Substation trench cover data acquisition and intelligent analysis management system and method

Abstract

The present application relates to the technical field of substations, and specifically discloses a data collection and intelligent analysis management system and method for substation trench cover plates, the method comprising: collecting cover plate vibration data and strain data through acceleration sensors and resistive strain gauges; extracting vibration direction changes, vibration attenuation, and cover plate deformation rebound time from acceleration data, and extracting maximum amplitude, change rate, and strain gradient from strain data to comprehensively characterize the stress and deformation characteristics of the cover plate; based on the stress, deformation, and rebound characteristics of the cover plate, a damage assessment model is established using a neural network model to achieve intelligent discrimination of the degree of damage to the cover plate; using a neural network to predict damage change trends and calculate the acceleration trend of cover plate damage; based on the damage assessment results and the damage acceleration trend, the remaining maintenance time of the cover plate is calculated, and an operation and maintenance warning is automatically triggered when it is lower than the warning threshold. The method of the present application can achieve intelligent monitoring, prediction, and active operation and maintenance management of the health status of the cover plate.

Description

Technical Field

[0001] The present application relates to the technical field of substations, and specifically to a data collection and intelligent analysis management system and method for substation trench cover plates. Background Art

[0002] Substation trench covers are a crucial component of power system infrastructure, primarily used to cover cable trenches and protect the safe operation of the cable infrastructure within. These covers are typically made of composite materials and are subject to long-term damage from vehicles, people, and environmental factors.

[0003] In actual operation and maintenance, the condition monitoring of substation trench cover plates primarily relies on manual inspections. Maintenance personnel visually inspect the cover plates for visible defects such as cracks and breakage to determine their condition. This traditional inspection method can detect visible damage, but it struggles to detect minor damage and performance degradation within the cover plates. Over long-term use, the cover plates' load-bearing capacity and structural performance gradually decline. If these defects are not detected and replaced promptly, they can break, impacting the substation's safe operation.

[0004] Therefore, how to accurately assess the damage status of the substation channel cover and how to scientifically predict the maintenance timing of the cover have become technical problems that need to be solved in the operation and maintenance management of substations.

[0005] In view of this, the present application proposes a system and method for collecting and intelligently analyzing substation trench cover data. Summary of the Invention

[0006] To achieve the above objectives, this application provides a substation trench cover data collection and intelligent analysis management system and method. The specific technical solutions are as follows:

[0007] Substation trench cover data collection and intelligent analysis management method, including:

[0008] Acceleration sensors and resistance strain gauges are embedded in the cover to collect vibration and strain data of the cover;

[0009] Extract the vibration direction change and vibration attenuation time of the acceleration sensor, calculate the deformation rebound time of the cover after the force is stopped, and extract the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate, and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover under stress.

[0010] According to the stress-deformation and rebound characteristics of the cover plate, a cover plate damage assessment model based on stress-deformation and deformation rebound time is constructed based on the neural network model to evaluate the damage degree of the cover plate;

[0011] By integrating the characteristics of vibration data and strain data, the machine learning model is used to predict the trend of damage degree changes and calculate the acceleration trend of cover damage;

[0012] Based on the currently assessed degree of cover plate damage and the accelerating trend of cover plate damage, the remaining time for cover plate maintenance is calculated. When the remaining time for maintenance is lower than the set threshold, an operation and maintenance warning message is triggered.

[0013] Preferably, an acceleration sensor is mounted at the geometric center of the lower surface of the cover plate, forming a rigid connection with the cover plate, and is used to collect vibration data of the cover plate;

[0014] A strain rosette consisting of multiple resistive strain gauges is set at the geometric center of the cover, and unidirectional strain gauges are arranged along the diagonal direction at the four corners of the cover to synchronously obtain strain data in the central area and corner areas of the cover.

[0015] Preferably, after the collected cover plate strain data is subjected to a Butterworth low-pass filter to remove high-frequency noise, a traversal search is performed within a fixed time window to obtain the maximum strain amplitude time series, and the corresponding strain change rate is calculated using the multi-scale central difference method. The maximum strain amplitude and strain change rate are used to characterize the magnitude and rate of the cover plate's stress deformation.

