A protection system and method for a high-voltage switchgear

Abstract

The application discloses a protection system and method for a high-voltage power distribution cabinet, and belongs to the technical field of high-voltage power distribution cabinet protection. The protection system comprises a data acquisition module and a fault test module. The data acquisition module is used for collecting data such as temperature and dust in the high-voltage cabinet and the machine room. The cloud management platform is used for machine deep learning, optimization and update processing, and improvement of prediction accuracy through an RNN recurrent network model. The temperature model can provide super-optimal parameters, so that the temperature model can realize accurate dynamic parameter adjustment. The display unit, the mobile terminal and the computer terminal are used for remote dynamic display. The RNN recurrent network model is used for predicting whether unknown load and dust amount and the like will exceed the temperature predetermined threshold during the operation of the high-voltage cabinet. The mobile terminal and the computer terminal are used for realizing the intelligent remote control purpose, realizing rapid response, and handling the abnormal temperature fault of the high-voltage cabinet.

Description

Technical Field

[0001] This application relates to the field of high-voltage switchgear protection technology, and more specifically, to a protection system and method for high-voltage switchgear. Background Technology

[0002] High-voltage switchgear is typically installed in a computer room, where multiple switchgear can be installed simultaneously and connected to the power grid for control. By installing various fire detection sensors, fire extinguishing systems, cabinet temperature detection sensors, equipment current sensors, and monitoring equipment in the computer room, a monitoring and protection system is formed. By constructing a protection system and a visual monitoring room, the operating status of the high-voltage switchgear in the computer room can be monitored and processed in real time, thereby effectively protecting the high-voltage switchgear. The prior art publication CN212114483U discloses a low-voltage distribution cabinet with temperature protection. This device is equipped with a heat dissipation device to cool the distribution cabinet when the temperature is high, an alarm device to alert people when the temperature approaches the danger temperature, and a power failure protector to disconnect the main circuit of the distribution cabinet when the danger temperature is reached. Through this three-level protection system, the distribution cabinet can be effectively protected to avoid accidents such as fire caused by excessive temperature of the distribution cabinet. While the existing technical solutions mentioned above can achieve the relevant beneficial effects through the existing technical structure, they still have the following defects: Compared with the power distribution cabinet protection methods in the literature, which limit the critical temperature threshold of the cabinet for protection, i.e., provide emergency circuit breaking protection for critical faults, they cannot protect against or respond to faults that cause abnormal temperatures under normal operation of the power distribution cabinet. They cannot perform dynamic temperature analysis and predict unknown abnormal faults in the power distribution cabinet, such as increased environmental dust, overload, undervoltage, grounding faults, etc., which cause abnormal temperature phenomena in the cabinet and computer room. They cannot make rapid operation and maintenance responses, and they cannot achieve the purpose of intuitive data display in a dynamic model and digital form. In view of the above-mentioned related technologies, we propose a protection system and method for high-voltage switchgear. Summary of the Invention

[0003] 1. The technical problem to be solved; The purpose of this application is to provide a protection system and method for high-voltage switchgear, which solves the problems of the switchgear protection methods in the prior art, which limit the critical temperature threshold of the switchgear for protection, i.e., provide emergency circuit breaking protection for critical faults. These methods cannot protect against or respond to faults that cause temperature anomalies during normal operation of the switchgear, cannot perform dynamic temperature analysis and predict unknown abnormal faults such as increased environmental dust, overload, undervoltage, grounding faults, etc., which cause abnormal temperatures in the switchgear and equipment room, and cannot provide rapid operation and maintenance response. Furthermore, they cannot achieve intuitive data display in a dynamic, model-based, and digital form. This system collects data on temperature, dust, and other parameters inside the high-voltage switchgear and in the computer room. The cloud management platform uses RNN (Recurrent Neural Network) machine deep learning to optimize and update the data, improving prediction accuracy. It can provide optimal parameters for the temperature model, enabling precise dynamic parameter adjustments. The system can be remotely and dynamically displayed via display units, mobile terminals, and computers. The RNN predictive capabilities can predict whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. Furthermore, it enables intelligent remote control via mobile terminals and computers, achieving rapid response and handling of abnormal temperature faults in the high-voltage switchgear.

