of high-voltage power transmission technology, and the expansion joint in the GIL pipe gallery is a key component for compensating the length change caused by thermal expansion and cold contraction to ensure safe and stable operation of the system. However, the expansion joint will be affected by various factors during long-term operation, leading to performance degradation or even failure, which threatens the normal operation of the GIL pipe gallery.

[0003] At present, the monitoring method for the expansion joint of the GIL pipe gallery has defects. The traditional method mainly focuses on a single parameter such as mechanical stress or displacement of the expansion joint, lacks comprehensive monitoring of environmental factors and material state, and is difficult to accurately evaluate the comprehensive state of the expansion joint. The state of the expansion joint is affected by various factors, and the existing method lacks deep fusion of multi-source heterogeneous data, cannot effectively mine the correlation between data, and limits the accuracy of state evaluation. For state changes caused by different factors, the existing method lacks targeted analysis means and processing strategies, and is difficult to effectively identify the fault type and position, and cannot predict the material aging and remaining service life. The existing method mainly relies on experience to set the warning threshold, lacks a self-adaptive adjustment mechanism, is difficult to adapt to the dynamic changes of the running state of the expansion joint, and is prone to false alarms or missed alarms. SUMMARY

[0004] The purpose of the present application is to provide a GIL pipe gallery expansion joint online monitoring system and method based on sensors to solve the problems in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions: In a first aspect, the present application provides a GIL pipe gallery expansion joint online monitoring method based on sensors, comprising: Deploying expansion joint sensors and environmental sensors to collect key state information of the expansion joint and surrounding environmental parameters affecting the state of the expansion joint, including stress, displacement, temperature, vibration, external temperature, humidity and air pressure, and performing preliminary processing on the collected data, including filtering and denoising; Deeply fusing the expansion joint sensor data and the environmental sensor data to construct a data-driven comprehensive state evaluation and prediction model; based on the model, analyzing the influence of different factors on the state of the expansion joint, which is divided into environmental factors, mechanical fault factors and material aging factors, and using corresponding analysis methods and processing strategies for different types of influencing factors; For state changes caused by environmental factors, the influence of environmental factors on the state parameters of the expansion joint is analyzed, and the state parameters of the expansion joint are corrected; for abnormal states caused by mechanical failures, vibration signal analysis and stress distribution analysis methods are used to identify the type and location of mechanical failures, evaluate the severity of the failure, and trigger appropriate warning information according to the severity of the failure; for performance degradation caused by material aging, real-time monitoring of the state is performed using material performance parameters and stress history data, and the performance degradation and remaining useful life of the expansion joint caused by material aging are predicted, and a maintenance strategy is developed based on the prediction results. Periodically evaluate model performance, update and optimize model parameters based on new operation monitoring data; dynamically adjust the warning threshold to achieve adaptive warning. Develop a GIL pipe gallery expansion joint data integration and visualization platform to integrate multi-source monitoring data and display the expansion joint state in a graphical interface to achieve visual display and interactive analysis of monitoring data.

[0006] In combination with the first aspect, in a first implementation manner of the first aspect of the present application, the expansion joint sensor and the environment sensor are deployed to collect key state information of the expansion joint and surrounding environmental parameters affecting the state of the expansion joint, including stress, displacement, temperature, vibration, external temperature, humidity and air pressure, and the collected data is preliminarily processed, including filtering and denoising, including: The expansion joint sensor and the environment sensor are deployed, the expansion joint sensor includes stress sensors, displacement sensors, temperature sensors and vibration sensors, the environment sensor includes external temperature sensors, humidity sensors and air pressure sensors, and stress, displacement, temperature, vibration, external temperature, humidity and air pressure data are collected through the sensors; the collected data is preprocessed, and filtering technology is used to remove noise and drift in the data, and a wavelet denoising method is used to select appropriate wavelet basis functions and decomposition levels to remove noise and interference in the data.

[0007] In combination with the first aspect, in a second implementation manner of the first aspect of the present application, the expansion joint sensor data and the environment sensor data are deeply fused to construct a comprehensive state evaluation and prediction model based on data driving, including: The preprocessed expansion joint sensor data and environment sensor data are time-synchronized and spatially aligned to construct a unified space-time data matrix; Correlation analysis algorithm is used to extract key fusion features reflecting state changes of the expansion joint, and a multi-source fusion feature set containing expansion joint parameters and environmental parameters is constructed to describe the comprehensive operating state; based on the fusion features, a long short-term memory network model capable of evaluating the comprehensive state and life prediction is trained to realize health scoring and fault type determination of the current state of the expansion joint, and to predict future state change trends and degradation of key performance indicators.

[0008] With reference to the first aspect, in a third implementation form of the first aspect of the present disclosure, the analysis of the influence of different factors on the state of the telescopic joint based on the model is divided into environmental factors, mechanical failure factors and material aging factors, and corresponding analysis methods and processing strategies are adopted for different types of influencing factors, including: The state parameters and environmental variables collected during the operation of the telescopic joint are classified and labeled to construct a data label system corresponding to the three main influencing factors; a model-based feature selection algorithm is used to identify the dominant variables of different types of factors affecting the state change of the telescopic joint, and the influence source type and corresponding key feature vector are divided; The current telescopic joint state anomaly is classified and judged to identify whether it belongs to changes caused by environmental fluctuations, mechanical structural abnormalities or material performance degradation, and according to the identified type of influencing factor, the corresponding processing strategy is selected for targeted analysis and processing.