[0016] Preferably, an interpolation strain field is established based on the strain values ​​of the four corner measuring points, and the strain gradients in each direction are obtained after least square fitting;

[0017] The covariance matrix of the acceleration data is constructed and the eigendecomposition is performed to determine the main vibration direction. The vibration attenuation time is obtained through the free vibration envelope attenuation analysis. The acceleration is then integrated twice and combined with high-pass filtering to determine the cover deformation rebound time.

[0018] Preferably, the maximum strain amplitude, strain change rate, corner strain gradient, vibration direction angle, vibration attenuation time and deformation rebound time are combined into a multidimensional feature vector; the comprehensive damage index is used as a supervision label, and the support vector regression algorithm is used to construct a cover damage assessment model.

[0019] Preferably, the penalty coefficient and kernel function width of the support vector regression model are optimized by grid search combined with cross validation, the root mean square error is used as the performance evaluation index, and the trained support vector regression model is used as the cover plate damage assessment model;

[0020] A confidence evaluation mechanism is established. When the cover damage assessment model outputs the prediction result of the cover damage, the confidence of the cover damage prediction result is calculated according to the threshold of the distance between the real-time sample and the support vector in the training set.

[0021] Preferably, the comprehensive damage index is stored in chronological order to form a damage time series; a long short-term memory network LSTM model is used to construct a damage trend prediction model, the input of the LSTM model is the historical damage time series, and the output is a sequence of cover damage trends in future time periods.

[0022] Preferably, the damage acceleration factor is obtained by calculating the cover plate damage growth rate and change rate;

[0023] When the change in damage growth rate exceeds the set threshold over multiple consecutive moments, it is determined to be in the acceleration stage; if the prediction error over multiple consecutive moments exceeds the preset threshold, the online incremental learning mechanism is triggered, and the long short-term memory network (LSTM) model parameters are updated using the latest monitoring data.

[0024] Preferably, the current damage rate is obtained by linear fitting of the comprehensive damage index within a preset recent time window, a damage development function is constructed, and the remaining maintenance time for the cover plate to reach the maintenance threshold is calculated;

[0025] Correct the remaining maintenance time based on the damage acceleration factor and the introduction of a safety margin factor, and set multi-level warning thresholds;

[0026] When the corrected remaining maintenance time is not greater than the corresponding warning threshold, hierarchical operation and maintenance warning information is generated, including the cover plate identification, current damage level, damage acceleration trend and recommended maintenance time window, and a time interval is set to suppress repeated warnings for the same cover plate.

[0027] Substation trench cover data acquisition and intelligent analysis management system, which is used in the substation trench cover data acquisition and intelligent analysis management method, including: a data acquisition module, a feature extraction module, a cover damage assessment module, a cover damage trend prediction module and an early warning module;

[0028] The data acquisition module embeds an acceleration sensor and a resistance strain gauge in the cover plate to collect vibration data and strain data of the cover plate;

[0029] The feature extraction module extracts the vibration direction change and vibration attenuation time of the acceleration sensor, calculates the deformation rebound time of the cover plate after the force is stopped, and extracts the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate, and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover plate under stress;

[0030] The cover plate damage assessment module constructs a cover plate damage assessment model based on stress deformation and deformation rebound time based on a neural network model according to the stress deformation and rebound characteristics of the cover plate, and assesses the degree of cover plate damage;

[0031] The cover plate damage trend prediction module integrates the features of vibration data and strain data, predicts the damage degree change trend based on the machine learning model, and calculates the acceleration trend of cover plate damage;

[0032] The warning module calculates the remaining maintenance time of the cover plate based on the currently assessed damage level of the cover plate and the acceleration trend of the cover plate damage, and triggers an operation and maintenance warning message when the remaining maintenance time is lower than a set threshold.

[0033] Beneficial effects of this application: This application collects cover plate vibration and strain data by embedding sensors, thereby realizing real-time and accurate monitoring of the cover plate's stress, vibration and deformation state, providing a comprehensive and reliable raw data basis for subsequent analysis, and effectively improving the comprehensiveness and accuracy of data collection.