[0004] 2. Technical solution; This application provides a protection system for a high-voltage switchgear, comprising a data acquisition module and a fault testing module. The data acquisition module collects temperature and environmental data information of the high-voltage switchgear during operation. After preprocessing the collected data, the data acquisition module synchronously inputs it into a central processing unit. The central processing unit receives the data and uploads it to a cloud management platform in real time. The cloud management platform integrates and standardizes the data, establishes a classified dataset, and stores it in a cloud database. The cloud management platform is equipped with an RNN recurrent network model for key data feature extraction. The RNN recurrent network model uses machine learning to optimize parameters. The cloud management platform uses the optimal parameters to construct a temperature model. The temperature model is connected to an analysis and comparison module, which is connected to an execution module. The execution module is used to handle abnormal temperature faults in the high-voltage switchgear. The cloud management platform is also connected to a display unit and a communication unit via a network. The communication unit is remotely connected to a mobile terminal and a computer via a network.

[0005] By adopting the above technical solution, high-voltage switchgear is distributed and installed in the equipment room. Data such as temperature and dust levels inside the switchgear and in the equipment room are collected. By setting up a fault testing module, the amount of data collected can be increased and stored in a cloud database. This allows for the use of a big data cluster for processing, improving the model training level. A cloud management platform is built to integrate and standardize the collected data, logically classify the data, and build different datasets. The cloud management platform uses an RNN recurrent network model to achieve machine deep learning, optimize and update the model, and gradually improve the prediction accuracy. It can provide optimal parameters for the temperature model, enabling precise dynamic parameter adjustment. The RNN recurrent network model can predict whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. The temperature model's three-dimensional dynamic temperature graph and dynamic data can be remotely and dynamically displayed through display units, mobile terminals, and computers. Furthermore, intelligent remote control can be achieved through mobile terminals and computers, enabling rapid response and maintenance of abnormal temperature faults in the high-voltage switchgear.

[0006] Optionally, the data acquisition module includes an internal temperature sensor, an external temperature sensor, a current sensor, a visual scanning device, and a dust monitoring sensor. The dust monitoring sensor is used to monitor the amount of dust entering the high-voltage cabinet. The visual scanning device is based on machine vision technology and uses software such as CAD to model and extract the surface information of the inside and outside of the high-voltage cabinet to construct a three-dimensional temperature model.

[0007] By adopting the above technical solution, the high-voltage switchgear internal temperature, equipment operating current, external temperature, and dust ingress amount can be dynamically monitored in real time. By using visual 3D scanning technology to perform 3D scanning of the high-voltage switchgear and the computer room, a 3D temperature model can be constructed. This model can intuitively display the operating status of the high-voltage switchgear and the heat dissipation efficiency of the computer room, and can effectively provide alarms and perform maintenance operations when the temperature exceeds the predetermined threshold.

[0008] Optionally, the central processing unit further includes a power module and a data processing module, wherein the data processing module is used for preprocessing the collected data, such as cleaning and segmentation.

[0009] By adopting the above technical solution, the collected data needs to undergo preprocessing operations, including image, dust quantity and temperature data segmentation, compression, cleaning and other preprocessing.

[0010] Optionally, the fault testing module includes a fault testing system, including overload testing, undervoltage testing, grounding fault testing, power supply testing, etc.

[0011] By adopting the above technical solutions, various fault simulation tests and fault diagnosis effects can be achieved through fault testing. It can provide a large amount of data information support from the cloud database, improve the training of RNN recurrent network models, and achieve the purpose of model optimization and parameter tuning.

[0012] Optionally, the execution module includes overload protection, a cooling fan, an air conditioning module, undervoltage protection, and an alarm module.

