Power equipment fault prediction method and system based on 5G communication

Abstract

This application provides a method and system for predicting power equipment faults based on 5G communication, relating to the field of fault prediction technology. The method includes: acquiring K transmission lines; obtaining first and second equipment monitoring data sets through asynchronous time-series extraction; traversing these data sets, performing fluctuation integration analysis, and determining the equipment fluctuation factor set; dynamically dividing power equipment into first and second-level dynamic monitoring equipment sets; generating first and second-level monitoring data sets through long and short time-series monitoring; and using a fault prediction network layer to perform fault prediction, obtaining K fault prediction results. This application addresses the shortcomings of existing technologies, which suffer from low efficiency and difficulty adapting to dynamic changes when processing large-scale power system data, making it difficult to reflect the operating status of equipment and predict potential faults in real time and accurately. It enables fault prediction of power system equipment, early identification of potential fault risks, reduction of fault incidence, and improvement of power system reliability and security.

Description

Technical Field

[0001] The present application relates to the field of fault prediction technology, and in particular to a method and system for predicting faults of power equipment based on 5G communication. Background Art

[0002] As the scale of power systems continues to expand, the number and types of power equipment have increased, and equipment failures have become increasingly prominent. Equipment failures may cause power systems to shut down, causing serious impacts on social production and people's lives.

[0003] At present, the existing technology relies on synchronous monitoring and fixed models to evaluate the status of power equipment and predict faults. It is inefficient when processing large-scale power system data and has difficulty adapting to dynamic changes. This makes it difficult to reflect the operating status of equipment and predict potential faults in real time and accurately, further affecting the stable operation of the power system and the timeliness of fault prevention. Summary of the invention

[0004] The purpose of this application is to provide a method and system for predicting power equipment faults based on 5G communication, so as to solve the problem that the prior art is inefficient in processing large-scale power system data and has difficulty adapting to dynamic changes, resulting in difficulty in real-time and accurately reflecting the operating status of equipment and predicting potential faults.

[0005] In view of the above problems, the present application provides a method and system for predicting power equipment faults based on 5G communication.

[0006] In the first aspect, the present application provides a method for predicting faults of electric power equipment based on 5G communication, which is implemented by a power equipment fault prediction system based on 5G communication, wherein the method for predicting faults of electric power equipment based on 5G communication comprises: obtaining K transmission lines in a target power area, wherein the K transmission lines include K sets of electric power equipment; performing asynchronous time series extraction on the equipment monitoring data of the K sets of electric power equipment within a preset monitoring window, respectively, to obtain K first equipment monitoring data sets and K second equipment monitoring data sets; traversing the K first equipment monitoring data sets and the K second equipment monitoring data sets; The method comprises the following steps: performing data fluctuation integration analysis on a set of monitoring data to determine K sets of equipment fluctuation factors; performing dynamic time series division of the K power equipment sets of the K transmission lines according to the K sets of equipment fluctuation factors to generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets; traversing the K first-level dynamic monitoring equipment sets and the K second-level dynamic monitoring equipment sets to perform long and short time series monitoring to generate K first-level monitoring data sets and K second-level monitoring data sets; performing fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets using a fault prediction network layer to obtain K fault prediction results.

[0007] In the second aspect, the present application also provides an electric power equipment fault prediction system based on 5G communication, which is used to execute the electric power equipment fault prediction method based on 5G communication as described in the first aspect, wherein the electric power equipment fault prediction system based on 5G communication includes: a transmission line acquisition module, used to acquire K transmission lines in the target power area, wherein the K transmission lines include K sets of electric power equipment; a monitoring data acquisition module, used to perform asynchronous time series extraction on the equipment monitoring data of the K sets of electric power equipment within a preset monitoring window, respectively, to obtain K first equipment monitoring data sets and K second equipment monitoring data sets; a fluctuation analysis module, used to traverse the K first equipment monitoring data sets and the K second equipment monitoring data sets. The invention relates to a method for performing data fluctuation integration analysis to determine K sets of equipment fluctuation factors; a timing division module is used to perform dynamic timing division of the K power equipment sets of the K transmission lines according to the K sets of equipment fluctuation factors to generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets; a long-short timing monitoring module is used to traverse the K first-level dynamic monitoring equipment sets and the K second-level dynamic monitoring equipment sets to perform long-short timing monitoring to generate K first-level monitoring data sets and K second-level monitoring data sets; a fault prediction result acquisition module is used to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets using a fault prediction network layer to obtain K fault prediction results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By acquiring K transmission lines in a target power area, wherein the K transmission lines include K power equipment sets; performing asynchronous time series extraction on the equipment monitoring data of the K power equipment sets within a preset monitoring window, respectively, to obtain K first equipment monitoring data sets and K second equipment monitoring data sets; performing data fluctuation integration analysis on the K first equipment monitoring data sets and the K second equipment monitoring data sets, to determine K equipment fluctuation factor sets; performing dynamic time series division of the transmission lines on the K power equipment sets of the K transmission lines according to the K equipment fluctuation factor sets, to generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets. The invention relates to a method for realizing a novel power system for controlling power system equipment and power management system. The method comprises the following steps: combining the K first-level dynamic monitoring equipment sets and the K second-level dynamic monitoring equipment sets to perform long and short time series monitoring, and generating K first-level monitoring data sets and K second-level monitoring data sets; using the fault prediction network layer to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets, and obtaining K fault prediction results. The method effectively solves the problem that the prior art is inefficient in processing large-scale power system data and is difficult to adapt to dynamic changes, resulting in difficulty in real-time and accurately reflecting the operating status of equipment and predicting potential faults, thereby realizing fault prediction of power system equipment, identifying potential fault risks in advance, reducing the occurrence rate of faults, and improving the reliability and safety of the power system.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 This is a flow chart of a method for predicting power equipment faults based on 5G communication in this application;

