Multi-level active power distribution network residual current early warning method and device

Abstract

The embodiment of the invention discloses a multi-level active power distribution network residual current early warning method and device, and the method comprises the steps: building a power distribution system simulation model, and determining the access point and position of a distributed photovoltaic system in a power distribution system; historical residual current data are acquired and preprocessed; after preprocessing, the change trend of the residual current is determined through a time sequence, and residual current modes of different grid levels are determined; combining the change trend of the residual current and the residual current mode screening correlation factor, constructing a prediction model, and performing baseline prediction on the residual current; setting a preset threshold value; and when the residual current exceeds a preset threshold value, early warning is triggered. By processing historical data and predicting the residual current, the residual current in the power distribution network can be effectively monitored and managed, an alarm is given when the residual current exceeds a preset threshold value, the safety, the reliability and the economical efficiency of the power distribution network can be effectively improved, and the method is particularly suitable for a multi-level active power distribution network containing a distributed photovoltaic system.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of power systems, and in particular to a method and device for early warning of residual current in a multi-level active distribution network. Background Art

[0002] In traditional distribution network systems, with the widespread access of distributed energy resources (such as photovoltaic systems), the network structure has become more complex and dynamic. This change has brought new challenges to the monitoring, protection and operation of distribution networks, especially in the management of residual current (also known as leakage current). Due to the uncertainty of the access points of distributed generation and its operation, traditional residual current protection methods are difficult to adapt.

[0003] Residual current usually refers to the current flowing in the non-working current path of the power system. It may be caused by insulation failure, equipment damage or improper wiring. High residual current will not only lead to waste of electric energy, but more importantly, it may cause electrical fires or the risk of electric shock to people. Therefore, monitoring and warning of residual current is crucial to ensure the safe operation of the distribution system. Summary of the invention

[0004] The embodiment of the present invention provides a multi-level active distribution network residual current early warning method and device, which can effectively monitor and manage the residual current in the distribution network by processing historical data and predicting the residual current, and issue an alarm reminder when the residual current exceeds a preset threshold, which can effectively improve the safety, reliability and economy of the distribution network, and is particularly suitable for a multi-level active distribution network containing a distributed photovoltaic system.

[0005] According to a first aspect of the present invention, a multi-level active power distribution network residual current early warning method is provided, comprising:

[0006] Establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the location of the distributed photovoltaic system in the distribution network;

[0007] Obtain historical residual current data of different grid levels in the grid structure of the distribution system, and pre-process the historical residual current data;

[0008] According to the pre-processed historical residual current data, the residual current change trend is determined through time series, and the residual current mode at different grid levels is determined;

[0009] Combine the residual current change trend and residual current mode to screen the correlation factors of residual current, build a residual current prediction model based on grey system theory, and make a baseline prediction of residual current according to the residual current change trend and residual current prediction model;

[0010] A residual current baseline prediction based on a residual current prediction model and a preset threshold value is set based on a distribution system simulation model;

[0011] When the prediction model predicts that the residual current exceeds the preset threshold, an early warning is triggered.

[0012] Optionally, the grid structure of the power distribution system includes a grid hierarchy and nodes and connection methods of the grid hierarchy; the grid hierarchy includes a main distribution layer, a secondary distribution layer and an end-user access layer.

[0013] Optionally, the historical residual current data includes residual current data of different grid levels when there is no photovoltaic access under different working conditions, residual current data of different grid levels when there is photovoltaic access under different working conditions, configuration parameters and operation data of distributed photovoltaic access points, and environmental data; wherein the environmental data includes ambient temperature and ambient humidity;

[0014] The correlation factors include at least one of the operating status of power grid equipment, environmental factors, load conditions, power grid parameters, power grid wiring methods and line conditions.

[0015] Optionally, historical residual current data of different grid levels in the grid structure of the power distribution system is obtained, and the historical residual current data is preprocessed, including:

[0016] Remove outliers and noise from historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization.

[0017] Optionally, remove outliers and noise from the historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization, including:

[0018] Machine learning algorithms are used to identify and remove outliers and noise, and principal component analysis is used for data dimensionality reduction and feature extraction.

[0019] Optionally, the residual current correlation factors are screened by combining the residual current change trend and the residual current mode, including:

[0020] The correlation factors are screened through grey correlation analysis or principal component analysis.