[0013] By adopting the above technical solution, when the predicted results are consistent with the actual monitoring results, the execution module works, including starting the cooling function of the air conditioning module in the computer room, protecting the high voltage cabinet from undervoltage and overload, and increasing ventilation and heat dissipation, thereby rapidly cooling down the high voltage cabinet. In addition, for some faults that cannot be remotely recovered, maintenance information can be prompted through mobile terminals and computers, improving the efficiency of maintenance response.

[0014] Optionally, the dataset is divided into a training set, a test set, and a validation set, with proportions of 70%, 20%, and 10%, respectively. The RNN recurrent network model is trained extensively on the training set. The RNN recurrent network model is equipped with a model prediction system to provide optimal parameter processing to the temperature model and to predict whether unknown loads, dust levels, and other data will affect the operation of the high-voltage switchgear beyond the predetermined temperature threshold.

[0015] By adopting the above technical solution, the RNN cyclic network model can provide optimal prediction capabilities through continuous optimization and parameter tuning, and provide super-optimal parameters to the temperature model, achieving the goal of accurate dynamic parameter adjustment of the temperature model. Through prediction accuracy, it can predict whether unknown loads, dust levels, and other data will affect the operation of the high-voltage switchgear and exceed the predetermined temperature threshold.

[0016] Optionally, the model prediction system includes the following processes: real-time event acquisition, data preprocessing, labeling feature variables and target variables, splitting the data into training and testing sets for RNN recurrent network model training and testing, machine learning and evaluating the average error value of the data, predicting output and providing optimal parameters to the temperature model when the average error value is zero, and dynamically adjusting the temperature model parameters.

[0017] By adopting the above technical solution, the accuracy of the prediction by the RNN recurrent network model is determined by comparing the average error value calculated by the model prediction system with the collected real-time data. If the data points match the actual data, the average error value can be directly and intelligently calculated using the formula, thereby determining the accuracy of the model prediction results.

[0018] Optionally, the formula for the average error value is: n; Where ME represents the average error, Xi This represents the i-th data point. This represents the average value of all data points, and n represents the number of data points.

[0019] By adopting the above technical solution, the deviation of each data point from the mean can be intelligently compared using the formula for the average error value, thereby achieving the purpose of machine deep learning and prediction.

[0020] Optionally, the average error value ranges from (0, infty). When the average error value is 0, it means that the deviation of each data point from the mean is zero, that is, the prediction is accurate. When the average error value is greater than 0, it means that there is a certain deviation between the data points and the mean, and the greater the dispersion of the data points, the greater the average error value.

[0021] By adopting the above technical solution, when the average error value is zero, it means that the model prediction is accurate; otherwise, it means that the model prediction is biased and needs to be returned to, and the data needs to be filtered and collected again, so as to make the dynamic information displayed by the temperature model more accurate.

[0022] Optional, the following steps may be included: S1. The high-voltage distribution cabinet is distributed and installed in the control room. The control room is equipped with a refrigeration and air conditioning module, a monitoring camera module and an indoor temperature sensor to collect indoor temperature data and image data in real time. The data is preprocessed and synchronously input into the central processor. S2. A cooling fan unit, a cabinet temperature sensor, a pressure sensor, a current sensor, and a dust monitoring sensor are installed inside the high-voltage distribution cabinet to collect data on the cabinet temperature, pressure, current, and dust entering the cabinet in real time. The data is preprocessed and synchronously input into the central processing unit. S3: The central processing unit processes the received data through edge computing and data compression, and uploads it to the cloud management platform via the communication network. The processing module integrates and standardizes the data, stores it in the cloud database, builds an RNN recurrent network model, extracts key features from the collected data, establishes a dataset, divides it into test set, training set and validation set, performs machine learning on the RNN recurrent network model, optimizes and tunes the parameters, calculates the average error value, and improves the prediction accuracy. S4. The model prediction system provides optimal parameters to build a three-dimensional temperature model of the cabinet and the room. Through intelligent comparison and verification analysis of the collected real-time data, it analyzes and predicts the heat generation data inside the high-voltage switchgear and the heat dissipation efficiency inside the room. It also predicts whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. The S5 cloud management platform uses a display unit to dynamically display the prediction results and temperature 3D model data. It also connects remotely to mobile terminals and computers via a communication module to achieve intelligent remote control, timely early warning and maintenance, improve the response rate of fault handling, and prevent the high-voltage cabinet temperature from being too high, which would affect the efficient operation of the equipment.