[0009] With reference to the first aspect, in a fourth implementation form of the first aspect of the present disclosure, for state changes caused by environmental factors, the influence of environmental factors on the state parameters of the telescopic joint is analyzed, and the state parameters of the telescopic joint are corrected, including: According to the comprehensive state evaluation and prediction model, the influence of environmental factors on the stress, displacement, temperature and vibration state parameters of the telescopic joint is analyzed, the environmental features related to the state parameters of the telescopic joint are selected, and the selected features are used to train a random forest model. The features are combined and analyzed by multiple decision trees, and each tree uses bootstrap sampling method during training and selects the optimal feature during node splitting; the model performance is evaluated by cross-validation, and the number and depth of trees are adjusted to optimize the model; A quantitative relationship model between environmental factors and state parameters of the telescopic joint is established, and the trained random forest model is used to calculate the influence of environmental factors on the state parameters of the telescopic joint, and the formula is: ; Where ΔS i is the influence of environmental factors on the i-th state parameter of the telescopic joint, f is the random forest model, and E is the environmental factor vector; The state parameters of the telescopic joint are dynamically corrected, and the formula is: ; Where S i ' is the corrected i-th state parameter value of the telescopic joint, and S i is the original value of the i-th state parameter measured by the sensor in real time; Based on historical data statistics, the normal range threshold of each state parameter is set, and the corrected state parameters of the telescopic joint are monitored and evaluated in real time. When S i ' is within the normal range, it is evaluated as normal state, and when Si When the abnormal state is determined, the corresponding high-frequency vibration signal and stress data are extracted for analysis.

[0010] In combination with the first aspect, in a fifth implementation manner of the first aspect of the present application, for the abnormal state caused by mechanical failure, a vibration signal analysis and stress distribution analysis method are used to identify the mechanical failure type and failure location, evaluate the failure severity, and trigger corresponding early warning information according to the failure severity, including: When the abnormal state of the telescopic joint is determined, the corresponding high-frequency vibration signal and stress data are extracted for analysis. The vibration signal is processed through time domain analysis, frequency domain analysis and time-frequency domain analysis; by analyzing the amplitude, frequency component, phase and impact characteristics of the vibration signal, the vibration signal is compared with the vibration characteristic library of the looseness, wear, misalignment and crack failure modes to identify the possible mechanical failure type and approximate location; in combination with the spatial distribution and real-time readings of the stress sensor, a stress distribution curve of the key parts of the telescopic joint is drawn to analyze the stress concentration area, the size and change trend of the stress value, and find out the high stress concentration area to further determine the failure location. In combination with the results of the vibration signal analysis and stress distribution analysis, a comprehensive judgment is made, based on the identified failure type, location and related vibration and stress parameter values, a failure severity level standard is divided, including slight failure, medium failure and serious failure, and the severity of the current failure is evaluated; the ranges of slight failure, medium failure and serious failure are set, when in the range of slight failure, it is slight failure; when in the range of medium failure, it is medium failure; when in the range of serious failure, it is serious failure. According to the evaluated failure severity level, different levels and forms of early warning information are automatically triggered; when the level is slight failure, a notification is sent to the maintenance personnel to prompt attention; when the level is medium failure, a medium failure alarm is triggered to prompt the maintenance personnel to arrange a near-term inspection; when the level is serious failure, a serious failure alarm is triggered to require immediate shutdown inspection and take emergency measures.

[0011] In combination with the first aspect, in a sixth implementation manner of the first aspect of the present application, for the performance degradation caused by material aging, material performance parameters and stress history data are used to monitor the state in real time, predict the performance degradation and remaining service life of the telescopic joint caused by material aging, and develop a maintenance strategy according to the prediction results, including: According to the comprehensive state evaluation and prediction model, a relationship model between the material performance parameters and the stress history data is established. The model uses a long short-term memory network model based on deep learning to reflect the aging law of the material in the long-term operation process. The key state parameters of the expansion joint, including stress, strain and temperature, are collected by the sensor, and the current state of the expansion joint is evaluated by a state evaluation method based on a physical information neural network, combined with the material performance parameters and the stress history data. Based on the evaluation result, the performance degradation and the remaining service life of the expansion joint in the future period of time are predicted by using the relationship model, and according to the prediction result, the corresponding maintenance strategy is formulated, including planned replacement, repair reinforcement and intensified monitoring.

[0012] In combination with the first aspect, in a seventh implementation manner of the first aspect of the present application, the model performance is periodically evaluated, and the model parameters are updated and optimized according to new operation monitoring data; the early warning threshold is dynamically adjusted to realize adaptive early warning, including: The model prediction result is compared with the actual monitoring data to calculate the accuracy and error of the model prediction, and the accuracy of the model prediction on the state of the expansion joint is evaluated; the new operation monitoring data is used to update and optimize the model parameters by using an incremental gradient descent algorithm; The early warning threshold is dynamically adjusted according to the model performance evaluation result, the actual operation state of the expansion joint and the historical early warning information; when the model prediction accuracy decreases or the operation state of the expansion joint deteriorates, the early warning threshold is lowered; when the model prediction accuracy increases or the operation state of the expansion joint is good, the early warning threshold is raised; By periodically evaluating the model performance, updating the model parameters and dynamically adjusting the early warning threshold, the early warning strategy can be adaptively adjusted according to the actual operation state of the expansion joint, the change of the model performance and the historical early warning effect.

[0013] In combination with the first aspect, in an eighth implementation manner of the first aspect of the present application, a GIL pipe gallery expansion joint data integration and visualization platform is developed to integrate multi-source monitoring data to display the state of the expansion joint in a graphical interface, realize the visual display and interactive analysis of the monitoring data, including: The expansion joint state parameter data is integrated to be displayed in a visual form of charts, curves and maps, including real-time data, historical data and trend analysis; an interactive analysis function is provided to deeply analyze and explore the data, including viewing the data of a specific time period, comparing the data of different sensors and analyzing the change trend of the data; the early warning information is displayed on the graphical interface to provide corresponding operation prompts, including confirming the early warning, viewing detailed information and recording the processing process.