[0034] This application realizes in-depth quantitative analysis of the cover plate's force response, deformation process, and rebound performance through multi-dimensional feature extraction of acceleration and strain signals, providing rich physical quantity feature support for damage characteristic discrimination and enhancing the ability to refine the structural state of the cover plate.

[0035] This application uses a machine learning model to model the stress deformation and rebound characteristics of the cover plate, thereby achieving quantitative and intelligent assessment of the degree of damage to the cover plate. It can timely detect early signs of degradation of the cover plate performance, and improve the accuracy and intelligence level of damage identification.

[0036] By integrating multi-source features and performing trend prediction, this application can identify the accelerated development trend of cover plate damage in advance, achieve dynamic tracking of the cover plate health status and risk prediction, and provide a scientific basis for subsequent operation and maintenance decisions.

[0037] This application intelligently calculates the remaining maintenance time of the cover plate based on the degree of damage and acceleration trend, and proactively pushes warnings at critical moments, effectively realizing proactive management and risk prevention of cover plate operation and maintenance, greatly improving the foresight and efficiency of operation and maintenance work.

[0038] This application realizes the integration of status monitoring, health assessment and intelligent operation and maintenance management of trench cover plates in substations; the technical solution of this application can accurately grasp the stress and damage status of the cover plates in real time, predict the potential structural risks of the cover plates in advance, significantly improve the scientificity, timeliness and economy of cover plate maintenance, effectively ensure the safety of substation equipment and operation and maintenance personnel, and provide strong support for the intelligent management and refined operation and maintenance of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Flowchart of the substation trench cover data collection and intelligent analysis management method provided for this application;

[0040] Figure 2Flowchart for extracting strain data and vibration data features provided for this application;

[0041] Figure 3 A flowchart for constructing a cover damage assessment model provided for this application;

[0042] Figure 4 A flowchart for constructing a cover damage trend prediction provided for this application;

[0043] Figure 5 Flowchart for calculation and early warning of the remaining maintenance time for the cover provided in this application;

[0044] Figure 6 Structural diagram of the substation trench cover data acquisition and intelligent analysis management system provided for this application. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the drawings in the specification.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without violating the connotation of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present application. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0048] Example 1:

[0049] Reference Figures 1 to 5 , which is the first embodiment of this application, provides a method for collecting and intelligently analyzing substation trench cover data.

[0050] Step 1: Embed an acceleration sensor and a resistance strain gauge in the cover to collect vibration data and strain data of the cover.

[0051] A high-precision three-axis acceleration sensor is installed at the geometric center of the lower surface of the cover to ensure that the vibration response of the cover when subjected to external force can be accurately captured; illustratively, the acceleration sensor is rigidly connected to the cover through epoxy resin adhesive or a preset card slot to ensure lossless transmission of the vibration signal.

[0052] A resistance strain gauge is placed on the lower surface of the cover. For example, a foil strain gauge with a resistance value of 350Ω is used. A strain rosette is placed at the geometric center of the cover. It consists of three resistance strain gauges arranged at 0°, 45°, and 90° to measure the principal strain direction and magnitude of the geometric center point. The arrangement of the strain rosette can accurately obtain the two-dimensional strain state of the central area of ​​the cover, and the principal strain value is obtained by calculation. and ,in , where 、 are the normal strains in the 0° and 90° directions, , It is the shear strain in the 45° direction; the layout of the strain rosette can fully reflect the complex stress state of the center of the cover plate.

[0053] For the key stress concentration areas of the cover plate, the four corner areas of the cover plate are taken as stress concentration points. A unidirectional strain gauge is arranged at each corner edge of the cover plate, and the long axis direction of the strain gauge is arranged along the diagonal direction of the cover plate. By arranging strain gauges at the corner edges of the cover plate, the stress concentration phenomenon in the corner area of ​​the cover plate can be effectively monitored. This is because the corners of the cover plate are often the first location to be damaged when the cover plate is subjected to load.

[0054] In this step, sensors are deployed and data is collected on the cover material to establish a complete monitoring system for the cover's stress deformation and vibration response. This system is used to quantitatively characterize the mechanical behavior of the cover under load, providing high-quality basic data support for subsequent damage assessment and operation and maintenance management.