[0013] Figure 2 This is a structural diagram of the power equipment fault prediction system based on 5G communication in this application.

[0014] Description of reference numerals:

[0015] Transmission line acquisition module 11, monitoring data acquisition module 12, fluctuation analysis module 13, time series division module 14, long and short time series monitoring module 15, fault prediction result acquisition module 16. DETAILED DESCRIPTION

[0016] This application provides a method and system for predicting power equipment faults based on 5G communication, which solves the problem that the prior art is inefficient in processing large-scale power system data and has difficulty adapting to dynamic changes, making it difficult to reflect the operating status of equipment in real time and accurately predict potential faults. This method predicts faults of power system equipment, identifies potential fault risks in advance, reduces the occurrence of faults, and improves the reliability and safety of the power system.

[0017] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0018] For example, please refer to the attached Figure 1 The present application provides a method for predicting a fault of an electric power device based on 5G communication, wherein the method for predicting a fault of an electric power device based on 5G communication is applied to a system for predicting a fault of an electric power device based on 5G communication, and the method for predicting a fault of an electric power device based on 5G communication specifically comprises the following steps:

[0019] S1: Acquire K power transmission lines in a target power area, wherein the K power transmission lines include K power equipment sets.

[0020] Specifically, the target power area is a specific power supply area that needs to be monitored and analyzed. In the target power area, K key transmission lines are selected for monitoring, that is, the main lines in the power grid. The set of power equipment on each transmission line includes all important equipment on the line, such as transformers, circuit breakers, compensators, line towers, etc. The grid structure data of the target power area is extracted from the database of the power grid management system, and the power grid analysis software is used to select K key transmission lines according to preset rules such as load rate, line length, voltage level, etc. Determine the selection criteria for the line, for example, select the line with the highest load rate or the line with the most faults in history. For each selected transmission line, obtain the equipment list, including information such as the type, location, operating status, maintenance history, etc. of the equipment.

[0021] S2: Perform asynchronous time series extraction on the device monitoring data of the K power device sets within a preset monitoring window respectively to obtain K first device monitoring data sets and K second device monitoring data sets.

[0022] Specifically, the size and time period of the monitoring window are determined. It can be a fixed time period, such as every hour, every day or every week, or it can be event-driven, such as a period of time after the equipment is started. In each monitoring window, the operating data of the equipment, including current, voltage, temperature, vibration, sound, etc., are collected in real time through sensors and monitoring systems installed on the power equipment. Asynchronous time series data extraction is performed for each set of power equipment. The extracted data is divided into two groups, a first equipment monitoring data set and a second equipment monitoring data set. The first equipment monitoring data set represents the monitoring data of the equipment under normal or expected operating conditions. The second equipment monitoring data set represents the monitoring data of the equipment under abnormal or unexpected conditions.

[0023] S3: Perform data fluctuation integration analysis on the K first device monitoring data sets and the K second device monitoring data sets to determine K device fluctuation factor sets.