[0014] The second aspect of the present application provides a sensor-based GIL pipe gallery expansion joint online monitoring system, including: The data acquisition module includes a telescopic joint sensor unit, an environment sensor unit, and a data preprocessing unit; the telescopic joint sensor unit collects key state information parameter data of the telescopic joint, including stress, displacement, temperature, and vibration; the environment sensor unit collects surrounding environmental parameter data that affect the state of the telescopic joint, including external temperature, humidity, and air pressure; the data preprocessing unit performs preliminary processing on the collected data, including filtering and denoising; The data fusion and modeling module includes a data synchronization and alignment unit, a feature extraction unit, and a state evaluation and prediction unit; the data synchronization and alignment unit synchronizes the telescopic joint sensor data and the environment sensor data in time and aligns them in space position, and constructs a unified space-time data matrix; The feature extraction unit extracts key fusion features reflecting the state change of the telescopic joint using a correlation analysis algorithm, and constructs a multi-source fusion feature set; the state evaluation and prediction unit trains a long short-term memory network model based on the fusion features, which can evaluate the comprehensive state and life prediction, and predicts the future state change trend and key performance indicators of the telescopic joint; The state analysis and processing module includes an environmental factor analysis unit, a mechanical fault analysis unit, and a material aging analysis unit; the environmental factor analysis unit analyzes the influence of environmental factors on the state parameters of the telescopic joint and makes corrections; the mechanical fault analysis unit uses vibration signal analysis and stress distribution analysis methods to identify the type and location of mechanical faults, evaluate the severity of the fault, and trigger warning information; the material aging analysis unit uses material performance parameters and stress historical data to monitor the state in real time, predict the performance degradation and remaining service life of the telescopic joint due to material aging, and develop maintenance strategies; The model evaluation and optimization module includes a model performance evaluation unit, a model parameter updating unit, and a warning threshold adjustment unit; the model performance evaluation unit compares the model prediction results with the actual monitoring data, calculates the accuracy and error, and evaluates the accuracy of the model in predicting the state of the telescopic joint; the model parameter updating unit uses incremental gradient descent algorithm to update and optimize the model parameters based on new operation monitoring data; the warning threshold adjustment unit dynamically adjusts the warning threshold based on the model performance evaluation results, the actual operation state of the telescopic joint, and historical warning information; The data visualization and interactive platform module includes a data integration unit, a visualization display unit, an interactive analysis unit, and a warning information display unit; the data integration unit integrates the state parameter data of the telescopic joint; the visualization display unit displays the data in the form of charts, curves, and maps; the interactive analysis unit provides interactive analysis functions for in-depth analysis and exploration of the data; the warning information display unit displays the warning information on the graphical interface and provides corresponding operation prompts.

[0015] Compared with the prior art, the application has the beneficial effects that: 1、 The application deeply fuses the telescopic joint sensor data and the environmental sensor data, constructs a comprehensive state evaluation and prediction model based on data driving, and overcomes the problem of insufficient data fusion in the prior art.

[0016] 2、 The application adopts corresponding analysis methods and processing strategies for state changes caused by different factors, and overcomes the problem of limited analysis means in the prior art.

[0017] 3、 The application predicts the performance degradation and residual service life of the telescopic joint caused by material aging based on material performance parameters and stress history data, formulates a maintenance strategy according to the prediction result, realizes the transformation from passive repair to active maintenance, and overcomes the problem of insufficient early warning capability in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0018] Fig. 1 A step schematic diagram of the online monitoring method of the telescopic joint of the GIL pipe gallery based on sensors according to the application; Fig. 2 A system structure diagram of the online monitoring system of the telescopic joint of the GIL pipe gallery based on sensors according to the application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0020] Embodiment: As shown in the drawings, Figs. 1-2 the application provides a technical solution, as shown in the drawings, Fig. 1 A step schematic diagram of the online monitoring method of the telescopic joint of the GIL pipe gallery based on sensors according to the application, the application provides a step schematic diagram of the online monitoring method of the telescopic joint of the GIL pipe gallery based on sensors, which comprises: Step S100: Deploy the telescopic joint sensor and the environmental sensor, collect the key state information of the telescopic joint and the surrounding environmental parameters affecting the state of the telescopic joint, including stress, displacement, temperature, vibration, external temperature, humidity and air pressure, and preliminarily process the collected data, including filtering and denoising; Specifically, a telescopic joint sensor and an environment sensor are deployed, the telescopic joint sensor includes a stress sensor, a displacement sensor, a temperature sensor and a vibration sensor, the environment sensor includes an external temperature sensor, a humidity sensor and a barometric pressure sensor, stress, displacement, temperature, vibration, external temperature, humidity and barometric pressure data are collected through the sensors; the collected data is preprocessed, filtering technology is used to remove noise and drift in the data, a wavelet denoising method is used, a suitable wavelet basis function and a decomposition layer number are selected to remove noise and interference in the data.

[0021] In a specific embodiment, the stress sensor selects a strain sensor with a range of 0-50 MPa, the sampling frequency is set to 100 Hz; the displacement sensor selects an LVDT sensor with a measurement range of 0-100 mm, the sampling frequency is set to 50 Hz; the temperature sensor selects a thermocouple sensor with a measurement range of -40℃ to 125℃, the sampling frequency is set to 1 Hz; the vibration sensor selects a piezoelectric acceleration sensor with a frequency range of 0.5 Hz to 10 kHz, the sampling frequency is set to 500 Hz; the external temperature sensor, the humidity sensor and the barometric pressure sensor respectively select commercially available SHT31 and BME280 sensors, the sampling frequency is set to 1 Hz. After data collection, first, filtering processing is performed, for stress, displacement and vibration data, a low-pass Butterworth filter with a cutoff frequency of 10 Hz is used to remove high-frequency noise; then wavelet denoising processing is performed, db4 wavelet basis function is selected, decomposition layer number is 3, temperature, humidity and barometric pressure data are denoised, random fluctuations and irrelevant interference in the data are effectively removed.

