A short circuit fault current prediction system based on data acquisition

Abstract

The application belongs to the technical field of power management, and specifically relates to a short-circuit fault current prediction system based on data collection, which comprises a data intelligent collection module, a data deep preprocessing module, a feature multi-dimensional extraction module, a fault mode intelligent identification module, a current dynamic prediction module and a result visualization display module; the original data is deeply preprocessed through the data deep preprocessing module, the preprocessed data is subjected to feature extraction through various feature extraction methods, the extracted features are classified and identified through a machine learning algorithm, the short-circuit fault current is dynamically predicted and visually displayed after it is determined that the power system is in a short-circuit fault mode, the accurate prediction of the short-circuit fault current of the power system is realized through the collaborative work of the modules, the timeliness of processing is improved, the power risk and automatic protection hidden danger of the power system can be reasonably analyzed and accurately warned, and the operation safety of the power system is ensured.

Description

Technical Field

[0001] This invention relates to the field of power management technology, specifically a short-circuit fault current prediction system based on data acquisition. Background Technology

[0002] In power systems, short-circuit faults are a common and serious type of fault. When a short-circuit fault occurs, it generates a huge short-circuit current, which can damage electrical equipment, even cause power outages, and seriously affect the normal operation of the power system.

[0003] Traditional short-circuit fault current prediction generally relies on a single data source or traditional machine learning model, which has problems such as response lag, single data dimension, and insufficient feature extraction capability. It is difficult to guarantee the accuracy of prediction results and improve the timeliness of processing. Furthermore, it cannot reasonably analyze and accurately warn of the power risks and automatic protection hazards of the power system, which is not conducive to ensuring the operational safety of the power system and has a low level of intelligence.

[0004] To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a short-circuit fault current prediction system based on data acquisition, which solves the problems of existing technologies such as slow response, single data dimension, and insufficient feature extraction capability. These problems make it difficult to guarantee the accuracy of prediction results and improve the timeliness of processing. Furthermore, they cannot reasonably analyze and accurately warn of the power risks and automatic protection hazards of the power system, which is not conducive to ensuring the operational safety of the power system and has a low level of intelligence.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A short-circuit fault current prediction system based on data acquisition includes a data intelligent acquisition module, a data deep preprocessing module, a feature multidimensional extraction module, a fault mode intelligent identification module, a current dynamic prediction module, and a result visualization module. The data intelligent acquisition module collects the operating data of the power system in real time by deploying various types of sensors at key nodes and equipment of the power system, and transmits the collected data to the data deep preprocessing module in the form of digital signals.

[0008] The data deep preprocessing module receives raw data from the data intelligent acquisition module and performs deep preprocessing on it, and then sends the deep preprocessed data to the feature multidimensional extraction module; the feature multidimensional extraction module uses a variety of feature extraction methods to extract features from the preprocessed data, and sends the extracted information to the fault mode intelligent identification module;

[0009] The fault mode intelligent identification module uses machine learning algorithms to classify and identify extracted features to determine whether the power system is in a short-circuit fault mode. When the fault mode intelligent identification module determines that the power system is in a short-circuit fault mode, the current dynamic prediction module uses a long short-term memory network to dynamically predict the short-circuit fault current based on historical data and currently collected data, and sends the short-circuit fault current prediction results to the result visualization module. The result visualization module displays the short-circuit fault current prediction results to the management personnel in an intuitive graphical and chart format.

[0010] Furthermore, the specific process of deep preprocessing is as follows:

[0011] First, the raw data is cleaned to remove noise, outliers, and missing values. Then, the raw data is normalized to unify data of different dimensions to the same scale range. Finally, data smoothing techniques are used to suppress random fluctuations in the data.