[0055] Step 2: Extract the vibration direction change and vibration attenuation time of the acceleration sensor, calculate the deformation rebound time after the cover stops being stressed, and extract the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate, and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover under stress; see Figure 2 , which is the flow chart of extracting strain data and vibration data features in this step.

[0056] The collected raw strain data are preprocessed and a Butterworth low-pass filter is used to filter out high-frequency noise to ensure that the effective strain data are fully retained while eliminating measurement noise interference.

[0057] For the filtered strain data , in the time interval Search for the maximum strain amplitude within , where is the independent variable, indicating time; To analyze the time window length, is the subscript index; the maximum strain amplitude is defined as , by traversing the entire monitoring period, the maximum strain amplitude time series is obtained ,in, is the maximum strain amplitude time series The characteristic value of the maximum strain amplitude directly reflects the maximum deformation degree of the cover plate in each time period, and provides a quantitative indicator for evaluating the bearing capacity of the cover plate.

[0058] The central difference method is used to calculate the strain change rate. discretely sampled strain data points , strain change rate The calculation formula is ,in is the sampling time interval, is the difference step size parameter.

[0059] In order to improve the calculation accuracy, the multi-scale difference method is used to calculate The strain change rate at that time is then calculated by weighted average to obtain the final strain change rate. ,in, For the The strain rate of the scale, is the weight coefficient and satisfies The magnitude of the strain change rate reflects the severity of the load. An excessively large strain change rate indicates that the cover plate has been subjected to an impact load, and the form of the impact load has a more significant effect on the damage to the cover plate.

[0060] Calculate the strain gradient in the stress concentration area and establish a local coordinate system using the strain measurement data at the four corners of the cover plate , where the origin is located at the geometric center of the cover. The strain field function is constructed by bilinear interpolation , where the coefficient The least squares fitting strain field function is obtained by testing at different positions in the cover plate to obtain the actual strain value and position of each strain measuring point at the four corners. The parameters in the calculation data.

[0061] The strain gradient in The direction component is ,exist The direction component is ; In each corner region, define the side length as In the square calculation domain, multiple calculation points are uniformly selected in the square calculation domain to calculate the strain gradient of each point. ,Pick The average value of is taken as the strain gradient characteristic value of the corner; a high strain gradient value indicates that there is stress concentration in this area, which is a potential damage development area.

[0062] The vibration direction change of the cover is analyzed, and the acceleration components measured by the three-axis acceleration sensor are analyzed through the covariance matrix , construct the covariance matrix ,in , is the analysis duration; perform eigenvalue decomposition on the covariance matrix ,in is the eigenvector matrix, is the eigenvalue diagonal matrix; the eigenvector corresponding to the largest eigenvalue That is the main vibration direction, is the eigenvector The three components correspond to the three-axis acceleration directions of X, Y, and Z.

[0063] Defining the vibration direction angle includes and , , is the angle with the vertical, and , Is the horizontal projection azimuth; continuously monitor the changes in vibration direction angle and ,in, Indicates the change in the angle between the main vibration direction at the current moment and the main vibration direction at the initial moment in the z-axis direction. Indicates the change in direction between the main vibration direction at the current moment and the main vibration direction at the initial moment in the horizontal projection plane (xy plane). Indicates the tilt angle of the main vibration direction in the z-axis direction at the current moment, Indicates the inclination angle of the main vibration direction in the z-axis direction at the initial moment, Indicates the direction angle of the main vibration direction in the xy plane at the current moment, It represents the orientation angle of the main vibration direction in the xy plane at the initial moment; it is used to identify changes in the cover support conditions and uneven degradation of the structural stiffness.

[0064] Extract vibration decay time based on free vibration response analysis and identify the moment when load is removed. ,definition The acceleration amplitude drops from the peak value to the threshold value moments, in which is the threshold coefficient, Indicates the peak acceleration. Start extracting free vibration segment data and calculating vibration envelope ; Simultaneous vibration envelope Satisfies the exponential decay form , logarithmically transform the vibration envelope , is the initial value of the vibration envelope, and the damping ratio is determined by linear regression and natural frequency .