[0024] Specifically, the data is quality checked, erroneous or abnormal data points are removed, fluctuation analysis is performed on each device monitoring data set, the data change trend over time is analyzed, and patterns such as stable, rising, falling or periodic fluctuations are identified. The basic statistics of the data, such as mean, standard deviation, maximum value, minimum value, etc., are calculated to quantify the degree of fluctuation, and abnormal fluctuations in the data are detected using threshold method, rate of change method or machine learning method. The fluctuation analysis results of the first device monitoring data set and the second device monitoring data set are integrated, and the fluctuation differences under normal and abnormal conditions are compared to identify the fluctuation characteristics that significantly increase or decrease under abnormal conditions. Based on the results of the fluctuation analysis, the key factors affecting the stability of equipment operation, namely the equipment fluctuation factors, are determined. The equipment fluctuation factors include specific monitoring indicators, such as temperature, vibration, current, etc., or a combination thereof. For each equipment set, a corresponding fluctuation factor set is formed according to the determined equipment fluctuation factors.

[0025] S4: performing dynamic time series division of the K power equipment sets of the K transmission lines according to the K equipment fluctuation factor sets, and generating K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets.

[0026] Specifically, according to the analysis results of the fluctuation factor, the standard of dynamic time series division is formulated. Including the threshold of the fluctuation factor, the equipment exceeding the specific threshold will be divided into the first-level monitoring. The importance of the equipment, key equipment may be directly divided into the first-level monitoring. Fault history, equipment with frequent faults in the past is divided into the first-level monitoring. The first-level dynamic monitoring equipment set includes equipment that the fluctuation factor indicates high risk or is important to the operation of the power grid. The second-level dynamic monitoring equipment set includes equipment that the fluctuation factor indicates lower risk or is less important to the operation of the power grid. According to the division results, K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets are generated. Risk assessment is performed on the fluctuation factor of each device to determine the risk level of the device. According to the risk assessment results, the equipment is prioritized, and the equipment with high priority will be divided into the first-level monitoring, and regular or real-time adjustments will be made according to the real-time operating status of the equipment and the changes in the fluctuation factor.

[0027] S5: Traverse the K first-level dynamic monitoring device sets and the K second-level dynamic monitoring device sets to perform long and short time series monitoring, and generate K first-level monitoring data sets and K second-level monitoring data sets.

[0028] Specifically, according to the characteristics of the first-level and second-level dynamic monitoring equipment sets, corresponding monitoring strategies are formulated to determine the specific time windows and monitoring frequencies of long and short time series monitoring. Long and short time series monitoring is performed on each first-level dynamic monitoring equipment set to collect data within a longer time range for analyzing the long-term trends and periodic changes of the equipment. Data within a shorter time range is collected to capture the immediate status and rapid changes of the equipment. Long and short time series monitoring is performed on each second-level dynamic monitoring equipment set, and the monitoring frequency and detail are lower than the first-level monitoring. Data is collected during the monitoring process, and the data is classified and integrated according to the equipment set to generate K first-level monitoring data sets and K second-level monitoring data sets. Long time series monitoring sets a longer time window, such as a few days, weeks or months, collects the operating data of the equipment during this period, and analyzes its trends, seasonal patterns, etc. Short time series monitoring sets a shorter time window, such as a few minutes, hours or a day, and intensively collects the real-time data of the equipment during this period to detect sudden changes or anomalies. Use sensors, data acquisition systems, SCADA systems and other technologies and tools to collect data, and preprocess the collected data, including cleaning, normalization, and interpolation of missing values.

[0029] S6: Perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets using the fault prediction network layer to obtain K fault prediction results.

[0030] Specifically, the fault prediction network layer is a model based on machine learning or deep learning algorithms. Select appropriate algorithms, such as random forest, support vector machine, neural network, convolutional neural network, recurrent neural network or long short-term memory network. Perform further data preprocessing on the primary and secondary monitoring data sets to ensure data quality. Extract and select features that are helpful for fault prediction, including time domain features (such as mean, standard deviation), frequency domain features (such as spectrum analysis), time-frequency features, etc. Use historical fault data and corresponding monitoring data to train the fault prediction model, adjust the model parameters, and optimize the model performance through methods such as cross-validation. Perform fault prediction on K primary monitoring data sets and K secondary monitoring data sets respectively, and use the trained fault prediction model to predict the monitoring data to identify potential fault risks. Analyze the fault prediction results, determine the fault probability or risk level of each device or line, and generate K fault prediction results, each corresponding to a monitoring data set.

[0031] Furthermore, the present application also includes:

[0032] Traverse the K sets of power equipment to extract historical fault time nodes and obtain K historical fault time node clusters, each historical fault time node cluster includes multiple historical fault time node sets; traverse the K historical fault time node clusters to calculate the difference between adjacent historical fault time nodes and generate K historical fault time node difference sets; perform asynchronous analysis on the K historical fault time node difference sets to determine K first asynchronous extraction steps and K second asynchronous extraction steps; perform asynchronous time series extraction of equipment monitoring data on the K sets of power equipment within a preset monitoring window according to the K first asynchronous extraction steps and the K second asynchronous extraction steps to obtain the K first equipment monitoring data sets and the K second equipment monitoring data sets.