[0012] Furthermore, the feature extraction methods employed by the multidimensional feature extraction module include time-domain feature extraction, frequency-domain feature extraction, and time-frequency-domain feature extraction. Among them, time-domain feature extraction focuses on the statistical characteristics of the data, including mean, variance, and peak value; frequency-domain feature extraction uses the Fourier transform method to transform the data to the frequency domain and extract features including frequency components and power spectrum; time-frequency-domain feature extraction combines time-domain and frequency-domain analysis methods to comprehensively reflect the dynamic characteristics of the data.

[0013] Furthermore, the fault mode intelligent identification module adopts a hybrid model that combines convolutional neural networks and recurrent neural networks in deep learning. The convolutional neural network performs local feature extraction and spatial feature learning, while the recurrent neural network is used to process sequential data and capture the temporal relationship between features.

[0014] Furthermore, the results visualization module is connected to the power risk output module. The power risk output module is used to set the detection period, analyze the predicted generation status of the short-circuit fault current of the power system during the detection period, judge the power risk through analysis, generate a high-risk power signal or a low-risk power signal, and send the high-risk power signal or low-risk power signal to the results visualization module. When the results visualization module receives the high-risk power signal, it issues an early warning.

[0015] Furthermore, the specific analysis process for the power risk output module includes:

[0016] The predicted number of short-circuit fault currents generated during the detection period is obtained and marked as the predicted generation frequency value. The predicted generation frequency value is compared with the preset predicted generation frequency threshold. If the predicted generation frequency value exceeds the preset predicted generation frequency threshold, a high-risk power signal is generated.

[0017] If the predicted frequency value does not exceed the preset predicted frequency threshold, the power risk output coefficient is obtained through analysis. The power risk output coefficient is then compared with the preset power risk output coefficient threshold. If the power risk output coefficient exceeds the preset power risk output coefficient threshold, a high-risk power signal is generated; if the power risk output coefficient does not exceed the preset power risk output coefficient threshold, a low-risk power signal is generated.

[0018] Furthermore, the method for obtaining the power risk output coefficient is as follows:

[0019] All predicted short-circuit fault current values ​​within the detection period are obtained. These short-circuit fault current values ​​are compared with a preset short-circuit fault current threshold. If a short-circuit fault current value exceeds the preset threshold, the corresponding short-circuit fault current value is marked as a short-circuit fault current out-of-current value. The number of short-circuit fault current out-of-current values ​​within the detection period is obtained and marked as high-risk frequency detection values. The average of all short-circuit fault current values ​​within the detection period is calculated to obtain the current prediction characteristic value. The power risk output coefficient is obtained by weighted summation of the predicted frequency value, the high-risk frequency detection value, and the current prediction characteristic value.

[0020] Furthermore, the power risk output module is communicatively connected to the power protection decision module. The power risk output module sends a low-risk power signal to the power protection decision module. When the power protection decision module receives the low-risk power signal, it analyzes the degree of automatic protection risk for short-circuit fault current during the detection period. After analysis, it generates a protection safety signal and a protection warning signal, and sends the protection safety signal or protection warning signal to the result visualization display module.

[0021] Furthermore, the specific analysis process of the power protection decision module includes:

[0022] When a short-circuit fault current is predicted, the corresponding automatic protection device is activated to disconnect the corresponding fault circuit. The reaction time of the corresponding protection process is collected and marked as the startup hazard value. The startup characteristic value is obtained by averaging all startup hazard values. The number of times the startup hazard value exceeds the preset startup hazard threshold during the detection period is marked as the startup abnormal value.

[0023] The activation characteristic value and activation anomaly value are compared with the preset activation characteristic threshold and preset activation anomaly threshold respectively. If the activation characteristic value or activation anomaly value exceeds the corresponding preset threshold, a protection warning signal is generated. If neither the activation characteristic value nor the activation anomaly value exceeds the corresponding preset threshold, the system is analyzed to determine whether there is a hidden danger protection device in the power system. If there is a hidden danger protection device in the power system, a protection warning signal is generated. If there is no hidden danger protection device in the power system, a protection safety signal is generated.