[0065] The vibration decay time is defined as the time it takes for the vibration envelope to decay to the initial value. times the time required, i.e. ,in, is the vibration decay time, is the attenuation proportional coefficient; the increase in vibration decay time indicates an increase in structural damping, which is usually caused by increased energy dissipation caused by internal damage to the cover.

[0066] Calculate the deformation rebound time of the cover plate, integrate the acceleration data to obtain the displacement response; use the trapezoidal integration method for numerical integration, and the velocity , displacement ,in and are the initial velocity and initial displacement, The total number of discrete time points, It is in The acceleration value sampled at a time point, In the The speed value obtained by sampling at a time point.

[0067] In order to eliminate the integral drift, a high-pass filter is used to process the displacement signal and identify the moment when the load is completely removed. ,Will It is defined as the moment when the acceleration returns to the static noise level, from Start monitoring the displacement recovery process at all times.

[0068] The static equilibrium position is defined as ,in For waiting time, is the average time window; the deformation rebound time is defined as the time when the displacement recovers from the maximum value to The time required, i.e. ,in is the convergence judgment threshold, is the deformation rebound time, The displacement recovers from the maximum value to The extension of deformation rebound time reflects the decline of elastic recovery ability of the cover plate and is an important indicator of the degradation of cover plate material performance.

[0069] This step achieves a comprehensive quantitative characterization of the static strain characteristics, dynamic vibration characteristics, and elastic recovery characteristics of the cover plate; it reflects the mechanical performance state of the cover plate from the perspective of multi-dimensional characteristic parameters.

[0070] Step 3: Based on the stress-deformation characteristics and rebound characteristics of the cover plate, a cover plate damage assessment model based on stress-deformation and deformation rebound time is constructed based on the neural network model to assess the damage degree of the cover plate; see Figure 3 , which is the flowchart of constructing the cover damage assessment model in this step.

[0071] The maximum strain amplitude , strain change rate , strain gradient , vibration direction change angle and , vibration decay time and deformation rebound time Combined into multidimensional feature vector .

[0072] In order to establish a supervised learning model, a quantitative indicator of the damage degree of the cover plate is defined and a comprehensive damage index is used. As a damage label, it is defined as ,in is the ultimate strain value of the cover plate, is the standard deformation rebound time of the new cover, is the critical value of the strain gradient, is the weight coefficient and satisfies ; Comprehensive injury index The value range is ,in Indicates no damage. Indicates serious damage. Through a combination of expert evaluation and on-site testing, accurate damage labels are annotated for historical data to form a training data set. ,in is the sample size.

[0073] The support vector regression (SVR) model is used to construct the cover plate damage assessment model, and the radial basis function (RBF) is selected as the kernel function. The SVR model can effectively handle nonlinear relationships, has good robustness to noise, and is suitable for processing complex mapping relationships in cover plate damage assessment.

[0074] The grid search combined with cross-validation method is used to optimize the hyperparameters of the SVR model, and the parameter grid is formed by uniform sampling on the logarithmic scale. The performance of each group of parameters is evaluated by k-fold cross-validation, and the performance indicator is the root mean square error (RMSE). The parameter combination that minimizes the cross-validation RMSE is selected as the optimal parameter.

[0075] Extract multidimensional feature vectors from real-time collected cover monitoring data , input the trained SVR model to obtain the damage prediction value ,in For the support vectors, is the kernel function, Support vector and the input multidimensional feature vector The inner product between Bias term, is the Lagrange multiplier, is the number of support vectors, is the support vector index.

[0076] In order to verify the reliability of the cover damage assessment model, a confidence assessment mechanism is established by analyzing the distance between the new sample and the support vector of the training set. ,when When , it indicates that the new sample deviates from the training data distribution. The distance threshold between the new sample and the support vector of the training set is set to issue a low confidence warning. The establishment of a confidence assessment mechanism can identify the applicable boundaries of the model and avoid erroneous damage assessment results under abnormal working conditions.