[0033] Specifically, traverse K sets of power equipment and extract the historical fault time nodes of each device from the historical fault records. For each device, form a historical fault time node cluster, which includes multiple historical fault time node sets; traverse K historical fault time node clusters, calculate the difference between adjacent historical fault time nodes in the cluster; generate K historical fault time node difference sets, each set contains the time difference between the historical fault time nodes of the corresponding device; perform asynchronous analysis on the K historical fault time node difference sets; determine K first asynchronous extraction steps and K second asynchronous extraction steps for subsequent asynchronous time series extraction of monitoring data. According to the K first asynchronous extraction steps and the K second asynchronous extraction steps, perform asynchronous time series extraction of device monitoring data on the K sets of power equipment within the preset monitoring window. Obtain K first device monitoring data sets and K second device monitoring data sets. Collect historical fault data from sources such as power system maintenance records and fault reports, and mark accurate time nodes for each fault event. Organize the historical fault time nodes of the same device together to form a historical fault time node cluster. Sort the time nodes in each historical fault time node cluster, calculate the time difference between adjacent time nodes, and organize the calculated differences into sets, each set corresponding to the historical fault time node difference of a device. Perform statistical analysis on the historical fault time node difference set, such as calculating the mean, median, standard deviation, etc. Based on the statistical analysis results, determine the first asynchronous extraction step and the second asynchronous extraction step, which reflects the statistical law of device failure. Determine the size and time period of the preset monitoring window, and extract the monitoring data of the device within the monitoring window from the monitoring system according to the determined asynchronous step. Organize the extracted data into the first device monitoring data set and the second device monitoring data set, respectively, for subsequent fault prediction analysis.

[0034] Furthermore, the present application also includes:

[0035] Respectively extract K maximum difference values ​​and K minimum difference values ​​from the K historical fault time node difference sets; take the K maximum difference values ​​and the K minimum difference values ​​as starting points, perform iterative edge identification in the K historical fault time node difference sets according to preset step sizes, and obtain K maximum difference neighborhoods and K minimum difference neighborhoods; perform mean calculation on the K maximum difference neighborhoods and the K minimum difference neighborhoods, and determine K first asynchronous extraction step sizes and the K second asynchronous extraction step sizes.

[0036] Specifically, for K historical fault time node difference sets, extract the maximum difference and the minimum difference in each set respectively, and obtain K maximum difference and K minimum difference. Taking the K maximum difference and the K minimum difference as the starting point, perform iterative edge identification in the K historical fault time node difference sets according to the preset step size. Calculate the mean of the K maximum difference neighborhoods and the K minimum difference neighborhoods respectively, and the calculated mean will be used to determine the asynchronous extraction step size. According to the calculated neighborhood mean, determine K first asynchronous extraction step sizes and K second asynchronous extraction step sizes. Traverse each historical fault time node difference set to find the maximum and minimum values ​​therein, and record the maximum and minimum values ​​of each set found to form two new sets. Use the extracted maximum and minimum difference values ​​as the starting point of iteration. Determine a preset step size, which can be a fixed value or based on a certain statistical standard, such as standard deviation. Starting from the starting point, iterate forward and backward in the difference set according to the preset step size to identify the neighborhood of the maximum and minimum values. Determine the range of the neighborhood. For example, you can select a certain number of points around the maximum and minimum values ​​as the neighborhood, average the differences in each neighborhood, and get the neighborhood mean. According to the neighborhood mean, determine the asynchronous extraction step size to guide the extraction of monitoring data to capture the law of fault occurrence.

[0037] Furthermore, the present application also includes:

[0038] Taking the K maximum difference values ​​as the starting point and the preset step size as the radius, K first neighborhoods are constructed, wherein each first neighborhood includes a plurality of historical fault time node differences whose distance from a historical fault time node difference set to a difference maximum value is within a preset step size range; calculating K first neighborhood densities of the K first neighborhoods; performing edge expansion on the K first neighborhoods according to the preset step size to obtain K second neighborhoods, wherein the K second neighborhoods include K second neighborhood densities; judging whether the difference between the K second neighborhood densities and the K first neighborhood densities is greater than or equal to a preset difference threshold, and if so, taking the K first neighborhoods as K maximum difference neighborhoods; taking the K minimum difference values ​​as the starting point, performing iterative edge identification in the K historical fault time node difference sets according to the preset step size to obtain the K minimum difference neighborhoods.