[0024] Furthermore, the analysis and judgment methods for hazard protection equipment are as follows:

[0025] All automatic protection devices in the power system are acquired, and the corresponding automatic protection devices are marked as target objects i, where i is a natural number greater than 1. The total usage time of target object i and the number of times target object i performs opening and closing operations in historical periods are collected and marked as opening and closing detection values. The total usage time and opening and closing detection values ​​are compared with the corresponding preset total usage time threshold and preset opening and closing detection threshold respectively. If the total usage time or opening and closing detection value exceeds the corresponding preset threshold, target object i is marked as a hidden danger protection device.

[0026] If the total usage time and the opening and closing detection values ​​do not exceed the corresponding preset thresholds, the vibration amplitude of the target object i is collected and marked as the amplitude detection value, and the temperature, humidity and pollution level of the environment where the target object i is located are collected and marked as the temperature detection value, humidity detection value and pollution detection value. The state evaluation value is calculated by weighted summation of the amplitude detection value, temperature detection value, humidity detection value and pollution detection value. The state evaluation value is compared with the preset state evaluation threshold. If the state evaluation value exceeds the preset state evaluation threshold, the target object i is judged to be in an unsafe state.

[0027] The total duration of target object i in an unsafe state during a historical period is obtained and marked as an unsafe situation value. The number of times that the duration of a single instance of target object i in an unsafe state during a historical period exceeds the corresponding preset duration threshold is obtained and marked as an unsafe frequency value.

[0028] The hazard decision value is calculated by weighting and summing the total usage time, opening and closing detection values, unsafe time condition values ​​and unsafe frequency condition values. The hazard decision value is then compared with the corresponding preset hazard decision threshold. If the hazard decision value exceeds the corresponding preset hazard decision threshold, the target object i is generated and marked as a hazard protection device.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. In this invention, the original data is preprocessed in depth by a data deep preprocessing module, and features are extracted from the preprocessed data using multiple feature extraction methods. The extracted features are classified and identified using machine learning algorithms. When the power system is determined to be in a short-circuit fault mode, the short-circuit fault current is dynamically predicted and visualized. Based on the collaborative work between the modules, the accurate prediction of the short-circuit fault current of the power system is achieved and the timeliness of processing is improved.

[0031] 2. In this invention, the power risk output module analyzes the predicted generation status of short-circuit fault current in the power system to determine the power risk. When a high-risk power signal is generated, the cause is investigated and analyzed, and corresponding optimization and improvement measures are taken for the power system. When a low-risk power signal is generated, the power protection decision module analyzes the degree of automatic protection risks for short-circuit fault current during the detection period. When a protection warning signal is generated, the subsequent maintenance and supervision of automatic protection equipment is strengthened, and the corresponding automatic protection equipment is replaced in a timely manner, which significantly improves the operational safety of the power system and has a high level of intelligence. Attached Figure Description

[0032] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0033] Figure 1 This is a system block diagram of Embodiment 1 of the present invention;

[0034] Figure 2 This is a system block diagram of Embodiments 2 and 3 of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1: As Figure 1 As shown, the present invention proposes a short-circuit fault current prediction system based on data acquisition, which includes a data intelligent acquisition module, a data deep preprocessing module, a feature multidimensional extraction module, a fault mode intelligent identification module, a current dynamic prediction module, and a result visualization display module.

[0037] The intelligent data acquisition module deploys various types of sensors, such as current sensors, voltage sensors, and temperature sensors, on key nodes and equipment in the power system to collect real-time operational data. These sensors are characterized by high precision and high reliability, accurately measuring parameters such as current, voltage, and power of the power system. The collected data is then transmitted to the data preprocessing module in the form of digital signals. By acquiring real-time and accurate operational data of the power system, a rich data foundation is provided for subsequent fault prediction, which helps improve the accuracy of predictions. Furthermore, the combined use of multiple types of sensors can comprehensively reflect the operating status of the power system, avoiding prediction biases caused by a single data source.