[0023] 3. Beneficial effects; One or more technical solutions provided in this application have at least the following technical effects or advantages: By collecting data such as temperature and dust inside the high-voltage switchgear and in the computer room, the cloud management platform optimizes and updates the processing through RNN cyclic network model machine deep learning, and improves the prediction accuracy. It can provide optimal parameters for the temperature model, enabling the temperature model to achieve precise dynamic parameter adjustment. It can also be remotely and dynamically displayed through display units, mobile terminals, and computers. The prediction capability of the RNN cyclic network model can predict whether the operation of the high-voltage switchgear will exceed the predetermined temperature threshold due to unknown loads, dust levels, and other data. Furthermore, it can achieve intelligent remote control through mobile terminals and computers, enabling rapid response and maintenance of abnormal temperature faults in the high-voltage switchgear. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the overall protection system structure of a protection system and method for a high-voltage switchgear disclosed in a preferred embodiment of this application; Figure 2 This is a schematic diagram of the model prediction system structure of a protection system and method for a high-voltage switchgear disclosed in a preferred embodiment of this application; Figure 3 This is a schematic diagram of the protection system operation method of a high-voltage distribution cabinet protection system and method disclosed in a preferred embodiment of this application. Detailed Implementation

[0025] The present application will be further described in detail below with reference to the accompanying drawings.

[0026] Reference Figure 1A protection system for a high-voltage switchgear includes a data acquisition module and a fault testing module. The data acquisition module collects temperature and environmental data during the operation of the high-voltage switchgear. After preprocessing the collected data, the module synchronously inputs it into a central processing unit (CPU). The CPU receives the data and uploads it to a cloud management platform in real time. The cloud management platform integrates and standardizes the data, establishes a categorized dataset, and stores it in a cloud database. The cloud management platform incorporates an RNN (Recurrent Neural Network) model for key data feature extraction. The RNN model is used for machine learning and parameter tuning. The cloud management platform constructs a temperature model using the optimal parameters. This temperature model is connected to an analysis and comparison module, which in turn is connected to an execution module. The execution module handles abnormal temperature faults in the high-voltage switchgear. The cloud management platform also connects to a display unit and a communication unit via a network. The communication unit remotely connects to mobile terminals and computers via the network, enabling the distributed operation of the high-voltage switchgear. Inside the installation room, data such as temperature and dust levels are collected from the high-voltage switchgear and the room itself. A fault testing module can be added to increase the amount of data collected. A cloud management platform integrates and standardizes the collected data, logically classifies it, and constructs different datasets. The cloud management platform utilizes an RNN (Recurrent Neural Network) model for machine deep learning, optimization, and updates, gradually improving prediction accuracy. It provides optimal parameters for the temperature model, enabling precise dynamic parameter adjustments. The RNN model's predictive capabilities allow for forecasting whether unknown loads and dust levels will cause the high-voltage switchgear to exceed predetermined temperature thresholds. The temperature model's 3D dynamic graph and data can be remotely displayed via display units, mobile terminals, and computers. Intelligent remote control via mobile terminals and computers enables rapid response and maintenance of abnormal temperature faults in the high-voltage switchgear.

[0027] Reference Figure 1 The data acquisition module includes an internal temperature sensor, an external temperature sensor, a current sensor, a visual scanning device, and a dust monitoring sensor. The dust monitoring sensor is used to monitor the amount of dust entering the high-voltage cabinet. The visual scanning device is based on machine vision technology and uses software such as CAD to model and extract the surface information inside and outside the high-voltage cabinet, and construct a three-dimensional temperature model. By dynamically monitoring the internal temperature of the high-voltage cabinet, the operating current of the equipment, the external temperature, and the amount of dust entering, and by performing three-dimensional scanning of the high-voltage cabinet and the computer room through visual three-dimensional scanning technology, it can realize the construction of a three-dimensional temperature model, which can intuitively display the operating status of the high-voltage cabinet and the heat dissipation efficiency of the computer room. It can effectively alarm and perform maintenance operations when the temperature exceeds the predetermined threshold.

