An artificial intelligence-based high-voltage cable temperature monitoring system

Abstract

The application discloses an artificial intelligence-based high-voltage cable temperature monitoring system, relates to the technical field of high-voltage cable monitoring, and aims at solving the problems of low prediction accuracy, lack of multi-factor interaction analysis, incomplete fault evaluation and untimely early warning in the existing high-voltage cable temperature monitoring. The data acquisition module is used for collecting multi-source monitoring data of the high-voltage cable in real time, storing and sending; the data preprocessing and fusion module is used for receiving the multi-source monitoring data, preprocessing the multi-source monitoring data, and screening out an effective feature data set through correlation analysis; the intelligent temperature prediction module constructs a neural network prediction model based on the effective feature data set, realizes accurate prediction of the core temperature, surface temperature and joint temperature of the high-voltage cable, and outputs a temperature prediction result. The application has the advantages of high temperature prediction accuracy, strong real-time performance, multi-dimensional data fusion and interaction analysis, comprehensive and quantitative accurate fault risk evaluation, timely early warning, strong operation pertinence, strong system compatibility and high scalability.

Description

Technical Field

[0001] This invention relates to the field of high-voltage cable monitoring technology. Background Technology

[0002] High-voltage cables are the core infrastructure of power transmission networks, and their core temperature is a key indicator for judging the cable's operating status and preventing insulation aging and faults. However, due to the complex cable laying environment (underground, tunnels, cable trays, etc.) and the concealed nature of monitoring points, the core temperature of high-voltage cables is difficult to measure directly and accurately, and existing monitoring technologies still have many shortcomings.

[0003] 1. Existing temperature calculation methods, such as the thermal circuit method, have large errors in complex laying environments, while the finite element method, although highly accurate, takes a long time to calculate and cannot meet the needs of real-time monitoring.

[0004] 2. Existing monitoring systems mostly focus on the acquisition of a single temperature parameter, lacking analysis of the multi-dimensional interaction effects of the environment, soil, and cables, and ignoring the coupled influence of factors such as soil pH, salinity, ambient temperature and humidity, and cable load parameters on cable temperature changes and fault risks.

[0005] 3. Some intelligent prediction models are built using only a single algorithm, which has problems such as low optimization efficiency, easy getting trapped in local optima, and weak generalization ability. The temperature prediction accuracy is insufficient, and environmental corrosion factors are not included in the fault assessment system, resulting in delayed fault warning and high false alarm rate.

[0006] 4. The data acquisition and analysis processes are disconnected, lacking a standardized multi-source data fusion processing flow, making it impossible to achieve integrated closed-loop management from data acquisition and temperature prediction to fault assessment and early warning.

[0007] Therefore, existing high-voltage cable temperature monitoring systems suffer from low prediction accuracy, lack of multi-factor interactive analysis, incomplete fault assessment, and untimely early warning. Summary of the Invention

[0008] The purpose of this invention is to solve the problems of low prediction accuracy, lack of multi-factor interactive analysis, incomplete fault assessment, and untimely early warning in existing high-voltage cable temperature monitoring systems, and proposes a high-voltage cable temperature monitoring system based on artificial intelligence.

[0009] The high-voltage cable temperature monitoring system based on artificial intelligence described in this invention includes a data acquisition module, a data preprocessing and fusion module, and an intelligent temperature prediction module;

[0010] The data acquisition module is used to collect, store, and transmit multi-source monitoring data of the high-voltage cable in real time.

[0011] The data preprocessing and fusion module is used to receive multi-source monitoring data, preprocess the multi-source monitoring data, and screen out effective feature datasets through correlation analysis.

[0012] The intelligent temperature prediction module constructs a neural network prediction model based on an effective feature dataset to accurately predict the temperature of the high-voltage cable core, surface, and joints, and outputs the temperature prediction results.

[0013] Furthermore, it also includes a fault risk assessment module, an early warning push module, and a back-end management module;

[0014] The fault risk assessment module calculates the initial fault assessment coefficient and soil correction factor based on temperature prediction results and soil parameters to obtain the target fault assessment coefficient.

[0015] The early warning push module compares the target fault assessment coefficient with a preset threshold to generate and push early warning information.