[0039] Specifically, with K maximum difference values ​​as the starting point and a preset step length as the radius, K first neighborhoods are constructed in the historical fault time node difference set, each first neighborhood includes a plurality of historical fault time node differences whose distance from a historical fault time node difference set to a maximum difference value is within the preset step length range, and the K first neighborhood densities of the K first neighborhoods are calculated. The first neighborhood density is the ratio of the number of historical fault time node differences in the first neighborhood to the area of ​​the first neighborhood, reflecting the degree of aggregation of the differences aggregated in the first neighborhood. The K first neighborhoods are edge expanded according to the preset step length to obtain K second neighborhoods. The second neighborhood includes K second neighborhood densities, reflecting the distribution of the differences in the expanded neighborhood. It is determined whether the difference between the K second neighborhood densities and the K first neighborhood densities is greater than or equal to the preset difference threshold. If the difference is greater than or equal to the preset threshold, the first neighborhood is considered to be valid and is used as the neighborhood of the maximum difference value. With the K minimum difference values ​​as the starting point, iterative edge identification is performed in the historical fault time node difference set according to the preset step length. Obtain K minimum difference neighborhoods, which reflect the fault mode near the minimum difference. Take each maximum difference as the starting point and the preset step length as the radius to determine the range of the neighborhood. Filter out the differences within the preset step length from the historical fault time node difference set, organize the filtered differences into neighborhoods to form K first neighborhoods, and perform density calculation on the differences in each first neighborhood. Select an expansion step length, and perform edge expansion on each first neighborhood according to the expansion step length. Perform density calculation on the expanded neighborhood to form the second neighborhood density. Set a preset difference threshold to determine the validity of the difference between the first neighborhood and the second neighborhood density. Compare whether the difference between the second neighborhood density and the first neighborhood density is greater than or equal to the preset threshold. If the difference is greater than or equal to the threshold, the first neighborhood is valid and is used as the maximum difference neighborhood.

[0040] Further, if not, iterative edge identification is performed on the K second neighborhoods according to the preset step size until the density difference between two adjacent iterations is greater than or equal to the preset difference threshold, thereby obtaining the K maximum difference neighborhoods.

[0041] Specifically, starting from K second neighborhoods, iterative edge identification is performed according to a preset step size. During the iteration, density calculation is performed on each second neighborhood to form a new density value. Compare the density difference between two adjacent iterations with the preset difference threshold. If the difference is less than the preset threshold, the iteration continues; if the difference is greater than or equal to the threshold, the iteration terminates. When the density difference between two adjacent iterations is greater than or equal to the preset difference threshold, the result of the current iteration is used as the K maximum difference neighborhood. A preset density difference threshold is set by a technician in this field to determine whether the iterative process is over. When the density difference between two adjacent iterations is greater than or equal to the threshold, the iteration ends, and the density value of the last iteration is used as the density value of the maximum difference neighborhood.

[0042] Furthermore, the present application also includes:

[0043] The fluctuation variances of the K first equipment monitoring data sets and the K second equipment monitoring data sets are calculated respectively to obtain K first fluctuation variance sets and K second fluctuation variance sets; the K first fluctuation variance sets and the K second fluctuation variance sets are weightedly calculated according to preset weight values ​​to obtain the K equipment fluctuation factor sets.

[0044] Specifically, the fluctuation variance is calculated for the K first equipment monitoring data sets and the K second equipment monitoring data sets respectively, which reflects the amplitude and range of data fluctuation and is an important indicator for measuring data volatility. The K first fluctuation variance sets and the K second fluctuation variance sets are weighted according to the preset weight values. The result of the weighted calculation forms K equipment fluctuation factor sets, each set containing the fluctuation factor of the corresponding equipment set. The K first equipment monitoring data sets and the K second equipment monitoring data sets are marked respectively, and the fluctuation variance is calculated for each data set. The fluctuation variance is the average of the square of the difference between the data point and the average value. According to the importance or failure risk of the equipment, a preset weight value is set, and the weight value can be fixed or dynamically adjusted. For each fluctuation variance set, a weighted calculation is performed according to the weight value, and the results of the weighted calculation are merged to form K equipment fluctuation factor sets. The fluctuation factor is extracted from each equipment fluctuation factor set, which reflects the status and risk of the equipment.

[0045] Furthermore, the present application also includes:

[0046] Construct the fault prediction network layer, wherein the fault prediction network layer includes a first feature recognition branch, a second feature recognition branch and a fault analysis fully connected layer; input the K first-level monitoring data sets into the first feature recognition branch to obtain K first-level feature recognition sets; input the K second-level monitoring data sets into the second feature recognition branch to obtain K second-level feature recognition sets; input the K first-level feature recognition sets and the K second-level feature recognition sets into the fault analysis fully connected layer respectively to obtain the K fault prediction results.