[0028] Reference Figure 1The central processing unit also includes a power module and a data processing module. The data processing module is used for preprocessing the collected data, such as cleaning and segmentation. The collected data needs to be preprocessed in advance, including segmentation, compression, and cleaning of image, dust, and temperature data.

[0029] Reference Figure 1 The fault testing module includes a fault testing system, which includes overload testing, undervoltage testing, grounding fault testing, power supply testing, etc. Through fault testing, various fault simulation tests and fault diagnosis effects can be achieved. It can provide a large amount of data information support from the cloud database, improve the training of RNN recurrent network models, and achieve the purpose of model optimization and parameter tuning.

[0030] Reference Figure 1 The execution module includes overload protection, cooling fans, air conditioning modules, undervoltage protection, and alarm modules. When the predicted results match the actual monitoring results, the execution module works, including starting the cooling function of the air conditioning modules in the computer room, undervoltage and overload protection of the high-voltage cabinet, and increasing ventilation and heat dissipation, thereby rapidly cooling down the high-voltage cabinet. For some faults that cannot be remotely recovered, maintenance information can be prompted through mobile terminals and computers, improving maintenance response efficiency.

[0031] Reference Figure 1 The dataset is divided into training, testing, and validation sets, with proportions of 70%, 20%, and 10%, respectively. The RNN recurrent network model is trained extensively on the training set. The RNN recurrent network model is equipped with a model prediction system that provides optimal parameters to the temperature model and is used to predict whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. Through continuous optimization and parameter tuning, the RNN recurrent network model can provide optimal prediction capabilities and provide optimal parameters to the temperature model, achieving the goal of accurate dynamic parameter adjustment of the temperature model. Through prediction accuracy, it can predict whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold.

[0032] Reference Figure 1 and Figure 2 The model prediction system includes the following processes: real-time event acquisition, data preprocessing, labeling feature variables and target variables, splitting the data into training and testing sets for RNN recurrent network model training and testing, machine learning and evaluating the average error value of the data, predicting output and providing optimal parameters to the temperature model when the average error value is zero, and dynamically adjusting the temperature model parameters. The accuracy of the RNN recurrent network model prediction is determined by comparing the average error value calculated by the model prediction system with the acquired real-time data. If the data points match the actual data, the value can be directly and intelligently calculated using the average error value calculation formula, thereby determining the accuracy of the model prediction result.

[0033] Reference Figure 1 and Figure 2 The formula for the average error value is: n Where ME represents the mean error, This represents the i-th data point. The average value represents the value of all data points, and n represents the number of data points. Using the formula for the average error value, the deviation of each data point from the mean can be intelligently compared and processed, achieving the purpose of machine deep learning and prediction. The average error value ranges from (0, infty). When the average error value is 0, it means that the deviation of each data point from the mean is zero, i.e., the prediction is accurate. When the average error value is greater than 0, it means that there is a certain deviation between the data point and the mean, and the greater the dispersion of the data points, the greater the average error value. When the average error value is zero, it means that the model prediction is accurate; otherwise, it means that the model prediction is biased and needs to be returned, and data should be filtered and collected again to make the dynamic information displayed by the temperature model more accurate.