[0016] The temperature prediction result input terminal of the background management module is connected to the temperature prediction result output terminal of the intelligent temperature prediction module; the preset threshold output terminal of the background management module is connected to the preset threshold input terminal of the early warning push module; and the early warning information receiving terminal of the background management module is connected to the early warning information sending terminal of the early warning push module.

[0017] Furthermore, the multi-source monitoring data includes cable operating parameters, environmental parameters, and soil parameters.

[0018] Furthermore, the data acquisition module includes a temperature sensor, a humidity sensor, a current transformer, a voltage transformer, a soil pH sensor, and a soil salinity sensor.

[0019] The cable operating parameters include cable surface temperature, cable joint temperature, cable load current, and cable load voltage; environmental parameters include laying environment temperature and environmental humidity; soil parameters include soil pH value and soil salinity.

[0020] Furthermore, the preprocessing operations of the data preprocessing and fusion module include data cleaning, outlier removal, missing value imputation, using Pearson correlation coefficient to measure the linear correlation between continuous variables, using Spearman correlation coefficient to measure the monotonic correlation between continuous and ordered variables, and dividing the effective feature dataset into training set and test set after removing redundant features.

[0021] Furthermore, the neural network prediction model constructed by the intelligent temperature prediction module is an improved BAS-GRNN neural network prediction model.

[0022] The BAS-GRNN improved neural network prediction model includes an input layer, a pattern layer, a summation layer, and an output layer.

[0023] The construction process of the BAS-GRNN improved neural network prediction model is as follows: the original GRNN model is smoothed using the whale optimization algorithm. To perform global optimization, a genetic algorithm is introduced to obtain the smoothness factor through selection, crossover, and mutation operations. The optimal solution is obtained; finally, the optimal smoothness factor is determined. Substituting the original GRNN model, we obtain the BAS-GRNN improved neural network prediction model.

[0024] Furthermore, the initial fault assessment coefficients in the fault risk assessment module (4) The calculation formula is:

[0025]

[0026] in, As the first weighting coefficient, It is the second weighting coefficient, and ; To predict the surface temperature of the cable; The optimal operating surface temperature for the cable; To predict cable joint temperature; The optimal operating joint temperature for the cable.

[0027] Furthermore, soil correction factors in the failure risk assessment module The calculation formula is:

[0028]

[0029] in, The third weighting coefficient, It is the fourth weighting coefficient, and ; Soil pH value; The soil pH value under optimal conditions for cable laying; Soil salinity.

[0030] Furthermore, the formula for calculating the target fault assessment coefficient of the fault risk assessment module is as follows:

[0031]

[0032] in, The target fault assessment coefficient.

[0033] This invention presents an artificial intelligence-based high-voltage cable temperature monitoring system. It integrates multi-source data acquisition technology, an improved artificial intelligence prediction algorithm, and a multi-factor fault assessment model to construct a high-precision, real-time, and intelligent high-voltage cable temperature monitoring system. This system enables accurate prediction of cable temperature and early warning of fault risks, improving the reliability and maintenance efficiency of high-voltage cable operation. Compared to existing technologies, it has the following significant advantages:

[0034] 1. High temperature prediction accuracy and strong real-time performance: The improved BAS-GRNN neural network model, jointly optimized by the whale algorithm and the genetic algorithm, solves the smoothing factor problem of the single GRNN model. The model avoids problems such as low optimization efficiency and easy getting trapped in local optima. It can accurately capture the complex nonlinear relationship between cable operating parameters, environmental parameters, soil parameters and cable temperature, greatly improve the prediction accuracy of cable core, surface and joint temperature, and has high model calculation efficiency to meet the needs of real-time monitoring.

[0035] 2. Achieve multi-dimensional data fusion and interactive analysis: Break through the limitations of single parameter monitoring, simultaneously collect multi-source data on cable operation, environment, and soil, and achieve feature screening and data fusion through correlation analysis. Fully consider the interactive coupling effect of environment, soil and cable, and better fit the actual operating conditions of high-voltage cables.