[0021] Optionally, a residual current prediction model is constructed based on grey system theory, including:

[0022] Establish a preliminary prediction model using grey system theory;

[0023] The parameters in the preliminary model were determined by the least squares method;

[0024] The preliminary model is tested and revised to determine the residual current prediction model.

[0025] Optionally, after the residual current prediction model is constructed based on the grey system theory, it also includes:

[0026] Regularly compare the residual current predicted by the residual current preset model with the actual monitored residual current, and adjust the parameters of the residual current prediction model according to the comparison results.

[0027] Optionally, the residual current prediction model adopts a Transformer model, a convolutional neural network model, or a long short-term memory network model.

[0028] According to a second aspect of the present invention, there is provided a multi-level active power distribution network residual current early warning device, comprising:

[0029] A simulation module is used to establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network;

[0030] A data collection and preprocessing module is used to obtain historical residual current data of different grid levels in the grid structure of the distribution system and preprocess the historical residual current data;

[0031] A residual current mode determination module is used to determine the residual current mode of different grid levels based on the pre-processed historical residual current data and the changing trend of the residual current through the time series;

[0032] A prediction module is used to screen the correlation factors of the residual current by combining the residual current variation trend and the residual current mode, to build a residual current prediction model based on the grey system theory, and to make a baseline prediction of the residual current according to the residual current variation trend and the residual current prediction model;

[0033] A preset threshold determination module, used to set a preset threshold based on a residual current baseline prediction of a residual current prediction model and a power distribution system simulation model;

[0034] The early warning module is used to trigger an early warning when the prediction model predicts that the residual current exceeds a preset threshold.

[0035] The multi-level active distribution network residual current early warning method provided by the embodiment of the present invention includes: establishing a distribution system simulation model based on the grid structure of the distribution system, and determining the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network, so as to ensure that the simulation model can accurately reflect the operating status of the actual distribution system; obtaining historical residual current data of different grid levels in the grid structure of the distribution system, and preprocessing the historical residual current data, so as to ensure the quality and accuracy of the data used for analysis; according to the preprocessed historical residual current data, and determining the changing trend of the residual current through time series, the correlation factor and the changing law analysis are helpful to understand the changing pattern of the residual current, and then optimize the operation and maintenance strategy of the distribution system, and determine the residual current pattern of different grid levels; combining the changing trend of the residual current and the residual current pattern to screen the correlation factor of the residual current, and constructing a residual current prediction model based on the grey system theory to predict the residual current. The combined use of the measurement model and the simulation model can provide decision support for the operation and maintenance personnel, help the operation and maintenance personnel to allocate resources and arrange maintenance work more effectively, improve the operation and maintenance efficiency, and combine historical data and prediction models to effectively monitor and predict the changes in residual current, timely discover abnormal conditions, and help prevent electrical fires and other related safety accidents, thereby improving the safety and reliability of the entire distribution system; and perform baseline prediction of the residual current according to the change trend of the residual current and the residual current prediction model; set a preset threshold based on the residual current baseline prediction of the residual current prediction model and the distribution system simulation model; when the prediction model predicts that the residual current exceeds the preset threshold, trigger an early warning, and can send out an early warning signal before the residual current exceeds the safe range, so that the operation and maintenance personnel can take measures in time to avoid potential safety risks. This intelligent early warning mechanism can reduce the frequency and intensity of manual inspections, reduce operation and maintenance costs, and ensure the continuous and stable operation of the system. The present invention can effectively monitor and manage the residual current in the distribution network by processing historical data and predicting the residual current, and send out an alarm reminder when the residual current exceeds the preset threshold, which can effectively improve the safety, reliability and economy of the distribution network, and is particularly suitable for multi-level active distribution networks containing distributed photovoltaic systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of a multi-level active distribution network residual current early warning method provided by an embodiment of the present invention;

[0037] Figure 2 It is a structural schematic diagram of a multi-level active distribution network residual current early warning device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0039] The embodiment of the present invention provides a multi-level active power distribution network residual current early warning method, Figure 1 is a flowchart of a multi-level active distribution network residual current early warning method provided by an embodiment of the present invention, with reference to Figure 1 ,The multi-level active distribution network residual current early warning method includes:

[0040] S110. Establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network.