[0022] Step S200: Deeply fuse the telescopic joint sensor data and the environment sensor data to construct a data-driven comprehensive state evaluation and prediction model; analyze the influence of different factors on the telescopic joint state based on the model, divide into environmental factors, mechanical failure factors and material aging factors, for different types of influence factors, use corresponding analysis methods and processing strategies; Specifically, the preprocessed telescopic joint sensor data and environment sensor data are time-synchronized and spatially aligned, and a unified space-time data matrix is constructed; Correlation analysis algorithm is used to extract key fusion features reflecting the state change of the telescopic joint, a multi-source fusion feature set containing telescopic joint parameters and environmental parameters is constructed to describe the comprehensive operating state; based on the fusion features, a long short-term memory network model capable of evaluating the comprehensive state and life prediction is trained to realize health scoring and fault type judgment of the current state of the telescopic joint, and to predict future state change trend and degradation of key performance indicators.

[0023] Further, the state parameters and environmental variables collected during the operation of the telescopic joint are classified and labeled to construct a data label system corresponding to the three main influencing factors; a model-based feature selection algorithm is used to identify the dominant variables of different types of factors affecting the state change of the telescopic joint, and the influence source type and corresponding key feature vector are divided; The current telescopic joint state anomaly is classified and judged to identify whether it belongs to environmental fluctuation, mechanical structural abnormality or material performance degradation, and according to the identified influence factor type, the corresponding processing strategy is selected for targeted analysis and processing.

[0024] In a specific embodiment, the pre-processed telescopic joint sensor data and environmental sensor data are time-synchronized to ensure that the timestamps of all data are accurate to the millisecond level, and the spatial positions are aligned to ensure that the data comes from the same telescopic joint position. The constructed spatio-temporal data matrix contains 1000 time points, each time point contains 7 feature dimensions, namely stress, displacement, temperature, vibration, external temperature, humidity and air pressure.

[0025] Pearson correlation analysis algorithm is used to calculate the correlation coefficient between each feature, and the features highly related to the state change of the telescopic joint are extracted to construct a multi-source fusion feature set containing these key fusion features.

[0026] Based on the fusion feature set, a long short-term memory network LSTM model is trained, the model structure is 3-layer LSTM layer, each layer contains 50 neurons, Adam optimization algorithm is used, learning rate is set to 0.001, batch size is 32, and training period is 100 rounds. Through the model output, the health score of the current state of the telescopic joint is 80 points, indicating that the state is good, and the fault type is determined as mechanical structural abnormality.

[0027] Further, 10000 state parameter and environmental variable data collected during the operation of the telescopic joint are classified and labeled to construct a data label system, including environmental factor label, mechanical failure factor label and material aging factor label. Using a model-based feature selection algorithm, through recursive feature elimination combined with a support vector machine algorithm, external temperature and humidity in the environmental factor, vibration and displacement in the mechanical failure factor, and stress and temperature in the material aging factor are identified as the dominant variables.

[0028] The current telescopic joint state anomaly is classified and judged, and it is identified that the state anomaly of a certain time is mainly caused by the sudden increase of external temperature by 5°C and the decrease of humidity by 10%, and the environmental control parameters are adjusted; when it is identified that the vibration amplitude exceeds the threshold value by 20%, the displacement exceeds the threshold value by 15%, and the stress exceeds the threshold value by 10%, structural inspection and maintenance are carried out; when it is identified that the stress relaxation rate exceeds 5%, the performance degradation index exceeds the threshold value by 8%, and the fatigue damage index exceeds the threshold value by 6%, the remaining service life is predicted and the maintenance strategy is formulated.

[0029] Step S300: For the state change caused by environmental factors, the influence of environmental factors on the telescopic joint state parameters is analyzed, and the telescopic joint state parameters are corrected; for the abnormal state caused by mechanical failure, the vibration signal analysis and stress distribution analysis method are used to identify the mechanical failure type and failure position, evaluate the failure severity, and trigger the corresponding warning information according to the severity of the failure; for the performance degradation caused by material aging, the material performance parameters and stress history data are used to monitor the state in real time, predict the performance degradation and remaining service life of the telescopic joint caused by material aging, and formulate a maintenance strategy according to the prediction results; Specifically, according to the comprehensive state evaluation and prediction model, the influence of environmental factors on the stress, displacement, temperature and vibration state parameters of the telescopic joint is analyzed, the environmental characteristics related to the telescopic joint state parameters are selected, the selected characteristics are used to train the random forest model, and the characteristics are combined and analyzed by multiple decision trees, and each tree uses the bootstrap sampling method during training and selects the optimal feature during node splitting; the model performance is evaluated by cross-validation, and the number and depth of trees are adjusted to optimize the model; A quantitative relationship model between environmental factors and telescopic joint state parameters is established, and the trained random forest model is used to calculate the influence of environmental factors on the state parameters of the telescopic joint, and the formula is: ; Where ΔS i is the influence of environmental factors on the i-th telescopic joint state parameter, f is the random forest model, and E is the environmental factor vector; The telescopic joint state parameters are dynamically corrected, and the formula is: ; Where S i ' is the corrected i-th telescopic joint state parameter value, and S i is the original value of the i-th state parameter measured by the sensor in real time; Based on historical data statistics, the normal range threshold of each state parameter is set, the corrected telescopic joint state parameters are monitored and evaluated in real time, and when S i ' is within the normal range, it is evaluated as normal state, and when S i ' is outside the normal range, it is evaluated as abnormal state.

[0030] When it is determined that the expansion joint enters an abnormal state, the corresponding high-frequency vibration signal and stress data are extracted for analysis; The vibration signal is processed by time domain analysis, frequency domain analysis and time-frequency domain analysis; by analyzing the amplitude, frequency component, phase and impact characteristics of the vibration signal, the vibration characteristic library of the looseness, wear, misalignment and crack fault modes is compared to identify the possible mechanical fault type and approximate location; combined with the spatial distribution and real-time readings of the stress sensor, the stress distribution curve of the key parts of the expansion joint is drawn, the stress concentration area, the size and change trend of the stress value are analyzed, and the high stress concentration area is found out to further determine the fault location; Combined with the results of vibration signal analysis and stress distribution analysis, comprehensive judgment is carried out, based on the identified fault type, location and related vibration, stress parameter value, the fault severity level standard is divided, including slight fault, medium fault and serious fault, the severity of the current fault is evaluated; set the range of slight fault, medium fault and serious fault, when in the range of slight fault, it is slight fault; when in the range of medium fault, it is medium fault; when in the range of serious fault, it is serious fault; According to the evaluated fault severity level, different levels and forms of early warning information are automatically triggered; when the level is slight fault, send a notification to the maintenance personnel and prompt attention; when the level is medium fault, trigger a medium fault alarm and prompt the maintenance personnel to arrange a near-term inspection; when the level is serious fault, trigger a serious fault alarm and require immediate shutdown inspection and take emergency measures.