[0047] Specifically, determine the structure of the fault prediction network layer, including the first feature recognition branch, the second feature recognition branch and the fault analysis fully connected layer, select a suitable deep learning architecture, such as a convolutional neural network, a recurrent neural network or a long short-term memory network, construct a network layer, input K first-level monitoring data sets into the first feature recognition branch, and input K second-level monitoring data sets into the second feature recognition branch. The first feature recognition branch performs feature recognition on the K first-level monitoring data sets and extracts key features, and the second feature recognition branch performs feature recognition on the K second-level monitoring data sets and extracts key features. The K first-level feature recognition sets and the K second-level feature recognition sets are respectively input into the fault analysis fully connected layer. The fault analysis fully connected layer further analyzes and processes the input features, and finally obtains K fault prediction results, which reflect the fault risk or probability of each device set. Design the number of network layers, the number of neurons in each layer and the activation function. The first feature recognition branch and the second feature recognition branch are used to extract the features of the monitoring data. The fault analysis fully connected layer receives the features from the two feature recognition branches and preprocesses the first-level and second-level monitoring data sets, such as standardization and normalization. The preprocessed data is input into the network layer, the network layer is trained using the labeled fault data, the network parameters are adjusted, and the monitoring data is input into the trained network layer for fault prediction. The fault prediction result can be the fault probability, fault type or fault level.

[0048] In summary, the power equipment fault prediction method based on 5G communication provided by this application has the following technical effects:

[0049] By acquiring K transmission lines in a target power area, wherein the K transmission lines include K power equipment sets; performing asynchronous time series extraction on the equipment monitoring data of the K power equipment sets within a preset monitoring window, respectively, to obtain K first equipment monitoring data sets and K second equipment monitoring data sets; performing data fluctuation integration analysis on the K first equipment monitoring data sets and the K second equipment monitoring data sets, to determine K equipment fluctuation factor sets; performing dynamic time series division of the transmission lines on the K power equipment sets of the K transmission lines according to the K equipment fluctuation factor sets, to generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets. The invention relates to a method for realizing a novel power system for controlling power system equipment and power management system. The method comprises the following steps: combining the K first-level dynamic monitoring equipment sets and the K second-level dynamic monitoring equipment sets to perform long and short time series monitoring, and generating K first-level monitoring data sets and K second-level monitoring data sets; using the fault prediction network layer to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets, and obtaining K fault prediction results. The method effectively solves the problem that the prior art is inefficient in processing large-scale power system data and is difficult to adapt to dynamic changes, resulting in difficulty in real-time and accurately reflecting the operating status of equipment and predicting potential faults, thereby realizing fault prediction of power system equipment, identifying potential fault risks in advance, reducing the occurrence rate of faults, and improving the reliability and safety of the power system.

[0050] Embodiment 2: Based on the power equipment fault prediction method based on 5G communication in the previous embodiment, the present application also provides a power equipment fault prediction system based on 5G communication, please refer to the attached Figure 2 , the power equipment fault prediction system based on 5G communication includes:

[0051] The transmission line acquisition module 11 is used to acquire K transmission lines in a target power area, wherein the K transmission lines include K sets of power equipment.

[0052] The monitoring data acquisition module 12 is used to perform asynchronous time series extraction on the equipment monitoring data of the K power equipment sets within a preset monitoring window to obtain K first equipment monitoring data sets and K second equipment monitoring data sets.

[0053] The fluctuation analysis module 13 is used to perform data fluctuation integration analysis on the K first device monitoring data sets and the K second device monitoring data sets to determine K device fluctuation factor sets.

[0054] The timing division module 14 is used to perform dynamic timing division of the K power equipment sets of the K transmission lines according to the K equipment fluctuation factor sets, and generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets.

[0055] The long-short time series monitoring module 15 is used to traverse the K first-level dynamic monitoring device sets and the K second-level dynamic monitoring device sets to perform long-short time series monitoring and generate K first-level monitoring data sets and K second-level monitoring data sets.

[0056] The fault prediction result acquisition module 16 is used to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets using a fault prediction network layer to obtain K fault prediction results.