[0041] Among them, the use of advanced geographic information system technology can realize accurate three-dimensional modeling of the distribution system grid structure in the grid level division and photovoltaic access point identification of the multi-level active distribution network residual current warning method, clarify the spatial layout of nodes and connection methods at different levels, and improve the identification accuracy of access points of distributed photovoltaic systems. This process can optimize system design, enhance visual management, improve operation and maintenance efficiency, support decision-making, promote resource management, improve safety, promote data sharing and collaboration, adapt to regulatory requirements and long-term planning support. The above advantages jointly improve the operating efficiency, safety and economy of the distribution system, and provide strong technical support for the sustainable development of the power system.

[0042] S120, obtaining historical residual current data of different grid levels in the grid structure of the power distribution system, and preprocessing the historical residual current data.

[0043] Among them, the Internet of Things sensor technology is used to monitor and collect historical residual current data and environmental variables of each level of the grid in real time, providing real-time, accurate and high-quality original data for the residual current early warning method of the multi-level active distribution network. It not only enhances the comprehensiveness and accuracy of the data, but also optimizes resource allocation, strengthens fault diagnosis capabilities, and improves the overall reliability of the early warning system, providing a solid data foundation for realizing intelligent energy management and system optimization.

[0044] S130, determining residual current patterns at different grid levels based on the pre-processed historical residual current data and determining the residual current variation trend through a time series.

[0045] S140. The correlation factors of the residual current are screened based on the residual current variation trend and the residual current mode, a residual current prediction model is constructed based on the grey system theory, and a baseline prediction of the residual current is performed based on the residual current variation trend and the residual current prediction model.

[0046] Among them, the grey system theory treats random variables as grey variables that vary within a certain range, and the random process as a grey process that varies within a certain range and a certain time zone. Grey system theory transforms irregular raw data into regular generated series through methods such as generating functions and grey differential equations, and then establishes a model for prediction and control. Baseline prediction refers to a long-term trend prediction of residual current at different grid levels in the distribution system based on the residual current prediction model. This prediction aims to provide a reference benchmark for comparison with the actual observed residual current data, so as to evaluate whether the operating status of the distribution system is abnormal.

[0047] Residual current modes include: Normal mode: the residual current is small and stable. Mutation mode: when a fault or abnormality occurs in the circuit, such as equipment damage, insulation aging, etc., the residual current will suddenly increase. Fluctuation mode: in some cases, the residual current may fluctuate periodically. Noise mode: affected by electromagnetic interference, internal noise of the equipment and other factors, the residual current may show certain noise characteristics.

[0048] Observe the change of residual current over time to determine whether it shows a gradual increase, decrease or fluctuation trend. You can plot a curve of residual current over time to intuitively understand its change pattern. Calculate statistical indicators such as the rate of change and variance of residual current to further quantify the degree of change of residual current, determine the trend and pattern of change, and use it as a basis to use the residual current prediction model to predict the residual current in the future. Based on the results of the prediction model and the residual current range during normal operation of the equipment, determine the baseline value. The baseline value can be a fixed value or a range.

[0049] S150, setting a preset threshold based on the residual current baseline prediction of the residual current prediction model and the power distribution system simulation model.

[0050] Among them, according to the determined baseline value, combined with the prediction of the future trend of residual current changes, the preset threshold is set. Generally speaking, the threshold can be set on the basis of the baseline value plus a reasonable deviation range. The distribution system simulation model is used to simulate the changes in residual current under different working conditions, and the response and safety of the system under different threshold settings are observed. According to the simulation results, the preset threshold is adjusted and optimized. If the simulation results show that the residual current is easy to exceed the threshold under certain working conditions, resulting in false alarms or equipment shutdown, then the threshold can be appropriately increased; if it is found that the threshold is set too high and the abnormal situation cannot be discovered in time, then the threshold can be lowered. When setting the preset threshold, it is necessary to fully consider the safety of the equipment and the actual operation requirements. The threshold cannot be set too high to avoid ignoring potential safety hazards; nor can it be set too low to cause frequent false alarms and equipment shutdowns, affecting normal production and life. At the same time, it is also necessary to consider the particularities of different equipment and different systems, and formulate personalized threshold setting plans according to actual conditions.

[0051] S160. When the prediction model predicts that the residual current exceeds a preset threshold, an early warning is triggered.