[0036] 3. Comprehensive and precise fault risk assessment: A two-layer fault risk assessment model is constructed. First, the initial fault assessment coefficient is calculated based on the temperature prediction results. Then, the soil correction factor is calculated by combining the corrosion effects of soil pH and soil salinity to obtain a comprehensive target fault assessment coefficient. The dual risks of temperature anomalies and soil corrosion are incorporated into the assessment system to achieve a precise quantitative assessment of fault risk and avoid the one-sidedness of assessment based on a single temperature index.

[0037] 4. Timely early warning and targeted operation and maintenance: A tiered early warning mechanism is set up, combined with multiple early warning push methods, to achieve real-time early warning of fault risks. The early warning information includes specific monitoring points, risk causes, and operation and maintenance suggestions, providing accurate decision-making basis for the refined operation and maintenance of high-voltage cables, effectively reducing the cable fault rate and improving the reliability of power transmission in the power system.

[0038] 5. Strong system compatibility and high scalability: The data acquisition module adopts a standardized SCADA system and configurable sampling parameters, supporting the access of various sensing devices; the parameters of the intelligent prediction module and fault assessment module can be flexibly configured through the backend to adapt to the monitoring needs of different laying environments (underground, tunnel, cable tray) and different specifications of high-voltage cables; the backend management module realizes full life cycle management of data, providing data support for subsequent big data analysis and life prediction of cable operation. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the principle structure of a high-voltage cable temperature monitoring system based on artificial intelligence, as described in Specific Implementation Method 1. Detailed Implementation

[0040] Specific Implementation Method 1: Combination Figure 1 This embodiment describes an artificial intelligence-based high-voltage cable temperature monitoring system, which includes a data acquisition module 1, a data preprocessing and fusion module 2, an intelligent temperature prediction module 3, a fault risk assessment module 4, an early warning push module 5, and a back-end management module 6. The modules are interconnected to form an integrated closed-loop system for high-voltage cable temperature monitoring.

[0041] Data acquisition module 1 is used to collect multi-source monitoring data of high-voltage cable operation in real time, including cable operation parameters, environmental parameters, and soil parameters. Through standardized sensing equipment and data transmission protocols, the collected data is uploaded to data preprocessing and fusion module 2 in real time. Among them, cable operation parameters include cable surface temperature, cable joint temperature, cable load current, and cable load voltage; environmental parameters include cable laying environment temperature and humidity; soil parameters (applicable to underground cables) include pH value and salinity of the soil where the cable is located.

[0042] Data acquisition module 1 deploys a SCADA system, equipped with temperature sensors, humidity sensors, current transformers, voltage transformers, soil pH sensors, and soil salinity sensors. It sets configurable sampling intervals and data transmission methods, and uses a PostgreSQL relational database to achieve real-time storage of the acquired data.

[0043] The data preprocessing and fusion module 2 is used to receive multi-source monitoring data uploaded by the data acquisition module 1, and perform preprocessing operations such as data cleaning, outlier removal, and missing value imputation in sequence. Then, feature screening and data fusion are achieved through correlation analysis to determine the effective feature dataset for temperature prediction.

[0044] Specifically, the Pearson correlation coefficient is used to measure the linear correlation between two continuous variables (such as load current and cable surface temperature, and ambient temperature and cable surface temperature), while the Spearman correlation coefficient is used to measure the monotonic correlation between continuous and ordered variables. Redundant features with low correlation are eliminated, and features highly correlated with cable core temperature among cable load current, load voltage, ambient temperature, ambient humidity, soil pH, soil salinity, cable surface temperature, and cable joint temperature are extracted as effective feature datasets and divided into training and testing sets.

[0045] The intelligent temperature prediction module 3, as the core module of the system, constructs an improved BAS-GRNN neural network prediction model based on an effective feature dataset to achieve accurate prediction of the core temperature of high-voltage cables and the surface and joint temperatures of cables at future times.

[0046] This module first constructs the original GRNN (Generalized Regression Neural Network) model using the training set. The GRNN model includes an input layer, a pattern layer, a summation layer, and an output layer. The input layer receives the effective feature dataset, the pattern layer realizes pattern memory based on the training samples, the summation layer completes the arithmetic sum and weighted sum calculation, and the output layer outputs the initial value of temperature prediction.