[0031] According to the comprehensive state evaluation and prediction model, a relationship model between material performance parameters and stress historical data is established, which uses a long short-term memory network model based on deep learning to reflect the aging law of the material in the long-term operation process; the key state parameters of the expansion joint collected by the sensor, including stress, strain and temperature, are combined with the material performance parameters and stress historical data to evaluate the current state of the expansion joint by a state evaluation method based on physical information neural network; based on the evaluation result, the relationship model is used to predict the performance degradation and remaining useful life of the expansion joint in a future period of time, according to the prediction result, the corresponding maintenance strategy is formulated, including planned replacement, repair reinforcement and enhanced monitoring.

[0032] In a specific embodiment, the influence of environmental factors on the state parameters of the expansion joint of an underground GIL pipe gallery in a certain city is analyzed. Through a random forest model, temperature, humidity, and pressure are selected as key environmental characteristics, and the model is trained using these characteristics. After cross-validation and parameter optimization, a random forest model containing 100 decision trees with a depth of 10 is determined. Using this model, the influence of environmental factors on the stress of the expansion joint is calculated as ΔS1=0.5 MPa, the influence on the displacement is ΔS2=0.2 mm, the influence on the temperature is ΔS3=1.5 ℃, and the influence on the vibration is ΔS4=0.1 g. The state parameters of the expansion joint are dynamically corrected, and the corrected stress value is S1'=15.5 MPa, the displacement value is S2'=0.8 mm, the temperature value is S3'=35.5 ℃, and the vibration value is S4'=0.3 g. Based on historical data statistics, the normal range of stress is set to 10-20 MPa, the normal range of displacement is set to 0-1 mm, the normal range of temperature is set to 30-40 ℃, and the normal range of vibration is set to 0-0.5 g. Real-time monitoring shows that the corrected state parameters are all within the normal range, and the evaluation is normal. At a certain time, the stress value is monitored to suddenly increase to S1'=22 MPa, which exceeds the normal range, and is determined to be an abnormal state. Further extraction of high-frequency vibration signals and stress data for analysis finds that there is a clear peak in the vibration signal with a frequency of 50 Hz, which matches the vibration characteristic library of the loose fault mode, and is initially identified as a loose fault. Combined with the stress distribution curve, the stress concentration area is located in the middle of the expansion joint, further determining the fault location. After comprehensive judgment, the current fault is evaluated as a medium fault, triggering a medium fault alarm and prompting maintenance personnel to arrange a near-term inspection. A long short-term memory network model is used to establish a relationship model between material performance parameters and stress historical data, and it is predicted that the performance degradation rate of the expansion joint in the next year is 5%, and the remaining service life is 8 years. According to the prediction results, a maintenance strategy of strengthening monitoring and planned replacement is developed.

[0033] Step S400: periodically evaluate the model performance, update and optimize the model parameters according to the new operation monitoring data; dynamically adjust the warning threshold to realize adaptive warning; Specifically, the model prediction results are compared with the actual monitoring data to calculate the accuracy and error of the model prediction, and the accuracy of the model prediction of the expansion joint state is evaluated; new operation monitoring data is used to update and optimize the model parameters using an incremental gradient descent algorithm; According to the model performance evaluation results, the actual operation state of the expansion joint, and the historical warning information, the warning threshold is dynamically adjusted; when the model prediction accuracy decreases or the operation state of the expansion joint deteriorates, the warning threshold is lowered; when the model prediction accuracy improves or the operation state of the expansion joint is good, the warning threshold is raised; By regularly evaluating model performance, updating model parameters, and dynamically adjusting early warning thresholds, the early warning strategy can adaptively adjust according to the actual operating state of the expansion joint, changes in model performance, and historical early warning effects.

[0034] In a specific embodiment, the model performance is evaluated regularly, with a comprehensive evaluation conducted once every quarter. The model prediction results are compared with the actual monitoring data, and the accuracy of the model prediction is calculated to be 92%, with an error of 0.8. To improve the accuracy of the model, new operating monitoring data is used, and the incremental gradient descent algorithm is used to update and optimize the model parameters. After 10 iterations, the model prediction accuracy is improved to 95%, and the error is reduced to 0.5. According to the model performance evaluation results, the actual operating state of the expansion joint, and the historical early warning information, the early warning threshold is dynamically adjusted. When the model prediction accuracy decreases to 90% or the expansion joint operating state deteriorates, the early warning threshold is reduced by 10%; when the model prediction accuracy increases to 96% or the expansion joint operating state is good, the early warning threshold is increased by 5%. By regularly evaluating model performance, updating model parameters, and dynamically adjusting early warning thresholds, the early warning strategy can adaptively adjust according to the actual operating state of the expansion joint, changes in model performance, and historical early warning effects, effectively improving the accuracy and timeliness of early warning.

[0035] Step S500: Develop a GIL pipe gallery expansion joint data integration and visualization platform, integrate multi-source monitoring data, and display the expansion joint state in a graphical interface to realize visual display and interactive analysis of monitoring data.

[0036] Specifically, the expansion joint state parameter data is integrated and displayed in the form of charts, curves, and maps, including real-time data, historical data, and trend analysis; interactive analysis functions are provided to conduct in-depth analysis and exploration of data, including viewing data for a specific time period, comparing data from different sensors, and analyzing data trends; early warning information is displayed on the graphical interface, and corresponding operation prompts are provided, including confirming the early warning, viewing detailed information, and recording the processing process.