[0077] This step builds a machine learning-based cover plate damage assessment model to accurately map multidimensional monitoring data to damage severity, overcoming the limitations of traditional threshold judgment methods. The constructed cover plate damage assessment model comprehensively considers the static deformation and dynamic response characteristics of the cover plate, accurately identifying different degrees of cover plate damage.

[0078] Step 4: Integrate the characteristics of vibration data and strain data, predict the damage degree change trend based on the machine learning model, and calculate the acceleration trend of cover damage; see Figure 4 , a flowchart of cover damage trend prediction is constructed for this step.

[0079] Comprehensive injury index Store in chronological order to form a damage time series ,in For the The monitoring time, The total number of monitoring times.

[0080] In order to accurately capture the damage evolution law, the long short-term memory network (LSTM) model is used to construct a damage trend prediction model. The input of the LSTM model is the historical damage time series, and the output of the LSTM is the sequence of cover plate damage trends in future time periods.

[0081] Constructing the input feature matrix ,in The length is The LSTM network can effectively learn the long-term dependency of damage development and overcome the gradient vanishing problem of traditional time series models.

[0082] Analyze the damage acceleration trend based on the second-order derivative of the damage growth rate and calculate the damage growth rate , the damage acceleration rate is defined as ; When continuous Time point satisfaction When , it is determined to be a state of accelerated damage growth, where is the acceleration growth rate threshold; defines the damage acceleration factor ,in is the average damage growth rate in the current period, is the average damage growth rate during the reference period; When , it indicates that the cover damage enters a rapid growth stage. is the damage acceleration threshold.

[0083] In order to improve the adaptive ability of the prediction model, an online learning mechanism is constructed and the prediction error is defined. , when continuous The prediction error at each time point satisfies hour, For time The actual damage value when For time The predicted damage value when is the error threshold, triggering the model update. Using the incremental learning strategy, the latest data points are added to the training set, and the earliest data points in the training set time series are removed. data points, keeping the size of the training set constant.

[0084] This step builds a deep learning-based system for predicting damage trends, determining damage, and assessing failure risk, enabling accurate prediction of the future health of the cover plate and quantitative risk assessment. This method not only predicts damage trends but also identifies critical periods of accelerated damage.

[0085] Step 5: Based on the current assessed damage level of the cover plate and the acceleration trend of the cover plate damage, calculate the remaining maintenance time of the cover plate. When the remaining maintenance time is lower than the set threshold, the operation and maintenance warning information is triggered; see Figure 5 , which is the flow chart of calculation and early warning of the remaining maintenance time of the cover in this step.

[0086] Based on the current level of damage and injury acceleration trends , build a calculation model for the remaining maintenance time of the cover.

[0087] For example, the damage development function is defined as ,in is the current damage rate, Indicates the time period starting from the current moment; the current damage rate is calculated by the recent The damage data of the time sampling points are linearly fitted to obtain: ,in and are the mean of time and damage, is the time sampling point index; set the damage threshold that the cover needs to maintain to , by solving the equation Get the remaining maintenance time :

[0088]

[0089] The calculation method based on physical degradation laws can relatively accurately predict the time point when the cover reaches the maintenance standard, providing a reliable basis for formulating maintenance plans.

[0090] Considering the uncertainty of damage development, the safety margin factor is introduced to correct the remaining maintenance time. Define the corrected remaining maintenance time ,in is the safety margin factor, and its value range is .

[0091] The safety margin coefficient is dynamically adjusted according to the damage acceleration trend. ,in =( ...

[0092] Set the time threshold for triggering operation and maintenance warnings , threshold Comprehensively consider the maintenance preparation time and safety reserve time; when the remaining maintenance time after correction Below the warning threshold When , triggering an operation and maintenance warning.

[0093] Warning information includes cover identification, current damage level , the trend of accelerated damage , Estimated remaining maintenance time and recommended maintenance time windows ,in , , is the half-width of the time window.

[0094] For example, the warning level can be divided into three levels according to the urgency of the remaining maintenance time: emergency warning ( )、Important Warning( ), General Warning ( ),in The hierarchical warning mechanism can help operation and maintenance personnel reasonably arrange maintenance priorities and optimize resource allocation.