[0047] Then, the whale optimization algorithm is used to globally optimize the smoothness factor of the GRNN model, which solves the problem of insufficient fitting or generalization ability caused by unreasonable parameter settings of the single GRNN model. At the same time, the genetic algorithm is introduced to optimize the whale optimization algorithm. Through selection, crossover and mutation operations, the stagnation problem of the whale optimization algorithm when it is close to the optimal solution is solved, so as to achieve the global optimal solution of the smoothness factor.

[0048] Finally, the optimized smoothing factor is substituted into the GRNN model to construct the BAS-GRNN improved neural network prediction model. The model performance is verified using the test set. The verified model is then used for temperature prediction of real-time input data, outputting the real-time temperature of the high-voltage cable core, the predicted cable surface temperature and joint temperature at a preset future time.

[0049] The fault risk assessment module 4, based on the temperature prediction results of the intelligent temperature prediction module 3, combines soil parameters to construct a two-layer fault risk assessment model, calculates the target fault assessment coefficient of the high-voltage cable, and realizes a comprehensive quantitative assessment of fault risk.

[0050] The first step is to calculate the initial fault assessment coefficients;

[0051] The second step is to calculate the soil correction factor;

[0052] The third step is to calculate the target failure assessment coefficient: combine the initial failure assessment coefficient with the soil correction factor, with the target being the final quantitative indicator of failure risk.

[0053] The early warning push module 5 presets the early warning threshold of the target fault assessment coefficient and compares the target fault assessment coefficient calculated by the fault risk assessment module with the early warning threshold in real time. If the target fault assessment coefficient is greater than the early warning threshold, the system automatically generates graded early warning information, which includes cable monitoring points, temperature prediction values, soil parameters, fault risk level, and operation and maintenance suggestions.

[0054] Warning information is pushed to the backend management module and maintenance personnel terminals through various means, including backend pop-ups, SMS, APP push, email, etc., to achieve real-time warning of fault risks; if the target fault assessment coefficient is less than or equal to the warning threshold, the system continuously monitors and updates the assessment results.

[0055] The backend management module 6, as the core of the system's human-computer interaction and data management, realizes functions such as visual display of monitoring data, configuration of model parameters, setting of early warning thresholds, and management of operation and maintenance records;

[0056] This module provides a visual interface that displays multi-source data, temperature prediction results, fault assessment coefficients, and early warning status at each monitoring point of the cable in real time. It supports flexible configuration of parameters of the intelligent temperature prediction model (such as the population size of the whale optimization algorithm, the number of iterations of the genetic algorithm, and the smoothing factor range of the original GRNN model) and the weight coefficients and early warning thresholds of the fault assessment model. At the same time, it records all early warning information and operation and maintenance records to form a full life cycle data archive for high-voltage cable operation monitoring, and supports data query, export, and analysis.

[0057] (I) System Deployment:

[0058] The hardware deployment of each module in this embodiment of the high-voltage cable temperature monitoring system based on artificial intelligence is as follows:

[0059] Data Acquisition Module 1: At monitoring points along the 110kV high-voltage cable laying section, install temperature sensors (monitoring cable surface and joint temperatures, accuracy ±0.5℃), ambient temperature and humidity sensors (accuracy ±1℃, ±3%RH), current transformers (0.2 class), voltage transformers (0.2 class), soil pH sensors (accuracy ±0.1pH), and soil salinity sensors (accuracy ±0.01%); deploy a SCADA system, including a server and monitoring terminals, setting the sampling interval to 1 minute, and using 4G / 5G wireless transmission to upload the collected data to a cloud database (PostgreSQL);

[0060] Data preprocessing and fusion module 2, intelligent temperature prediction module 3, and fault risk assessment module 4 are deployed on cloud servers, using Python to build algorithm models, and based on the TensorFlow / PyTorch framework to implement the training and inference of neural network models.

[0061] Early warning push module 5: Connects to SMS platform, power operation and maintenance APP, and enterprise email system to realize multi-channel early warning push;

[0062] Backend management module: A visual management platform built with B / S architecture is supported by computer and mobile devices, enabling data display, parameter configuration, and record management functions.