[0037] In a specific embodiment, a GIL pipe gallery expansion joint data integration and visualization platform is developed, which integrates multi-source monitoring data and displays the expansion joint state in a graphical interface. The platform integrates expansion joint state parameter data, including stress, displacement, temperature, vibration, and environmental parameters, and displays them in the form of charts, curves, and maps.

[0038] The real-time data is displayed at a frequency of one update per second, and the historical data can be traced back to the records of the past year. Trend analysis shows that the stress value of the expansion joint fluctuates between 30 MPa and 40 MPa in the past month, with an average of 35.6 MPa, a maximum of 38.9 MPa, and a minimum of 31.2 MPa. Displacement data shows that the maximum displacement of the expansion joint in the normal working range is 5 mm, the minimum displacement is 0.5 mm, and the average displacement is 2.8 mm. Temperature monitoring shows that the temperature of the expansion joint varies between 20℃ and 50℃, with an average temperature of 35℃. Vibration data analysis shows that the vibration frequency of the expansion joint is mainly concentrated in the range of 50Hz to 100Hz, with a maximum vibration acceleration of 0.5g. Environmental parameter monitoring shows that the humidity varies between 40% and 80%, with an average humidity of 60%, and the air pressure fluctuates between 1013hPa and 1020hPa, with an average air pressure of 1016hPa. The platform also provides interactive analysis functions, allowing users to view data in specific time periods through zoom-in, zoom-out, and translation operations, or to perform statistical analysis on data through filtering, sorting, and grouping operations.

[0039] The platform displays early warning information on the graphical interface and provides corresponding operation prompts according to the warning level. When a medium fault alarm is triggered, the platform will pop up a dialog box to prompt the maintenance personnel to confirm the warning and view detailed information, and record the processing process for subsequent analysis and improvement. Through this platform, maintenance personnel can intuitively understand the running state of the expansion joint and timely discover and handle potential problems.

[0040] As Fig. 2 A system structure diagram of a sensor-based GIL pipe gallery expansion joint online monitoring system is shown, the present application provides a sensor-based GIL pipe gallery expansion joint online monitoring system, comprising: The data acquisition module includes an expansion joint sensor unit, an environmental sensor unit, and a data preprocessing unit. The expansion joint sensor unit collects key state information parameter data of the expansion joint, including stress, displacement, temperature, and vibration. The environmental sensor unit collects surrounding environmental parameter data that affect the state of the expansion joint, including external temperature, humidity, and air pressure. The data preprocessing unit performs preliminary processing on the collected data, including filtering and denoising. The data fusion and modeling module includes a data synchronization and alignment unit, a feature extraction unit, and a state evaluation and prediction unit. The data synchronization and alignment unit synchronizes the expansion joint sensor data and environmental sensor data in time and aligns them in space position, constructing a unified spatio-temporal data matrix. The feature extraction unit adopts a correlation analysis algorithm to extract key fusion features reflecting the state change of the expansion joint, and constructs a multi-source fusion feature set; the state evaluation and prediction unit trains a long short-term memory network model capable of evaluating the comprehensive state and life prediction based on the fusion features, to score the state health of the expansion joint and determine the fault type, and to predict the future state change trend and degradation of key performance indicators; The state analysis and processing module includes an environmental factor analysis unit, a mechanical fault analysis unit, and a material aging analysis unit; the environmental factor analysis unit analyzes the influence of environmental factors on the state parameters of the expansion joint and performs correction; the mechanical fault analysis unit identifies the mechanical fault type and location, evaluates the fault severity, and triggers the warning information by using vibration signal analysis and stress distribution analysis methods; the material aging analysis unit uses material performance parameters and stress historical data to monitor the state in real time, predicts the performance degradation and remaining service life of the expansion joint caused by material aging, and formulates a maintenance strategy; The model evaluation and optimization module includes a model performance evaluation unit, a model parameter updating unit, and a warning threshold adjustment unit; the model performance evaluation unit compares the model prediction results with the actual monitoring data, calculates the accuracy and error, and evaluates the accuracy of the model in predicting the state of the expansion joint; the model parameter updating unit updates and optimizes the model parameters using incremental gradient descent algorithm based on new operation monitoring data; the warning threshold adjustment unit dynamically adjusts the warning threshold based on the model performance evaluation results, the actual operating state of the expansion joint, and historical warning information; The data visualization and interactive platform module includes a data integration unit, a visualization display unit, an interactive analysis unit, and a warning information display unit; the data integration unit integrates the state parameter data of the expansion joint; the visualization display unit displays the data in the form of charts, curves, and maps; the interactive analysis unit provides interactive analysis functions for in-depth analysis and exploration of the data; the warning information display unit displays the warning information on the graphical interface and provides corresponding operation prompts.

[0041] It is apparent for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be considered as limiting the claims involved.

Claims

1. A sensor-based online monitoring method for GIL tube gallery expansion joint, characterized in that, The method comprises the following steps: deploying telescopic joint sensors and environmental sensors to collect key state information of the telescopic joint and surrounding environmental parameters affecting the state of the telescopic joint, including stress, displacement, temperature, vibration, external temperature, humidity and air pressure, and performing preliminary processing on the collected data, including filtering and denoising; deeply fusing the telescopic joint sensor data and the environmental sensor data to construct a data-driven comprehensive state evaluation and prediction model; based on the model, the influence of different factors on the state of the telescopic joint is analyzed, which is divided into environmental factors, mechanical failure factors and material aging factors, and corresponding analysis methods and processing strategies are adopted for different types of influencing factors; for the state change caused by environmental factors, the influence of environmental factors on telescopic joint state parameters is analyzed, and the telescopic joint state parameters are corrected; for abnormal state caused by mechanical failure, vibration signal analysis and stress distribution analysis method are used to identify mechanical failure type and failure position, evaluate failure severity, and trigger corresponding warning information according to the severity of the failure; for performance degradation caused by material aging, the material performance parameters and stress history data are used to monitor the state in real time, predict the performance degradation and remaining service life of the telescopic joint caused by material aging, and develop maintenance strategies according to the prediction results; periodically evaluate the performance of the model, update and optimize the model parameters according to the new operation monitoring data; dynamically adjust the warning threshold to realize adaptive warning; develop a GIL pipe gallery telescopic joint data integration and visualization platform, integrate multi-source monitoring data, and display the telescopic joint state through a graphical interface to realize visual display and interactive analysis of monitoring data.