[0034] Reference Figure 3 This includes the following steps: S1. The high-voltage distribution cabinet is distributed and installed in the control room. The control room is equipped with a refrigeration and air conditioning module, a monitoring camera module and an indoor temperature sensor to collect indoor temperature data and image data in real time. The data is preprocessed and synchronously input into the central processor. S2. A cooling fan unit, a cabinet temperature sensor, a pressure sensor, a current sensor, and a dust monitoring sensor are installed inside the high-voltage distribution cabinet to collect data on the cabinet temperature, pressure, current, and dust entering the cabinet in real time. The data is preprocessed and synchronously input into the central processing unit. S3: The central processing unit processes the received data through edge computing and data compression, and uploads it to the cloud management platform via the communication network. The processing module integrates and standardizes the data, stores it in the cloud database, builds an RNN recurrent network model, extracts key features from the collected data, establishes a dataset, divides it into test set, training set and validation set, performs machine learning on the RNN recurrent network model, optimizes and tunes the parameters, calculates the average error value, and improves the prediction accuracy. S4. The model prediction system provides optimal parameters to build a three-dimensional temperature model of the cabinet and the room. Through intelligent comparison and verification analysis of the collected real-time data, it analyzes and predicts the heat generation data inside the high-voltage switchgear and the heat dissipation efficiency inside the room. It also predicts whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. The S5 cloud management platform uses a display unit to dynamically display the prediction results and temperature 3D model data. It also connects remotely to mobile terminals and computers via a communication module to achieve intelligent remote control, timely early warning and maintenance, improve the response rate of fault handling, and prevent the high-voltage cabinet temperature from being too high, which would affect the efficient operation of the equipment.

[0035] This application provides a protection system and method for a high-voltage switchgear. The working principle is as follows: The high-voltage switchgear is distributed and installed in a control room. The control room is equipped with a cooling air conditioning module, a monitoring camera module, and an indoor temperature sensor to collect indoor temperature and image data in real time. The data is preprocessed and synchronously input into the central processing unit (CPU). The high-voltage switchgear is also equipped with a cooling fan unit, an internal temperature sensor, a pressure sensor, a current sensor, and a dust monitoring sensor to collect real-time data on internal temperature, pressure, current, and dust levels. This data is preprocessed and synchronously input into the CPU. The CPU processes the received data through edge computing and data compression, and uploads it to a cloud management platform via a communication network. The processing module integrates and standardizes the data, stores it in a cloud database, and constructs an RNN (Recurrent Neural Network) model to process the collected data. Key feature extraction is performed to establish a dataset, which is then divided into test, training, and validation sets. An RNN recurrent network model is used for machine learning, with parameters optimized and tuned to calculate the average error value and improve prediction accuracy. The model prediction system provides optimal parameters to build a three-dimensional temperature model of the cabinet and room. Through intelligent comparison and verification analysis of the collected real-time data, it analyzes and predicts the heat generation data inside the high-voltage switchgear and the indoor heat dissipation efficiency. It also predicts whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. The cloud management platform dynamically displays the prediction results and the three-dimensional temperature model data using a display unit. It also connects remotely to mobile terminals and computers via a communication module to achieve intelligent remote control, timely warnings, and maintenance, improving the response rate for fault handling and preventing the high-voltage switchgear from overheating and affecting the efficient operation of the equipment.

Claims

1. A protection system for a high-voltage switchgear, characterized in that: The system includes a data acquisition module and a fault testing module. The data acquisition module collects temperature and environmental data for the high-voltage switchgear. After preprocessing the collected data, the module synchronously inputs it into a central processing unit (CPU). The CPU receives the data and uploads it to a cloud management platform in real time. The cloud management platform integrates and standardizes the data, establishes a categorized dataset, and stores it in a cloud database. The cloud management platform is equipped with an RNN (Recurrent Neural Network) model for key data feature extraction. The RNN model undergoes machine learning and parameter tuning. The cloud management platform uses the optimal parameters to construct a temperature model. The temperature model is connected to an analysis and comparison module, which is connected to an execution module. The execution module handles abnormal temperature faults in the high-voltage switchgear. The cloud management platform is also connected to a display unit and a communication unit via a network. The communication unit is remotely connected to a mobile terminal and a computer via the network.