[0063] (II) System Operation Steps:

[0064] Multi-source data acquisition: The sensors of the data acquisition module collect cable operating parameters (surface temperature, joint temperature, load current, load voltage), environmental parameters (ambient temperature, humidity), and soil parameters (pH value, salinity) at the monitoring points in real time, and store the data in the PostgreSQL cloud database in real time through the SCADA system;

[0065] Data Preprocessing and Fusion: The data preprocessing and fusion module reads multi-source data from the cloud database, first performing data cleaning (removing invalid and garbled data) and outlier removal (using 3D algorithms). The principles are as follows: data that deviates from the normal range is removed, and missing values ​​are filled (using linear interpolation to fill missing data); then, the correlation between each feature and cable core temperature is analyzed using Pearson correlation coefficient and Spearman correlation coefficient, redundant data with low correlation in ambient humidity are removed, and cable load current, load voltage, ambient temperature, cable surface temperature, cable joint temperature, soil pH value, and soil salinity are extracted as effective feature datasets, which are divided into training set and test set in a 7:3 ratio;

[0066] Intelligent temperature prediction:

[0067] (1) Construct the original GRNN model: The input layer receives 7-dimensional effective feature data, the number of nodes in the pattern layer is consistent with the number of training samples, the summation layer completes the arithmetic sum and weighted sum calculation, and the output layer initially outputs the predicted values ​​of cable core temperature, surface temperature and joint temperature.

[0068] (2) Smoothing factor Optimization: Initialize whale optimization algorithm parameters (population size SN=30, maximum number of iterations Tmax=100), and adjust the smoothing factor of the original GRNN model. Global optimization is performed; when the whale algorithm stagnates near the optimal solution, a genetic algorithm is introduced, using selection (roulette wheel), crossover (single-point crossover), and mutation (basic bit mutation) operations to achieve a smoothing factor. The optimal solution is obtained. =0.85;

[0069] (3) Model validation and prediction: The optimal smoothing factor Substitute the original GRNN model into the model to construct the BAS-GRNN improved model. Validate the model using a test set. The model's mean square error (MSE) is ≤0.3, meeting the accuracy requirements. Input the real-time collected effective feature data into the validated model to output the real-time temperature of the cable core, the predicted cable surface temperature and joint temperature for the next 3 hours.

[0070] Failure risk assessment:

[0071] (1) Setting parameters: Optimal operating surface temperature of the cable =40℃, optimal operating joint temperature =45℃, first weighting coefficient =0.6, second weighting coefficient =0.4; Optimal soil pH for cables =7.0, third weighting coefficient =0.5, fourth weighting coefficient =0.5;

[0072] (2) Calculate the initial fault assessment coefficient: If the model predicts the cable surface temperature in the next 3 hours =42℃, predicted joint temperature =48℃, then P=0.6×(42 / 40)+0.4×(48 / 45)=1.07;

[0073] (3) Calculate soil correction factor: If the collected soil pH value =5.5, salinity D=0.8%, then =0.5×∣5.5-7.0∣ / 7.0+0.5×0.8=0.607;

[0074] (4) Calculate the target fault assessment coefficient =1.07 + 1.07 × 0.607 ≈ 1.72;

[0075] Early warning push: The preset target fault assessment coefficient early warning threshold is 1.2. Based on the target calculated this time, the system automatically generates a level 1 early warning message, which reads: Level 1 early warning for 110kV high-voltage cable section. Monitoring point: predicted surface temperature 42℃, joint temperature 48℃, soil pH 5.5 (slightly acidic), salinity 0.8% in the next 3 hours. The fault risk level is high, and it is recommended to go to the site for inspection immediately. The early warning message is sent to maintenance personnel through background pop-up windows, maintenance APP push, SMS and email.

[0076] Backend Management and Operation: Operation and maintenance personnel can view early warning information and monitoring data through the backend management module 6, go to the site for inspection and maintenance, and enter the processing results into the backend management module 6 after the inspection and maintenance is completed to form an operation and maintenance record. The system continuously monitors the cable operation status at that point.