2. The sensor-based online monitoring method of GIL tube gallery expansion joint according to claim 1, characterized in that, The method comprises the following steps: deploying telescopic joint sensors and environmental sensors to collect key state information of the telescopic joint and surrounding environmental parameters affecting the state of the telescopic joint, including stress, displacement, temperature, vibration, external temperature, humidity and air pressure, and performing preliminary processing on the collected data, including filtering and denoising, which comprises:

3. The sensor-based online monitoring method of GIL tube gallery expansion joint according to claim 1, characterized in that, deploying telescopic joint sensors and environmental sensors, the telescopic joint sensors including stress sensors, displacement sensors, temperature sensors and vibration sensors, and the environmental sensors including external temperature sensors, humidity sensors and air pressure sensors, collecting stress, displacement, temperature, vibration, external temperature, humidity and air pressure data through the sensors; pre-processing the collected data, removing noise and drift in the data by using filtering technology, and removing noise and interference in the data by using wavelet denoising method, selecting appropriate wavelet basis function and decomposition level. The method comprises the following steps: time synchronization and spatial position alignment are performed on the pre-processed telescopic joint sensor data and environmental sensor data to construct a unified space-time data matrix; The correlation analysis algorithm is used to extract key fusion features reflecting the state change of the expansion joint, and a multi-source fusion feature set containing expansion joint parameters and environmental parameters is constructed to describe the comprehensive operating state. Based on the fusion features, a long short-term memory network model capable of evaluating the comprehensive state and life prediction is trained to realize the health score of the current state of the expansion joint and the fault type judgment, and to predict the future state change trend and the degradation of the key performance indicators.

4. The sensor-based online monitoring method of GIL tube gallery expansion joint according to claim 1, characterized in that, The model-based analysis of the influence of different factors on the state of the expansion joint is divided into environmental factors, mechanical failure factors and material aging factors. For different types of influencing factors, corresponding analysis methods and processing strategies are adopted, including: The state parameters and environmental variables collected during the operation of the expansion joint are classified and labeled to construct a data label system corresponding to the three main influencing factors. A model-based feature selection algorithm is used to identify the dominant variables of different types of factors affecting the state change of the expansion joint, and to divide the influence source type and the corresponding key feature vector; The current state of the expansion joint is classified and judged to identify whether it belongs to environmental fluctuation, mechanical structural abnormality or material performance degradation. According to the identified type of influencing factor, the corresponding processing strategy is selected for targeted analysis and processing.

5. The sensor-based online monitoring method of GIL tube gallery expansion joint according to claim 1, characterized in that, For the state change caused by environmental factors, the influence of environmental factors on the state parameters of the expansion joint is analyzed, and the state parameters of the expansion joint are corrected, including: According to the comprehensive state evaluation and prediction model, the influence of environmental factors on the stress, displacement, temperature and vibration state parameters of the expansion joint is analyzed, the environmental features related to the state parameters of the expansion joint are selected, and the selected features are used to train a random forest model. Each tree uses the bootstrap sampling method during training and selects the optimal feature during node splitting. The model performance is evaluated by cross-validation, and the number and depth of trees are adjusted to optimize the model. A quantitative relationship model between environmental factors and expansion joint state parameters is established, and the trained random forest model is used to calculate the influence of environmental factors on the state parameters of the expansion joint. The formula is: ； where ΔS i is the influence of environmental factors on the state parameter of the ith expansion joint, f is the random forest model, and E is the environmental factor vector. The state parameters of the expansion joint are dynamically corrected, and the formula is: ； wherein S i is the corrected value of the i-th telescopic joint state parameter, S i is the raw value of the i-th state parameter measured by the sensor in real time; Based on historical data statistics, set the normal range threshold of each state parameter, monitor and evaluate the modified telescopic joint state parameter in real time, when S i ′ is in the normal range, evaluate as normal state, when S i ′ exceeds the normal range, evaluate as abnormal state.

6. The sensor-based online monitoring method of GIL conduit expansion joints according to claim 1, characterized in that, For the abnormal state caused by mechanical failure, vibration signal analysis and stress distribution analysis methods are used to identify the type and location of mechanical failure, evaluate the severity of the failure, and trigger corresponding warning information according to the severity of the failure, including: When the expansion joint enters an abnormal state, the corresponding high-frequency vibration signal and stress data are extracted for analysis; The vibration signal is processed by time domain analysis, frequency domain analysis and time-frequency domain analysis. By analyzing the amplitude, frequency component, phase and impact characteristics of the vibration signal, it is compared with the vibration feature library of loose, wear, misalignment and crack fault modes to identify the possible mechanical failure type and approximate location. Combined with the spatial distribution and real-time readings of the stress sensor, the stress distribution curve of the key parts of the expansion joint is drawn to analyze the stress concentration area, the size and change trend of the stress value, and to find the high stress concentration area to further determine the fault location; In combination with the results of vibration signal analysis and stress distribution analysis, comprehensive judgment is performed, based on the identified fault type, location and related vibration, stress parameter values, a fault severity level standard is divided, including slight fault, medium fault and serious fault, and the severity of the current fault is evaluated; a slight fault, medium fault and serious fault range is set, when being in the slight fault range, it is a slight fault; when being in the medium fault range, it is a medium fault; when being in the serious fault range, it is a serious fault; According to the evaluated fault severity level, different levels and forms of early warning information are automatically triggered; when the level is a slight fault, a notification is sent to the maintenance personnel, prompting attention; when the level is a medium fault, a medium fault alarm is triggered, prompting the maintenance personnel to arrange a near-term inspection; when the level is a serious fault, a serious fault alarm is triggered, requiring immediate shutdown inspection and taking emergency measures.