[0052] Specifically, based on the grid structure of the distribution system, a distribution system simulation model is built using professional simulation software to determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network, and the historical residual current data of different grid levels are obtained from the grid structure of the distribution system. The historical residual current data is preprocessed, including data cleaning, denoising, missing value processing, etc. According to the preprocessed historical residual current data, the correlation factors of the residual current are screened using methods such as correlation analysis, and the changing trend of the residual current is determined by methods such as time series analysis. According to the changing trend of the residual current and the related factors, the residual current mode of different grid levels is determined. Based on the gray system theory, combined with the changing trend of the residual current and the residual current mode, a residual current prediction model is constructed. According to the residual current prediction model, a baseline prediction of the residual current is performed. Based on the baseline prediction results of the distribution system simulation model and the residual current prediction model, a preset threshold is set. When the prediction model predicts that the residual current exceeds the preset threshold, an early warning mechanism is triggered to promptly notify relevant personnel to handle it.

[0053] The multi-level active distribution network residual current early warning method provided by the embodiment of the present invention includes: establishing a distribution system simulation model based on the grid structure of the distribution system, and determining the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network, so as to ensure that the simulation model can accurately reflect the operating status of the actual distribution system; obtaining historical residual current data of different grid levels in the grid structure of the distribution system, and preprocessing the historical residual current data, so as to ensure the quality and accuracy of the data used for analysis; according to the preprocessed historical residual current data, and determining the changing trend of the residual current through time series, the correlation factor and the changing law analysis are helpful to understand the changing pattern of the residual current, and then optimize the operation and maintenance strategy of the distribution system, and determine the residual current pattern of different grid levels; combining the changing trend of the residual current and the residual current pattern to screen the correlation factor of the residual current, and constructing a residual current prediction model based on the grey system theory to predict the residual current. The combined use of the measurement model and the simulation model can provide decision support for the operation and maintenance personnel, help the operation and maintenance personnel to allocate resources and arrange maintenance work more effectively, improve the operation and maintenance efficiency, and combine historical data and prediction models to effectively monitor and predict the changes in residual current, timely discover abnormal conditions, and help prevent electrical fires and other related safety accidents, thereby improving the safety and reliability of the entire distribution system; and perform baseline prediction of the residual current according to the change trend of the residual current and the residual current prediction model; set a preset threshold based on the residual current baseline prediction of the residual current prediction model and the distribution system simulation model; when the prediction model predicts that the residual current exceeds the preset threshold, trigger an early warning, and can send out an early warning signal before the residual current exceeds the safe range, so that the operation and maintenance personnel can take measures in time to avoid potential safety risks. This intelligent early warning mechanism can reduce the frequency and intensity of manual inspections, reduce operation and maintenance costs, and ensure the continuous and stable operation of the system. The present invention can effectively monitor and manage the residual current in the distribution network by processing historical data and predicting the residual current, and send out an alarm reminder when the residual current exceeds the preset threshold, which can effectively improve the safety, reliability and economy of the distribution network, and is particularly suitable for multi-level active distribution networks containing distributed photovoltaic systems.

[0054] Optionally, the grid structure of the power distribution system includes a grid level and nodes and connection methods of the grid level; the grid level includes a main distribution layer, a secondary distribution layer and an end-user access layer.

[0055] Among them, through the three-point division in the main distribution layer, secondary distribution layer and end-user access layer, the access point of the distributed photovoltaic system and the specific location of the distributed photovoltaic system in the distribution network can be determined more accurately, which is helpful for subsequent data processing and analysis, ensuring the accuracy and completeness of the data, so as to improve the efficiency and accuracy of system management, provide a clear framework for subsequent data collection and preprocessing, facilitate the identification and location of potential problem points, and provide a basis for system optimization.

[0056] Optionally, the historical residual current data includes residual current data of different grid levels when there is no photovoltaic access under different working conditions, residual current data of different grid levels when there is photovoltaic access under different working conditions, configuration parameters and operation data of distributed photovoltaic access points, and environmental data; wherein the environmental data includes ambient temperature and ambient humidity;

[0057] The correlation factors include at least one of the operating status of power grid equipment, environmental factors, load conditions, power grid parameters, power grid wiring methods and line conditions.

[0058] Specifically, IoT sensor technology is used to monitor the historical residual current data of each level of the grid, as well as environmental variables such as temperature and humidity, in real time to ensure the real-time and accuracy of the data and provide high-quality original data for subsequent analysis.