[0057] Furthermore, the 5G communication-based power equipment fault prediction system also includes an asynchronous timing extraction module, which is used to:

[0058] Traverse the K sets of power equipment to extract historical fault time nodes and obtain K historical fault time node clusters, each historical fault time node cluster includes multiple historical fault time node sets; traverse the K historical fault time node clusters to calculate the difference between adjacent historical fault time nodes and generate K historical fault time node difference sets; perform asynchronous analysis on the K historical fault time node difference sets to determine K first asynchronous extraction steps and K second asynchronous extraction steps; perform asynchronous time series extraction of equipment monitoring data on the K sets of power equipment within a preset monitoring window according to the K first asynchronous extraction steps and the K second asynchronous extraction steps to obtain the K first equipment monitoring data sets and the K second equipment monitoring data sets.

[0059] Furthermore, the 5G communication-based power equipment fault prediction system also includes a mean value calculation module, which is used to:

[0060] Respectively extract K maximum difference values ​​and K minimum difference values ​​from the K historical fault time node difference sets; take the K maximum difference values ​​and the K minimum difference values ​​as starting points, perform iterative edge identification in the K historical fault time node difference sets according to preset step sizes, and obtain K maximum difference neighborhoods and K minimum difference neighborhoods; perform mean calculation on the K maximum difference neighborhoods and the K minimum difference neighborhoods, and determine K first asynchronous extraction step sizes and the K second asynchronous extraction step sizes.

[0061] Furthermore, the 5G communication-based power equipment fault prediction system also includes a threshold judgment module, which is used to:

[0062] Taking the K maximum difference values ​​as the starting point and the preset step size as the radius, K first neighborhoods are constructed, wherein each first neighborhood includes a plurality of historical fault time node differences whose distance from a historical fault time node difference set to a difference maximum value is within a preset step size range; calculating K first neighborhood densities of the K first neighborhoods; performing edge expansion on the K first neighborhoods according to the preset step size to obtain K second neighborhoods, wherein the K second neighborhoods include K second neighborhood densities; judging whether the difference between the K second neighborhood densities and the K first neighborhood densities is greater than or equal to a preset difference threshold, and if so, taking the K first neighborhoods as K maximum difference neighborhoods; taking the K minimum difference values ​​as the starting point, performing iterative edge identification in the K historical fault time node difference sets according to the preset step size to obtain the K minimum difference neighborhoods.

[0063] Furthermore, the threshold judgment module in the power equipment fault prediction system based on 5G communication is also used for:

[0064] If not, iterative edge identification is performed on the K second neighborhoods according to the preset step size until the density difference between two adjacent iterations is greater than or equal to the preset difference threshold, thereby obtaining the K maximum difference neighborhoods.

[0065] Furthermore, the 5G communication-based power equipment fault prediction system also includes a weighted calculation module, which is used to:

[0066] The fluctuation variances of the K first equipment monitoring data sets and the K second equipment monitoring data sets are calculated respectively to obtain K first fluctuation variance sets and K second fluctuation variance sets; the K first fluctuation variance sets and the K second fluctuation variance sets are weightedly calculated according to preset weight values ​​to obtain the K equipment fluctuation factor sets.

[0067] Furthermore, the 5G communication-based power equipment fault prediction system also includes a fault analysis module, which is used to:

[0068] Construct the fault prediction network layer, wherein the fault prediction network layer includes a first feature recognition branch, a second feature recognition branch and a fault analysis fully connected layer; input the K first-level monitoring data sets into the first feature recognition branch to obtain K first-level feature recognition sets; input the K second-level monitoring data sets into the second feature recognition branch to obtain K second-level feature recognition sets; input the K first-level feature recognition sets and the K second-level feature recognition sets into the fault analysis fully connected layer respectively to obtain the K fault prediction results.