7. The sensor-based online monitoring method of GIL conduit expansion joints according to claim 1, characterized in that, For performance degradation caused by material aging, material performance parameters and stress historical data are used to monitor the state in real time, predict performance degradation and remaining useful life of the expansion joint caused by material aging, and develop a maintenance strategy according to the prediction results, including: According to the comprehensive state evaluation and prediction model, a relationship model between material performance parameters and stress historical data is established, which uses a long short-term memory network model based on deep learning to reflect the aging law of the material in the long-term operation process; the key state parameters of the expansion joint collected by the sensor, including stress, strain and temperature, are combined with the material performance parameters and stress historical data, and the current state of the expansion joint is evaluated by a state evaluation method based on a physical information neural network; based on the evaluation results, the relationship model is used to predict the performance degradation and remaining useful life of the expansion joint in a future period of time, and a corresponding maintenance strategy is developed according to the prediction results, including planned replacement, repair reinforcement and enhanced monitoring.

8. The sensor-based online monitoring method of GIL tunnel expansion joint according to claim 1, characterized in that, The model performance is periodically evaluated, and the model parameters are updated and optimized according to new operation monitoring data; The early warning threshold is dynamically adjusted to realize adaptive early warning, including: The model prediction results are compared with the actual monitoring data to calculate the accuracy and error of the model prediction, and the accuracy of the model prediction of the state of the expansion joint is evaluated; new operation monitoring data is used to update and optimize the model parameters by using an incremental gradient descent algorithm; According to the model performance evaluation results, the actual operation state of the expansion joint and the historical early warning information, the early warning threshold is dynamically adjusted; when the model prediction accuracy decreases or the operation state of the expansion joint deteriorates, the early warning threshold is lowered; when the model prediction accuracy improves or the operation state of the expansion joint is good, the early warning threshold is raised; Through periodic evaluation of the model performance, updating of the model parameters and dynamic adjustment of the early warning threshold, the early warning strategy can be adaptively adjusted according to the actual operation state of the expansion joint, the changes in the model performance and the historical early warning effect.

9. The sensor-based online monitoring method of GIL tube gallery expansion joint according to claim 1, characterized in that, The GIL pipe gallery expansion joint data integration and visualization platform is developed, multi-source monitoring data is integrated, and the state of the expansion joint is displayed in a graphical interface to realize visual display and interactive analysis of the monitoring data, including: The telescopic joint state parameter data is integrated and displayed in the form of charts, curves and maps, including real-time data, historical data and trend analysis; interactive analysis functions are provided for in-depth analysis and exploration of the data, including viewing data for a specific time period, comparing data from different sensors and analyzing the trend of the data; warning information is displayed on the graphical interface, and corresponding operation prompts are provided, including confirming the warning, viewing detailed information and recording the processing process.

10. A sensor-based online monitoring system for GIL pipe gallery expansion joint, using the sensor-based online monitoring method for GIL pipe gallery expansion joint according to any one of claims 1-9, characterized in that, It comprises: a data acquisition module, including a telescopic joint sensor unit, an environmental sensor unit and a data preprocessing unit; wherein the telescopic joint sensor unit collects key state information parameter data of the telescopic joint, including stress, displacement, temperature and vibration; the environmental sensor unit collects surrounding environmental parameter data that affect the state of the telescopic joint, including external temperature, humidity and air pressure; the data preprocessing unit performs preliminary processing on the collected data, including filtering and denoising; a data fusion and modeling module, including a data synchronization and alignment unit, a feature extraction unit and a state evaluation and prediction unit; wherein the data synchronization and alignment unit synchronizes the telescopic joint sensor data and the environmental sensor data in time and aligns them in space position, constructing a unified spatio-temporal data matrix; the feature extraction unit uses correlation analysis algorithm to extract key fusion features reflecting the state change of the telescopic joint, and constructs a multi-source fusion feature set; the state evaluation and prediction unit trains a long short-term memory network model based on the fusion features, which can evaluate the comprehensive state and life prediction, and predicts the future state change trend and key performance indicators of the telescopic joint; a state analysis and processing module, including an environmental factor analysis unit, a mechanical fault analysis unit and a material aging analysis unit; wherein the environmental factor analysis unit analyzes the influence of environmental factors on the state parameters of the telescopic joint and makes corrections; the mechanical fault analysis unit uses vibration signal analysis and stress distribution analysis methods to identify the type and location of mechanical faults, evaluate the severity of the faults and trigger warning information; the material aging analysis unit uses material performance parameters and stress historical data to monitor the state in real time, predict the performance degradation and remaining service life of the telescopic joint due to material aging, and develop maintenance strategies; a model evaluation and optimization module, including a model performance evaluation unit, a model parameter updating unit and a warning threshold adjustment unit; wherein the model performance evaluation unit compares the model prediction results with the actual monitoring data, calculates the accuracy and error, and evaluates the accuracy of the model in predicting the state of the telescopic joint; the model parameter updating unit uses new operation monitoring data and adopts incremental gradient descent algorithm to update and optimize the model parameters; the warning threshold adjustment unit dynamically adjusts the warning threshold according to the model performance evaluation results, the actual operation state of the telescopic joint and the historical warning information; The data visualization and interaction platform module comprises a data integration unit, a visualization display unit, an interactive analysis unit and a warning information display unit; the data integration unit integrates the telescopic joint state parameter data; the visualization display unit displays the data in the form of a chart, a curve and a map; the interactive analysis unit provides an interactive analysis function for in-depth analysis and exploration of the data; and the warning information display unit displays the warning information on a graphical interface and provides corresponding operation prompts.

Images