[0095] After the warning is triggered, the warning information will be pushed to the operation and maintenance management platform. In order to avoid information overload caused by frequent warnings, a warning suppression mechanism is set up. The system filters out duplicate warnings within the system and only retains the highest-level warning information; the intelligent warning release mechanism ensures that operation and maintenance personnel can obtain key information in a timely and accurate manner, and avoids unnecessary interruptions caused by information redundancy.

[0096] This step achieves accurate prediction and timely warning of cover plate operation and maintenance needs through the technical solution of calculating the remaining maintenance time, considering the time correction of safety margin, triggering warning by graded threshold comparison, and intelligently providing warning information; it can effectively avoid the safety risks brought by cover plate failure, and at the same time optimize the scheduling of maintenance resources through a reasonable warning mechanism, significantly improving the operation and maintenance management efficiency and reliability of substation trench cover plates.

[0097] Example 2:

[0098] Reference Figure 6 , which is the second embodiment of the present application, provides a substation trench cover data collection and intelligent analysis and management system.

[0099] The system includes: a data acquisition module, a feature extraction module, a cover plate damage assessment module, a cover plate damage trend prediction module and an early warning module.

[0100] The data acquisition module embeds an acceleration sensor and a resistance strain gauge in the cover plate to collect vibration data and strain data of the cover plate.

[0101] The feature extraction module extracts the vibration direction change and vibration attenuation time of the acceleration sensor, calculates the deformation rebound time after the cover plate stops being subjected to force, and extracts the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover plate under stress.

[0102] The cover plate damage assessment module constructs a cover plate damage assessment model based on stress deformation and deformation rebound time based on a neural network model according to the stress deformation characteristics and rebound characteristics of the cover plate to assess the degree of cover plate damage.

[0103] The cover plate damage trend prediction module integrates the features of vibration data and strain data, predicts the damage degree change trend based on a machine learning model, and calculates the acceleration trend of cover plate damage.

[0104] The warning module calculates the remaining maintenance time of the cover plate based on the currently assessed damage level of the cover plate and the acceleration trend of the cover plate damage, and triggers an operation and maintenance warning message when the remaining maintenance time is lower than a set threshold.

[0105] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0106] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose of this application and the scope of protection of the claims, which are all within the protection of this application.

Claims

1. Substation trench cover data collection and intelligent analysis management method, characterized by: include: Acceleration sensors and resistance strain gauges are embedded in the cover to collect vibration and strain data of the cover; Extract the vibration direction change and vibration attenuation time of the acceleration sensor, calculate the deformation rebound time of the cover after the force is stopped, and extract the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate, and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover under stress. The interpolation strain field is established based on the strain values ​​of the four corner measuring points, and the strain gradients in each direction are obtained after least square fitting. The covariance matrix of the acceleration data is constructed and eigendecomposition is performed to determine the main vibration direction. The vibration attenuation time is obtained through free vibration envelope attenuation analysis. The acceleration is then integrated twice and combined with high-pass filtering to determine the cover deformation rebound time. According to the stress-deformation and rebound characteristics of the cover plate, a cover plate damage assessment model based on stress-deformation and deformation rebound time is constructed based on the neural network model to evaluate the damage degree of the cover plate; By integrating the characteristics of vibration data and strain data, the machine learning model is used to predict the trend of damage degree changes and calculate the acceleration trend of cover damage; Based on the currently assessed degree of cover plate damage and the accelerating trend of cover plate damage, the remaining time for cover plate maintenance is calculated. When the remaining time for maintenance is lower than the set threshold, an operation and maintenance warning message is triggered.

2. The substation trench cover data collection and intelligent analysis management method according to claim 1 is characterized in that: An acceleration sensor is installed at the geometric center of the lower surface of the cover plate to form a rigid connection with the cover plate to collect vibration data of the cover plate; A strain rosette consisting of multiple resistive strain gauges is set at the geometric center of the cover, and unidirectional strain gauges are arranged along the diagonal direction at the four corners of the cover to synchronously obtain strain data in the central area and corner areas of the cover.