[0077] (III) System Performance Verification:

[0078] The system in this embodiment was tested on a 110kV high-voltage cable line for 3 months. Compared with a single temperature monitoring system, the average error of cable temperature prediction in this invention was reduced to within ±1℃, the accuracy of fault early warning was increased to over 95%, and the cable fault incidence rate was reduced by 60%, effectively realizing accurate monitoring of high-voltage cable temperature and early prevention of fault risks.

[0079] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A high-voltage cable temperature monitoring system based on artificial intelligence, characterized in that, It includes a data acquisition module (1), a data preprocessing and fusion module (2), and an intelligent temperature prediction module (3). The data acquisition module (1) is used to collect, store and transmit multi-source monitoring data of high-voltage cables in real time; The data preprocessing and fusion module (2) is used to receive multi-source monitoring data, preprocess the multi-source monitoring data, and screen out effective feature datasets through correlation analysis. The intelligent temperature prediction module (3) constructs a neural network prediction model based on an effective feature dataset to achieve accurate prediction of the temperature of the high-voltage cable core, surface and joint, and outputs the temperature prediction results.

2. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 1, characterized in that, It also includes a fault risk assessment module (4), an early warning push module (5), and a back-end management module (6); The fault risk assessment module (4) calculates the initial fault assessment coefficient and soil correction factor based on the temperature prediction results and soil parameters to obtain the target fault assessment coefficient. The early warning push module (5) compares the target fault assessment coefficient with the preset threshold, generates and pushes early warning information; The temperature prediction result input terminal of the background management module (6) is connected to the temperature prediction result output terminal of the intelligent temperature prediction module (3); the preset threshold output terminal of the background management module (6) is connected to the preset threshold input terminal of the early warning push module (5); the early warning information receiving terminal of the background management module (6) is connected to the early warning information sending terminal of the early warning push module (5).

3. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 1, characterized in that, The multi-source monitoring data includes cable operating parameters, environmental parameters, and soil parameters.

4. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 3, characterized in that, The data acquisition module (1) includes a temperature sensor, a humidity sensor, a current transformer, a voltage transformer, a soil pH sensor, and a soil salinity sensor. The cable operating parameters include cable surface temperature, cable joint temperature, cable load current, and cable load voltage; environmental parameters include laying environment temperature and environmental humidity; soil parameters include soil pH value and soil salinity.

5. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 1, characterized in that, The preprocessing operations of the data preprocessing and fusion module (2) include data cleaning, outlier removal, missing value imputation, Pearson correlation coefficient to measure the linear correlation between continuous variables, Spearman correlation coefficient to measure the monotonic correlation between continuous and ordered variables, and after removing redundant features, the effective feature dataset is divided into training set and test set.

6. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 1, characterized in that, The neural network prediction model constructed by the intelligent temperature prediction module (3) is the BAS-GRNN improved neural network prediction model; The BAS-GRNN improved neural network prediction model includes an input layer, a pattern layer, a summation layer, and an output layer. The construction process of the BAS-GRNN improved neural network prediction model is as follows: the original GRNN model is smoothed using the whale optimization algorithm. To perform global optimization, a genetic algorithm is introduced to obtain the smoothness factor through selection, crossover, and mutation operations. The optimal solution is obtained; finally, the optimal smoothness factor is determined. Substituting the original GRNN model into the model yields the improved BAS-GRNN neural network prediction model.

7. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 1, characterized in that, Initial fault assessment coefficients in the fault risk assessment module (4) The calculation formula is: in, As the first weighting coefficient, It is the second weighting coefficient, and ; To predict the surface temperature of the cable; The optimal operating surface temperature for the cable; To predict cable joint temperature; The optimal operating joint temperature for the cable.

8. The high-voltage cable temperature monitoring system based on artificial intelligence according to claim 7, characterized in that, Soil correction factors in the fault risk assessment module (4) The calculation formula is: in, This is the third weighting coefficient. It is the fourth weighting coefficient, and ; Soil pH value; The soil pH value under optimal conditions for cable laying; Soil salinity.

9. A high-voltage cable temperature monitoring system based on artificial intelligence as described in claim 8, characterized in that, The formula for calculating the target fault assessment coefficient in the fault risk assessment module (4) is as follows: in, The target fault assessment coefficient.

Images