[0038] The deep data preprocessing module receives raw data from the intelligent data acquisition module and performs deep preprocessing on it. The preprocessed data is then sent to the feature multidimensional extraction module. This effectively improves data quality and usability, reduces the impact of noise and outliers on subsequent analysis, and normalization and data smoothing techniques help improve model training efficiency and prediction accuracy. The specific process of deep preprocessing is as follows:

[0039] First, the raw data is cleaned to remove noise, outliers, and missing values. Then, the raw data is normalized to unify data of different dimensions to the same scale range for subsequent feature extraction and model training. In addition, data smoothing techniques are used to suppress random fluctuations in the data and make the data more stable.

[0040] The feature multidimensional extraction module uses various feature extraction methods to extract features from the preprocessed data, such as time domain feature extraction, frequency domain feature extraction, and time-frequency domain feature extraction. The extracted information is then sent to the fault mode intelligent identification module, which can extract more valuable information from the original data and provide richer feature basis for fault mode identification and current prediction. Furthermore, the comprehensive analysis of multidimensional features helps to improve the system's identification and prediction capabilities, enabling it to better cope with different types of short-circuit faults.

[0041] Among them, time-domain feature extraction mainly focuses on the statistical characteristics of data, such as mean, variance, peak value, etc.; frequency-domain feature extraction transforms data to the frequency domain through methods such as Fourier transform and extracts features including frequency components and power spectrum; time-frequency domain feature extraction combines time-domain and frequency-domain analysis methods, which can more comprehensively reflect the dynamic characteristics of data. By comprehensively applying these feature extraction methods, feature information of data can be extracted from multiple dimensions.

[0042] The fault mode intelligent identification module uses machine learning algorithms to classify and identify extracted features to determine whether the power system is in a short-circuit fault mode. It can quickly and accurately identify the short-circuit fault mode of the power system and provide key information for subsequent current prediction.

[0043] The fault mode intelligent identification module adopts a hybrid model that combines convolutional neural networks (CNN) and recurrent neural networks (RNN) in deep learning. CNN extracts local features and learns spatial features, while RNN is good at processing sequential data and capturing the temporal relationship between features. By combining the advantages of the two network models, fault modes in power systems can be accurately identified.

[0044] Once the fault mode intelligent identification module determines that the power system is in a short-circuit fault mode, the current dynamic prediction module starts working. Based on historical data and currently collected data, it uses a long short-term memory network (LSTM) to dynamically predict the short-circuit fault current and sends the short-circuit fault current prediction results to the result visualization module. It can predict the magnitude and trend of the short-circuit fault current in real time and accurately, providing an important basis for the protection and control of the power system.

[0045] It should be noted that LSTM is a special type of recurrent neural network that can effectively solve the gradient vanishing and gradient explosion problems that occur when traditional RNNs process long sequence data. It can better capture the dynamic trend of current changes. By learning from historical data and analyzing current data, the LSTM model can predict the short-circuit fault current value in the future.

[0046] The results visualization module presents the predicted short-circuit fault current results to management personnel in intuitive graphical and chart formats. This module employs visualization technologies such as line charts, bar charts, and pie charts to visually display the predicted short-circuit fault current values, trends, and related power system operating parameters. Simultaneously, the module provides alarm functions; for example, it promptly issues an alarm signal when the predicted short-circuit fault current exceeds a set threshold, alerting management personnel to take appropriate measures.

[0047] Example 2: Figure 2 As shown, the difference between this embodiment and Embodiment 1 is that the result visualization module is connected to the power risk output module. The power risk output module is used to set the detection period, preferably fifteen days. The predicted generation status of the short-circuit fault current of the power system during the detection period is analyzed, and the power risk is judged by the analysis, thereby generating a high-risk power signal or a low-risk power signal.