2. The protection system for a high-voltage distribution cabinet according to claim 1, characterized in that: The data acquisition module includes an internal temperature sensor, an external temperature sensor, a current sensor, a visual scanning device, and a dust monitoring sensor. The dust monitoring sensor is used to monitor the amount of dust entering the high-voltage cabinet. The visual scanning device is based on machine vision technology and uses software such as CAD to model and extract the surface information of the inside and outside of the high-voltage cabinet to construct a three-dimensional temperature model.

3. The protection system for a high-voltage distribution cabinet according to claim 1, characterized in that: The central processing unit also includes a power module and a data processing module. The data processing module is used for preprocessing the collected data, such as cleaning and segmentation.

4. The protection system for a high-voltage distribution cabinet according to claim 1, characterized in that: The fault testing module includes a fault testing system, which includes overload testing, undervoltage testing, grounding fault testing, power supply testing, etc.

5. The protection system for a high-voltage distribution cabinet according to claim 1, characterized in that: The execution module includes overload protection, a cooling fan, an air conditioning module, undervoltage protection, and an alarm module.

6. The protection system for a high-voltage distribution cabinet according to claim 1, characterized in that: The dataset is divided into a training set, a test set, and a validation set, with proportions of 70%, 20%, and 10%, respectively. The RNN recurrent network model is trained extensively on the training set. The RNN recurrent network model is equipped with a model prediction system, which provides optimal parameter processing to the temperature model and is used to predict whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold.

7. The protection system for a high-voltage distribution cabinet according to claim 6, characterized in that: The model prediction system includes the following processes: real-time event acquisition, data preprocessing, labeling feature variables and target variables, splitting the data into training and testing sets for RNN recurrent network model training and testing, machine learning and evaluating the average error value of the data, predicting output and providing optimal parameters to the temperature model when the average error value is zero, and dynamically adjusting the temperature model parameters.

8. The protection system for a high-voltage distribution cabinet according to claim 7, characterized in that: Calculate the average error value ME.

9. The protection system for a high-voltage distribution cabinet according to claim 8, characterized in that: The average error value ranges from (0, infty). When the average error value is 0, it means that the deviation of each data point from the mean is zero, that is, the prediction is accurate. When the average error value is greater than 0, it means that there is a certain deviation between the data points and the mean, and the greater the dispersion of the data points, the greater the average error value.

10. A method of using the protection system for a high-voltage distribution cabinet according to any one of claims 1-9, characterized in that, Includes the following steps: S1. The high-voltage distribution cabinet is distributed and installed in the control room. The control room is equipped with a refrigeration and air conditioning module, a monitoring camera module and an indoor temperature sensor to collect indoor temperature data and image data in real time. The data is preprocessed and synchronously input into the central processor. S2. A cooling fan unit, a cabinet temperature sensor, a pressure sensor, a current sensor, and a dust monitoring sensor are installed inside the high-voltage distribution cabinet to collect data on the cabinet temperature, pressure, current, and dust entering the cabinet in real time. The data is preprocessed and synchronously input into the central processing unit. S3: The central processing unit processes the received data through edge computing and data compression, and uploads it to the cloud management platform via the communication network. The processing module integrates and standardizes the data, stores it in the cloud database, builds an RNN recurrent network model, extracts key features from the collected data, establishes a dataset, divides it into test set, training set and validation set, performs machine learning on the RNN recurrent network model, optimizes and tunes the parameters, calculates the average error value, and improves the prediction accuracy. S4. The model prediction system provides optimal parameters to build a three-dimensional temperature model of the cabinet and the room. Through intelligent comparison and verification analysis of the collected real-time data, it analyzes and predicts the heat generation data inside the high-voltage switchgear and the heat dissipation efficiency inside the room. It also predicts whether unknown loads, dust levels, and other data will cause the high-voltage switchgear to exceed the predetermined temperature threshold. The S5 cloud management platform uses a display unit to dynamically display the prediction results and temperature 3D model data. It also connects remotely to mobile terminals and computers via a communication module to achieve intelligent remote control, timely early warning and maintenance, improve the response rate of fault handling, and prevent the high-voltage cabinet temperature from being too high, which would affect the efficient operation of the equipment.

Images