[0059] Optionally, historical residual current data of different grid levels in the grid structure of the power distribution system is obtained, and the historical residual current data is preprocessed, including:

[0060] Remove outliers and noise from historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization.

[0061] Among them, normalization or standardization is used to eliminate the impact of dimensions, increase the speed of model convergence, and improve model performance. By preprocessing the data, the errors and deviations of subsequent analysis are reduced. Normalization or standardization helps to improve the generalization ability of the model, ensure the quality and consistency of the data, and improve the reliability of data analysis.

[0062] Optionally, remove outliers and noise from the historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization, including:

[0063] Machine learning algorithms are used to identify and remove outliers and noise, and principal component analysis is used for data dimensionality reduction and feature extraction.

[0064] Among them, the introduction of machine learning algorithms and advanced data processing techniques such as principal component analysis into the data preprocessing process can automatically identify and eliminate outliers and noise, and perform data dimension reduction and feature extraction, thereby improving data quality and processing efficiency. This method can improve data quality, reduce manual intervention, improve processing efficiency, strengthen data standardization, reduce data dimensions, highlight key variables, improve model performance, enhance model generalization capabilities, support complex analysis and maintain system stability, and can ensure the accuracy, reliability and efficiency of the early warning system.

[0065] Optionally, the residual current correlation factors are screened by combining the residual current change trend and the residual current mode, including:

[0066] The correlation factors are screened through grey correlation analysis or principal component analysis.

[0067] Specifically, it is used to extract features that may affect the variation of residual current, identify the main factors affecting the residual current, and screen the associated factors, analyze the variation trend of the residual current, and identify the residual current patterns at different grid levels.

[0068] Optionally, a residual current prediction model is constructed based on grey system theory, including:

[0069] Establish a preliminary prediction model using grey system theory;

[0070] The parameters in the preliminary model were determined by the least squares method;

[0071] The preliminary model is tested and revised to determine the residual current prediction model.

[0072] Among them, the grey system theory is suitable for processing incomplete information and is suitable for complex systems such as distribution networks. Constructing a current prediction model based on grey theory Data collection and preprocessing: Collect historical data of residual current, including data on residual current values, ambient temperature, humidity and other related factors under different working conditions. Clean and preprocess the data to remove outliers and missing values ​​to ensure the accuracy and reliability of the data. Establish a grey system model: The residual current system is regarded as a grey system containing known and unknown information. According to the trend and pattern of residual current changes, the input (independent variable) and output (dependent variable) of the system are determined, that is, the factors affecting the residual current are used as input and the residual current value is used as output. The modeling method in the grey system theory, such as the GM (1,1) model, is used to establish a residual current prediction model. The GM (1,1) model is a commonly used grey prediction model. It generates and processes the original data once, weakens randomness and enhances regularity, and then establishes a differential equation model for prediction. Model solution and parameter estimation: Estimate the parameters in the model by methods such as the least squares method. Solve the model to obtain the predicted value of the residual current. Model verification and correction: The established grey prediction model is verified. Commonly used verification methods include residual error test, correlation test and posterior difference test. If the accuracy of the model does not meet the requirements, the model can be corrected and optimized by adding correction items, adjusting model parameters, etc. Result analysis and application: According to the predicted residual current value, analysis and application are carried out in combination with actual conditions. For example, it can be determined whether there is an abnormality in the residual current and take corresponding measures in advance; or the prediction results can be combined with other safety monitoring systems to improve the overall safety performance.

[0073] Optionally, after the residual current prediction model is constructed based on the grey system theory, it also includes:

[0074] Regularly compare the residual current predicted by the residual current preset model with the actual monitored residual current, and adjust the parameters of the residual current prediction model according to the comparison results.

[0075] Among them, the prediction results are evaluated and compared with the actual monitoring data, the accuracy of the prediction model is evaluated, and the model parameters are adjusted regularly to adapt to system changes. This not only improves the accuracy and reliability of the prediction model, but also enhances decision-making support, improves the stability of the system and resource utilization efficiency, while also supporting the long-term planning of the power system and improving user trust.

[0076] Optionally, the residual current prediction model adopts a Transformer model, a convolutional neural network model, or a long short-term memory network model.