[0048] Furthermore, high-risk or low-risk power signals are sent to the results visualization module. Upon receiving a high-risk power signal, the results visualization module issues an early warning to remind management personnel to investigate and analyze the causes and take corresponding optimization and improvement measures for the power system, thereby reducing the subsequent operational risks of the power system. The specific analysis process of the power risk output module is as follows:

[0049] The predicted number of short-circuit fault currents generated during the detection period is obtained and marked as the predicted generation frequency value. The predicted generation frequency value is compared with the preset predicted generation frequency threshold. If the predicted generation frequency value exceeds the preset predicted generation frequency threshold, it indicates that the power system has a high power risk due to the short-circuit fault current, and a high power risk signal is generated.

[0050] If the predicted frequency value does not exceed the preset predicted frequency threshold, then all predicted short-circuit fault current values ​​within the detection period are obtained, and the short-circuit fault current values ​​are compared with the preset short-circuit fault current threshold. If the short-circuit fault current value exceeds the preset short-circuit fault current threshold, then the corresponding short-circuit fault current value is marked as a short-circuit fault current abnormal value. The number of short-circuit fault current abnormal values ​​within the detection period is obtained and marked as current high-risk frequency detection values. The average value of all short-circuit fault current values ​​within the detection period is calculated to obtain the current prediction characteristic value.

[0051] The power risk output coefficient is calculated by weighting and summing the predicted frequency generation value, the high-risk current detection frequency value, and the predicted current characteristic value. Specifically, each of the predicted frequency generation value, the high-risk current detection frequency value, and the predicted current characteristic value is assigned a corresponding preset weight coefficient, and then each of these values ​​is multiplied by its respective preset weight coefficient. The sum of these three products is then labeled as the power risk output coefficient. It should be noted that the larger the value of the power risk output coefficient, the higher the overall power risk of the power system due to short-circuit fault current.

[0052] The power risk output coefficient is compared with the preset power risk output coefficient threshold. If the power risk output coefficient exceeds the preset power risk output coefficient threshold, it indicates that the overall power risk of the power system due to short-circuit fault current is relatively high, and a high-risk power signal is generated. If the power risk output coefficient does not exceed the preset power risk output coefficient threshold, it indicates that the overall power risk of the power system due to short-circuit fault current is relatively low, and a low-risk power signal is generated.

[0053] Example 3: Figure 2As shown, the difference between this embodiment and Embodiment 1 and Embodiment 2 is that the power risk output module is communicatively connected to the power protection decision module. The power risk output module sends a low-risk power signal to the power protection decision module. When the power protection decision module receives the low-risk power signal, it analyzes the degree of automatic protection risk for short-circuit fault current during the detection period, and generates a protection safety signal and a protection warning signal through analysis.

[0054] Furthermore, the protection safety signal or protection early warning signal is sent to the result visualization module. When the result visualization module receives the protection early warning signal, it issues a corresponding warning to remind management personnel to investigate the cause and strengthen the subsequent maintenance and supervision of automatic protection equipment, and to replace the corresponding automatic protection equipment in a timely manner. This helps to improve the automatic protection performance of the power system and further ensure the subsequent operational safety of the power system. The specific analysis process of the power protection decision module is as follows:

[0055] When a short-circuit fault current is predicted, the corresponding automatic protection device is activated to disconnect the corresponding fault circuit. The reaction time of the corresponding protection process is collected and marked as the startup hazard value. The startup characteristic value is obtained by averaging all startup hazard values. The number of times the startup hazard value exceeds the preset startup hazard threshold during the detection period is marked as the startup abnormal value.

[0056] The starting characteristic value and the starting abnormal value are compared with the preset starting characteristic threshold and the preset starting abnormal threshold respectively. If the starting characteristic value or the starting abnormal value exceeds the corresponding preset threshold, it indicates that the timeliness of the power protection operation during the detection period is not good, the potential danger of power automatic protection is high, and it is necessary to investigate the cause in time and take corresponding improvement measures. Then, a protection warning signal is generated.