3. The substation trench cover data collection and intelligent analysis management method according to claim 2 is characterized in that: After using a Butterworth low-pass filter to remove noise from the collected cover plate strain data, a traversal search is performed within a fixed time window to obtain the maximum strain amplitude time series. The corresponding strain change rate is calculated using the multi-scale central difference method, and the maximum strain amplitude and strain change rate are used to characterize the magnitude and rate of the cover plate's stress deformation.

4. The substation trench cover data collection and intelligent analysis management method according to claim 3 is characterized in that: The maximum strain amplitude, strain change rate, corner strain gradient, vibration direction angle, vibration decay time and deformation rebound time are combined into a multi-dimensional feature vector; The comprehensive damage index is used as the supervision label and the support vector regression algorithm is used to construct the cover damage assessment model.

5. The substation trench cover data collection and intelligent analysis management method according to claim 4 is characterized in that: The penalty coefficient and kernel function width of the support vector regression model were optimized through grid search combined with cross-validation. The root mean square error was used as the performance evaluation indicator, and the trained support vector regression model was used as the cover plate damage assessment model. A confidence evaluation mechanism is established. When the cover damage assessment model outputs the prediction result of the cover damage, the confidence of the cover damage prediction result is calculated according to the threshold of the distance between the real-time sample and the support vector in the training set.

6. The substation trench cover data collection and intelligent analysis management method according to claim 5 is characterized in that: The comprehensive damage index is stored in chronological order to form a damage time series. The long short-term memory network (LSTM) model is used to construct a damage trend prediction model. The input of the LSTM model is the historical damage time series, and the output is a sequence of cover damage trends in future time periods.

7. The substation trench cover data collection and intelligent analysis management method according to claim 6 is characterized in that: The damage acceleration factor is obtained by calculating the cover plate damage growth rate and change rate; When the change in damage growth rate exceeds the set threshold over multiple consecutive moments, it is determined to be in the acceleration stage; if the prediction error over multiple consecutive moments exceeds the preset threshold, the online incremental learning mechanism is triggered, and the long short-term memory network (LSTM) model parameters are updated using the latest monitoring data.

8. The method for collecting and intelligently analyzing substation trench cover data according to claim 7 is characterized in that: The current damage rate is obtained by linear fitting of the comprehensive damage index within the preset recent time window, and a damage development function is constructed to calculate the remaining maintenance time for the cover plate to reach the maintenance threshold. Correct the remaining maintenance time based on the damage acceleration factor and the introduction of a safety margin factor, and set multi-level warning thresholds; When the corrected remaining maintenance time is not greater than the corresponding warning threshold, hierarchical operation and maintenance warning information is generated, including the cover plate identification, current damage level, damage acceleration trend and recommended maintenance time window, and a time interval is set to suppress repeated warnings for the same cover plate.

9. A system for collecting and intelligently analyzing substation trench cover data, which is used to implement the method for collecting and intelligently analyzing substation trench cover data according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, feature extraction module, cover plate damage assessment module, cover plate damage trend prediction module and early warning module; The data acquisition module embeds an acceleration sensor and a resistance strain gauge in the cover plate to collect vibration data and strain data of the cover plate; The feature extraction module extracts the vibration direction change and vibration attenuation time of the acceleration sensor, calculates the deformation rebound time of the cover plate after the force is stopped, and extracts the characteristic values ​​of the strain data, including the maximum strain amplitude, strain change rate, and strain gradient in the stress concentration area, to characterize the stress deformation characteristics of the cover plate under stress; The cover plate damage assessment module constructs a cover plate damage assessment model based on stress deformation and deformation rebound time based on a neural network model according to the stress deformation and rebound characteristics of the cover plate, and assesses the degree of cover plate damage; The cover plate damage trend prediction module integrates the features of vibration data and strain data, predicts the damage degree change trend based on the machine learning model, and calculates the acceleration trend of cover plate damage; The warning module calculates the remaining maintenance time of the cover plate based on the currently assessed damage level of the cover plate and the acceleration trend of the cover plate damage, and triggers an operation and maintenance warning message when the remaining maintenance time is lower than a set threshold.

Images