[0077] Among them, the Transformer model has the advantages of strong ability to handle long-distance dependencies, strong parallel computing capabilities, high scalability and flexibility. It can more accurately predict the changing trend of residual current and improve the accuracy and timeliness of early warning. Due to its high computing efficiency, it can generate early warning results faster, providing real-time guarantee for the safe operation of the distribution network. The convolutional neural network model has the advantages of local perception and weight sharing, strong feature extraction capabilities, and sensitivity to image and sequence data. It can extract useful features from residual current data, improve the accuracy of prediction, and is sensitive to abnormal changes in the distribution network. It can detect and warn potential faults in time. The long short-term memory network model has the advantages of strong ability to process sequence data, sensitivity to time series data, and strong robustness. It can accurately predict the future trend of residual current and provide reliable guarantee for the safe operation of the distribution network. It is sensitive to abnormal changes in time series data and can detect and warn potential faults in time.

[0078] The embodiment of the present invention also provides a multi-level active power distribution network residual current early warning device, Figure 2 is a schematic diagram of a structure of a multi-level active distribution network residual current early warning device provided by an embodiment of the present invention, with reference to Figure 2 , the multi-level active distribution network residual current warning device includes:

[0079] The simulation module 210 is used to establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network;

[0080] The data collection and preprocessing module 220 is used to obtain the historical residual current data of different grid levels in the grid structure of the power distribution system and preprocess the historical residual current data;

[0081] The residual current mode determination module 230 is used to determine the residual current mode of different grid levels according to the pre-processed historical residual current data and the change trend of the residual current through the time series;

[0082] A prediction module 240 is used to screen the correlation factors of the residual current in combination with the residual current variation trend and the residual current mode, construct a residual current prediction model based on the grey system theory, and perform a baseline prediction of the residual current according to the residual current variation trend and the residual current prediction model;

[0083] A preset threshold determination module 250, for setting a preset threshold based on a residual current baseline prediction of a residual current prediction model and a power distribution system simulation model;

[0084] The early warning module 260 is used to trigger an early warning when the prediction model predicts that the residual current exceeds a preset threshold.

[0085] The technical solution of this embodiment is used to establish a distribution system simulation model based on the grid structure of the distribution system through a simulation module, and determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network; through a data collection and preprocessing module, it is used to obtain historical residual current data of different grid levels in the grid structure of the distribution system, and preprocess the historical residual current data; through a residual current mode determination module, it is used to determine the residual current mode of different grid levels based on the preprocessed historical residual current data and the changing trend of the residual current through a time series; through a prediction module, it is used to screen the correlation factors of the residual current in combination with the changing trend of the residual current and the residual current mode, build a residual current prediction model based on the grey system theory, and perform a baseline prediction of the residual current according to the changing trend of the residual current and the residual current prediction model; through a preset threshold determination module, it is used to set a preset threshold based on the residual current baseline prediction of the residual current prediction model and the distribution system simulation model; through an early warning module, it is used to trigger an early warning when the prediction model predicts that the residual current exceeds the preset threshold. By processing historical data and predicting residual current, the residual current in the distribution network can be effectively monitored and managed, and an alarm will be issued when the residual current exceeds the preset threshold, which can effectively improve the safety, reliability and economy of the distribution network. It is especially suitable for multi-level active distribution networks containing distributed photovoltaic systems.

[0086] Optionally, the grid structure of the power distribution system includes a grid level and nodes and connection methods of the grid level; the grid level includes a main distribution layer, a secondary distribution layer and an end-user access layer.

[0087] Optionally, the data collection and preprocessing module 220 is specifically used for:

[0088] Remove outliers and noise from historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization.

[0089] Optionally, the data collection and preprocessing module 220 is specifically used for:

[0090] Machine learning algorithms are used to identify and remove outliers and noise, and principal component analysis is used for data dimensionality reduction and feature extraction.

[0091] Optionally, the residual current mode determination module 230 is specifically used for:

[0092] The correlation factors are screened through grey correlation analysis or principal component analysis.

[0093] Optionally, the prediction module 240 is specifically used for:

[0094] Establish a preliminary prediction model using grey system theory;

[0095] The parameters in the preliminary model were determined by the least squares method;

[0096] The preliminary model is tested and revised to determine the residual current prediction model.

[0097] Optionally, the prediction module 240 further includes:

[0098] Regularly compare the residual current predicted by the residual current preset model with the actual monitored residual current, and adjust the parameters of the residual current prediction model according to the comparison results.