[0057] Furthermore, if neither the start characteristic value nor the start abnormal value exceeds the corresponding preset threshold, then all automatic protection devices of the power system (mainly referring to all circuit breakers involved in the power system) are acquired, and the corresponding automatic protection device is marked as target object i, where i is a natural number greater than 1; the total usage time of target object i is collected, as well as the number of times target object i performs opening and closing operations in the historical stage are collected and marked as opening and closing detection values;

[0058] The total usage time and the opening and closing detection value are compared with the corresponding preset total usage time threshold and preset opening and closing detection threshold respectively. If the total usage time or the opening and closing detection value exceeds the corresponding preset threshold, it indicates that the overall safety hazard of target object i is high. Then target object i is marked as a hazard protection device.

[0059] If the total usage time and the opening and closing detection values ​​do not exceed the corresponding preset thresholds, the vibration amplitude of the target object i is collected and marked as the amplitude detection value, and the temperature, humidity and pollution level (concentration of pollutants such as dust and salt spray) of the environment where the target object i is located are collected and marked as the temperature detection value, humidity detection value and pollution detection value.

[0060] The state assessment value is obtained by weighted summation of amplitude detection values, temperature detection values, humidity detection values, and pollution detection values. Specifically, each of these values ​​is assigned a corresponding preset weight coefficient, and then multiplied by its respective preset weight coefficient. The sum of these four products is then marked as the state assessment value. It should be noted that the larger the state assessment value, the worse the current state of the target object i, and the more easily its performance and lifespan are damaged.

[0061] The status evaluation value is compared with the preset status evaluation threshold. If the status evaluation value exceeds the preset status evaluation threshold, it indicates that the current status of target object i is poor and is likely to damage its performance and lifespan. Therefore, target object i is judged to be in an unsafe state.

[0062] The total duration of target object i in an unsafe state in a historical period is obtained and marked as an unsafe situation value. The duration of a single instance of target object i being in an unsafe state is compared with the corresponding preset duration threshold. Based on this, the number of times the duration of a single instance of target object i being in an unsafe state in a historical period exceeds the corresponding preset duration threshold is obtained and marked as an unsafe frequency value.

[0063] The hazard decision value is obtained by weighting and summing the total usage time, circuit breaker detection values, unsafe time conditions, and unsafe frequency conditions. Specifically, a corresponding preset weight coefficient is assigned to each of the four values, and each is multiplied by its respective preset weight coefficient. The sum of these four products is then marked as the hazard decision value. It should be noted that the larger the hazard decision value, the higher the overall safety hazard of the target object i.

[0064] The hazard decision value is compared with the corresponding preset hazard decision threshold. If the hazard decision value exceeds the corresponding preset hazard decision threshold, it indicates that the overall safety hazard of target object i is relatively high, and target object i is marked as a hazard protection device. If there is a hazard protection device in the power system, it indicates that the automatic power protection hazard during the detection period is relatively high, and a protection warning signal is generated. If there is no hazard protection device in the power system, it indicates that the automatic power protection hazard during the detection period is relatively low, and a protection safety signal is generated.

[0065] The working principle of this invention is as follows: In use, the intelligent data acquisition module collects real-time operating data of the power system; the deep data preprocessing module performs deep preprocessing on the raw data; the multi-dimensional feature extraction module uses various feature extraction methods to extract features from the preprocessed data; the intelligent fault mode identification module uses machine learning algorithms to classify and identify the extracted features; when it is determined that the power system is in a short-circuit fault mode, the dynamic current prediction module dynamically predicts the short-circuit fault current; and the result visualization module displays the short-circuit fault current prediction results to the management personnel in an intuitive graphical and chart-like format. Based on the collaborative work between the modules, accurate prediction of the short-circuit fault current of the power system is achieved, and the timeliness of processing is improved.