[0069] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1 The power equipment fault prediction method based on 5G communication and the specific examples in Example 1 are also applicable to the power equipment fault prediction system based on 5G communication in this embodiment. Through the above detailed description of the power equipment fault prediction method based on 5G communication, those skilled in the art can clearly know the power equipment fault prediction system based on 5G communication in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0070] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0071] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A method for predicting power equipment faults based on 5G communication, characterized in that: include: Acquire K power transmission lines in a target power area, wherein the K power transmission lines include K power equipment sets; Performing asynchronous time series extraction on the device monitoring data of the K power device sets within a preset monitoring window respectively, to obtain K first device monitoring data sets and K second device monitoring data sets; Performing data fluctuation integration analysis on the K first device monitoring data sets and the K second device monitoring data sets to determine K device fluctuation factor sets; According to the K equipment fluctuation factor sets, the K power equipment sets of the K transmission lines are dynamically divided into transmission line time series to generate K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets; Traversing the K first-level dynamic monitoring device sets and the K second-level dynamic monitoring device sets to perform long and short time series monitoring, and generating K first-level monitoring data sets and K second-level monitoring data sets; Using the fault prediction network layer to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets to obtain K fault prediction results; Traversing the K power equipment sets to extract historical fault time nodes, and obtaining K historical fault time node clusters, each historical fault time node cluster including multiple historical fault time node sets; Traversing the K historical fault time node clusters to calculate the difference between adjacent historical fault time nodes, and generating K historical fault time node difference sets; Performing asynchronous analysis on the K historical fault time node difference sets to determine K first asynchronous extraction step lengths and K second asynchronous extraction step lengths; Performing asynchronous time series extraction of device monitoring data on the K power device sets within a preset monitoring window according to the K first asynchronous extraction step sizes and the K second asynchronous extraction step sizes to obtain the K first device monitoring data sets and the K second device monitoring data sets; Respectively extract K maximum difference values ​​and K minimum difference values ​​from the K historical fault time node difference value sets; Taking the K maximum difference values ​​and the K minimum difference values ​​as starting points, iterative edge identification is performed in the K historical fault time node difference sets according to a preset step size to obtain K maximum difference neighborhoods and K minimum difference neighborhoods; Calculate the mean of the K maximum difference neighborhoods and the K minimum difference neighborhoods respectively, and determine the K first asynchronous extraction step lengths and the K second asynchronous extraction step lengths; Taking the K maximum difference values ​​as the starting point and the preset step length as the radius, K first neighborhoods are constructed, wherein each first neighborhood includes a plurality of historical fault time node differences whose distance from a historical fault time node difference set to a maximum difference value is within a preset step length range; Calculating K first neighborhood densities of the K first neighborhoods; Expanding the edges of the K first neighborhoods according to the preset step size to obtain K second neighborhoods, wherein the K second neighborhoods include K second neighborhood densities; Determine whether the difference between the K second neighborhood densities and the K first neighborhood densities is greater than or equal to a preset difference threshold, and if so, use the K first neighborhoods as K maximum difference neighborhoods; Taking the K minimum difference values ​​as the starting point, performing iterative edge identification in the K historical fault time node difference sets according to the preset step size to obtain the K minimum difference neighborhoods; If not, iteratively identifying edges on the K second neighborhoods according to the preset step size until the density difference between two adjacent iterations is greater than or equal to the preset difference threshold, thereby obtaining the K maximum difference neighborhoods; Calculating the fluctuation variances of the K first device monitoring data sets and the K second device monitoring data sets respectively to obtain K first fluctuation variance sets and K second fluctuation variance sets; The K first fluctuation variance sets and the K second fluctuation variance sets are weightedly calculated according to preset weight values ​​to obtain the K equipment fluctuation factor sets.

2. The method for predicting power equipment faults based on 5G communication according to claim 1, characterized in that: include: Constructing the fault prediction network layer, wherein the fault prediction network layer includes a first feature recognition branch, a second feature recognition branch and a fault analysis fully connected layer; Inputting the K primary monitoring data sets into the first feature recognition branch to obtain K primary feature recognition sets; Inputting the K secondary monitoring data sets into the second feature recognition branch to obtain K secondary feature recognition sets; The K primary feature recognition sets and the K secondary feature recognition sets are respectively input into the fault analysis fully connected layer to obtain the K fault prediction results.

3. The power equipment fault prediction system based on 5G communication is characterized by: The steps for implementing the method for predicting power equipment faults based on 5G communication according to any one of claims 1 to 2, wherein the power equipment fault prediction system based on 5G communication comprises: A transmission line acquisition module, used to acquire K transmission lines in a target power area, wherein the K transmission lines include K sets of power equipment; A monitoring data acquisition module, used to perform asynchronous time series extraction on the device monitoring data of the K power device sets within a preset monitoring window, to obtain K first device monitoring data sets and K second device monitoring data sets; A fluctuation analysis module, used for traversing and performing data fluctuation integration analysis on the K first device monitoring data sets and the K second device monitoring data sets to determine K device fluctuation factor sets; A timing division module, used for performing dynamic timing division of the K power equipment sets of the K transmission lines according to the K equipment fluctuation factor sets, and generating K first-level dynamic monitoring equipment sets and K second-level dynamic monitoring equipment sets; A long-short time series monitoring module, used to traverse the K first-level dynamic monitoring device sets and the K second-level dynamic monitoring device sets to perform long-short time series monitoring, and generate K first-level monitoring data sets and K second-level monitoring data sets; The fault prediction result acquisition module is used to perform fault prediction on the K first-level monitoring data sets and the K second-level monitoring data sets using the fault prediction network layer to obtain K fault prediction results.

Images