[0099] In some embodiments, the residual current prediction model adopts a Transformer model, a convolutional neural network model, or a long short-term memory network model.

[0100] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0101] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-level active distribution network residual current early warning method, characterized in that: include: Establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the location of the distributed photovoltaic system in the distribution network; Obtain historical residual current data of different grid levels in the grid structure of the distribution system, and pre-process the historical residual current data; According to the pre-processed historical residual current data, the residual current change trend is determined through time series, and the residual current mode at different grid levels is determined; Combine the residual current change trend and residual current mode to screen the correlation factors of residual current, build a residual current prediction model based on grey system theory, and make a baseline prediction of residual current according to the residual current change trend and residual current prediction model; A residual current baseline prediction based on a residual current prediction model and a preset threshold value is set based on a distribution system simulation model; When the prediction model predicts that the residual current exceeds the preset threshold, an early warning is triggered.

2. The multi-level active distribution network residual current early warning method according to claim 1 is characterized in that: The grid structure of the power distribution system includes a grid level and nodes and connection methods of the grid level; the grid level includes a main distribution layer, a secondary distribution layer and an end-user access layer.

3. The multi-level active distribution network residual current early warning method according to claim 1 is characterized in that: The historical residual current data includes residual current data of different grid levels under different working conditions when there is no photovoltaic access, residual current data of different grid levels under different working conditions when there is photovoltaic access, configuration parameters and operation data of distributed photovoltaic access points, and environmental data; wherein the environmental data includes ambient temperature and ambient humidity; The correlation factors include at least one of the operating status of power grid equipment, environmental factors, load conditions, power grid parameters, power grid wiring mode and line status.

4. The multi-level active distribution network residual current early warning method according to claim 1, characterized in that: Obtain historical residual current data of different grid levels in the grid structure of the distribution system, and pre-process the historical residual current data, including: Remove outliers and noise from historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization.

5. The multi-level active distribution network residual current early warning method according to claim 4 is characterized in that: Remove outliers and noise from historical residual current data, perform data dimension reduction and feature extraction, and perform normalization or standardization, including: Machine learning algorithms are used to identify and remove outliers and noise, and principal component analysis is used for data dimensionality reduction and feature extraction.

6. The multi-level active distribution network residual current early warning method according to claim 1, characterized in that: Combine the residual current change trend and residual current mode to screen the associated factors of residual current, including: The correlation factors are screened through grey correlation analysis or principal component analysis.

7. The multi-level active distribution network residual current early warning method according to claim 1, characterized in that: The residual current prediction model is constructed based on the grey system theory, including: Establish a preliminary prediction model using grey system theory; The parameters in the preliminary model were determined by the least squares method; The preliminary model is tested and revised to determine the residual current prediction model.

8. The multi-level active distribution network residual current early warning method according to claim 1, characterized in that: After building the residual current prediction model based on the grey system theory, it also includes: Regularly compare the residual current predicted by the residual current preset model with the actual monitored residual current, and adjust the parameters of the residual current prediction model according to the comparison results.

9. The multi-level active distribution network residual current early warning method according to claim 1, characterized in that: The residual current prediction model adopts a Transformer model, a convolutional neural network model or a long short-term memory network model.

10. A multi-level active distribution network residual current early warning device, characterized in that: include: A simulation module is used to establish a distribution system simulation model based on the grid structure of the distribution system, and determine the access point of the distributed photovoltaic system in the distribution system and the position of the distributed photovoltaic system in the distribution network; A data collection and preprocessing module is used to obtain historical residual current data of different grid levels in the grid structure of the distribution system and preprocess the historical residual current data; A residual current mode determination module is used to determine the residual current mode of different grid levels based on the pre-processed historical residual current data and the changing trend of the residual current through the time series; A prediction module is used to screen the correlation factors of the residual current by combining the residual current variation trend and the residual current mode, to build a residual current prediction model based on the grey system theory, and to make a baseline prediction of the residual current according to the residual current variation trend and the residual current prediction model; A preset threshold determination module, used to set a preset threshold based on a residual current baseline prediction of a residual current prediction model and a power distribution system simulation model; The early warning module is used to trigger an early warning when the prediction model predicts that the residual current exceeds a preset threshold.

Images