[0066] In this invention, the threshold, preset value, or preset range settings are for result comparison and analysis to determine whether the result is good or bad. The magnitude of these values ​​is determined by a combination of large-scale model analysis of sample data and human experience, and can also be appropriately adjusted based on seasonal or common-sense influence conditions. Similarly, the preset weight coefficients and influence factors are assigned specific values ​​based on the magnitude of each parameter's influence on the result, ultimately reflecting the impact on the result. These settings are also determined by a combination of large-scale model analysis of sample data and human experience, and can also be appropriately adjusted based on seasonal or common-sense influence conditions.

[0067] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, enabling those skilled in the art to better understand and utilize it. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A short circuit fault current prediction system based on data collection, characterized by, The system comprises a data intelligent acquisition module, a data deep preprocessing module, a feature multi-dimensional extraction module, a fault mode intelligent identification module, a current dynamic prediction module and a result visualization display module; the data intelligent acquisition module acquires the operation data of the power system in real time, and transmits the acquired data in the form of digital signals to the data deep preprocessing module; The data deep preprocessing module receives the original data from the data intelligent acquisition module and performs deep preprocessing thereon, and the feature multi-dimensional extraction module extracts features from the preprocessed data by using multiple feature extraction methods; The fault mode intelligent identification module classifies and identifies the extracted features by using a machine learning algorithm to determine whether the operation state of the power system is in a short-circuit fault mode; when the fault mode intelligent identification module determines that the power system is in a short-circuit fault mode, the current dynamic prediction module dynamically predicts the short-circuit fault current by using a long short-term memory network based on historical data and currently acquired data, and sends the short-circuit fault current prediction result to the result visualization display module; The result visualization display module is communicatively connected to a power risk output module, the power risk output module is configured to set a detection period, analyze the prediction generation condition of the short-circuit fault current of the power system in the detection period, determine the power risk by analysis, generate a power high-risk signal or a power low-risk signal according to the determination result, and send the power high-risk signal or the power low-risk signal to the result visualization display module; the result visualization display module issues a warning when the power high-risk signal is received; The specific analysis process of the power risk output module comprises: acquiring the number of times of prediction generation of the short-circuit fault current in the detection period and marking the number of times as a prediction frequency value, comparing the prediction frequency value with a preset prediction frequency threshold value, and generating the power high-risk signal if the prediction frequency value exceeds the preset prediction frequency threshold value; if the prediction frequency value does not exceed the preset prediction frequency threshold value, acquiring a power risk output coefficient by analysis, comparing the power risk output coefficient with a preset power risk output coefficient threshold value, generating the power high-risk signal if the power risk output coefficient exceeds the preset power risk output coefficient threshold value, and generating the power low-risk signal if the power risk output coefficient does not exceed the preset power risk output coefficient threshold value; The analysis and acquisition method of the power risk output coefficient is as follows: acquiring all the predicted short-circuit fault current values in the detection period, comparing the short-circuit fault current values with a preset short-circuit fault current threshold value, marking the corresponding short-circuit fault current value as a short-circuit fault current abnormal current value if the short-circuit fault current value exceeds the preset short-circuit fault current threshold value, acquiring the number of short-circuit fault current abnormal current values in the detection period and marking the number as a current high-risk detection frequency value, performing mean value calculation on all the short-circuit fault current values in the detection period to obtain a current prediction characteristic value, and calculating the power risk output coefficient by weighted summation of the prediction frequency value, the current high-risk detection frequency value and the current prediction characteristic value. The power risk output module is connected to the power protection decision module. The power risk output module sends a low-risk power signal to the power protection decision module. When the power protection decision module receives the low-risk power signal, it analyzes the degree of automatic protection risk for short-circuit fault current during the detection period. After analysis, it generates a protection safety signal and a protection warning signal, and sends the protection safety signal or protection warning signal to the result visualization display module. The specific analysis process of the power protection decision module includes: When a short-circuit fault current is predicted, the corresponding automatic protection device is activated to disconnect the corresponding fault circuit. The reaction time of the corresponding protection process is collected and marked as the startup hazard value. The startup characteristic value is obtained by averaging all startup hazard values. The number of times the startup hazard value exceeds the preset startup hazard threshold during the detection period is marked as the startup abnormal value. The activation characteristic value and activation anomaly value are compared with the preset activation characteristic threshold and preset activation anomaly threshold respectively. If the activation characteristic value or activation anomaly value exceeds the corresponding preset threshold, a protection warning signal is generated. If neither the activation characteristic value nor the activation anomaly value exceeds the corresponding preset threshold, the system is analyzed to determine whether there is a hidden danger protection device in the power system. If there is a hidden danger protection device in the power system, a protection warning signal is generated. If there is no hidden danger protection device in the power system, a protection safety signal is generated. The analysis and judgment methods for hidden danger protection equipment are as follows: All automatic protection devices in the power system are acquired, and the corresponding automatic protection devices are marked as target objects i, where i is a natural number greater than 1. The total usage time of target object i and the number of times target object i performs opening and closing operations in historical periods are collected and marked as opening and closing detection values. The total usage time and opening and closing detection values ​​are compared with the corresponding preset total usage time threshold and preset opening and closing detection threshold respectively. If the total usage time or opening and closing detection value exceeds the corresponding preset threshold, target object i is marked as a hidden danger protection device. If the total usage time and the opening and closing detection values ​​do not exceed the corresponding preset thresholds, the vibration amplitude of the target object i is collected and marked as the amplitude detection value, and the temperature, humidity and pollution level of the environment where the target object i is located are collected and marked as the temperature detection value, humidity detection value and pollution detection value. The state evaluation value is calculated by weighted summation of the amplitude detection value, temperature detection value, humidity detection value and pollution detection value. The state evaluation value is compared with the preset state evaluation threshold. If the state evaluation value exceeds the preset state evaluation threshold, the target object i is judged to be in an unsafe state. The total duration of target object i in an unsafe state during a historical period is obtained and marked as an unsafe situation value. The number of times that the duration of a single instance of target object i in an unsafe state during a historical period exceeds the corresponding preset duration threshold is obtained and marked as an unsafe frequency value. The hazard decision value is calculated by weighting and summing the total usage time, opening and closing detection values, unsafe time condition values ​​and unsafe frequency condition values. The hazard decision value is then compared with the corresponding preset hazard decision threshold. If the hazard decision value exceeds the corresponding preset hazard decision threshold, the target object i is generated and marked as a hazard protection device.

2. A short circuit fault current prediction system based on data collection as claimed in claim 1, wherein, The specific process of depth preprocessing is as follows: First, the raw data is cleaned to remove noise, outliers, and missing values. Then, the raw data is normalized to unify data of different dimensions to the same scale range. Finally, data smoothing techniques are used to suppress random fluctuations in the data.

3. A short circuit fault current prediction system based on data collection as claimed in claim 1, wherein, The feature extraction module employs time-domain feature extraction, frequency-domain feature extraction, and time-frequency-domain feature extraction. Time-domain feature extraction focuses on the statistical characteristics of the data, including mean, variance, and peak value. Frequency-domain feature extraction uses Fourier transform to convert the data to the frequency domain and extracts features including frequency components and power spectrum. Time-frequency-domain feature extraction combines time-domain and frequency-domain analysis methods to comprehensively reflect the dynamic characteristics of the data.

4. A short circuit fault current prediction system based on data collection as claimed in claim 1, wherein, The fault mode intelligent identification module adopts a hybrid model that combines convolutional neural networks and recurrent neural networks in deep learning. The convolutional neural network performs local feature extraction and spatial feature learning, while the recurrent neural network is used to process sequential data and capture the temporal relationship between features.

Images