Power distribution network line loss abnormity identification method and system

Abstract

The invention discloses a power distribution network line loss abnormity identification method and system, and relates to the technical field of power systems, and the method comprises the steps: collecting and preprocessing historical data, meteorological data and geographic position information, extracting and fusing features to form a comprehensive feature vector, building a power distribution network topological structure diagram based on the comprehensive feature vector, and obtaining a power distribution network line loss abnormity identification result; analyzing by using a graph neural network to obtain a power grid result; a threshold is set in combination with historical line loss data, a line loss prediction model is constructed to output a prediction result, the prediction result is compared with historical data to calculate an actual line loss value, and a monitoring system is triggered to take corresponding measures once the actual line loss value exceeds the threshold. And the fused feature vector is converted into a comprehensive feature vector by using nonlinear conversion, so that the nonlinear fitting capability of the model is enhanced, and the distinction degree of the features and the generalization capability of the model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method and system for identifying abnormal line loss in a distribution network. Background Art

[0002] The power system is one of the important infrastructures that are indispensable to modern society. In recent years, with the development of smart grid technology, the data collection, processing and analysis capabilities of the power system have been significantly improved, which has put forward higher requirements for the abnormal identification of line loss in the distribution network. Traditional line loss abnormality identification methods mainly rely on manual experience and simple statistical analysis. This method is not only inefficient, but also easy to miss potential problems. With the advancement of big data and artificial intelligence technology, data-driven methods have gradually become a research hotspot.

[0003] Most existing line loss prediction models are based on a single data source, ignoring the interaction between different data, which reduces the accuracy of the prediction. In addition, existing line loss anomaly detection methods often lack the time scale matching between actual line loss values ​​and historical line loss data, which leads to false positives or false negatives when performing anomaly detection. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a distribution network line loss anomaly identification method and system to solve the problems of insufficient data fusion and low model prediction accuracy in the existing distribution network line loss anomaly identification method.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for identifying abnormal line loss in a distribution network, which comprises:

[0008] Collect historical data, meteorological data and geographic location information, and perform pre-processing;

[0009] Extract features from the preprocessed data and fuse the extracted features to form a comprehensive feature vector;

[0010] The topological structure diagram of the distribution network is established based on the comprehensive feature vector, and the topological structure of the distribution network is analyzed using the graph neural network to obtain the power grid analysis results;

[0011] Establish line loss thresholds based on historical line loss data;

[0012] Build a line loss prediction model based on the power grid analysis results, output the line loss prediction results, calculate the line loss prediction results with the historical line loss data, and obtain the actual line loss value;

[0013] When the actual line loss value exceeds the line loss threshold, the monitoring system will take immediate action.

[0014] As a preferred solution of the method for identifying abnormal line loss in a distribution network of the present invention, the historical data includes historical monitoring operation data, user power consumption data, and distribution automation system data;

[0015] The meteorological data include temperature, humidity, and wind speed;

[0016] The geographical location information includes location information of each device in the distribution network;

[0017] The preprocessing of the collected data includes data cleaning, filling missing values, unifying data formats and normalization.

[0018] As a preferred solution of the method for identifying abnormal line loss in a distribution network of the present invention, the features of line loss are extracted from the preprocessed data;

[0019] The line loss features are fused to obtain a fused feature vector;

[0020] The fused feature vector is transformed into a comprehensive feature vector through nonlinear transformation.

[0021] As a preferred solution of the method for identifying abnormal line loss in the distribution network of the present invention, the topological structure diagram of the distribution network is established based on the comprehensive feature vector, and the specific steps are as follows:

[0022] The system receives a comprehensive feature vector;

[0023] Establish a topological structure graph containing nodes, edges, attributes, and partitions based on the connection relationship between devices;

[0024] Define the propagation layer of the graph neural network, where each node receives information from its neighbor nodes and fuses this information with its own features;

[0025] By iterating the operation of the propagation layer multiple times, the hidden state of each node is gradually updated;

[0026] Repeat the operation of the propagation layer to obtain the final hidden state;

[0027] Perform global pooling based on the final hidden state to obtain the global feature vector;

[0028] The global eigenvector is transformed nonlinearly to obtain the power grid analysis results.

[0029] As a preferred solution of the method for identifying abnormal line loss in a distribution network according to the present invention, wherein: based on historical line loss data, a line loss threshold is determined by statistical analysis;

[0030] Use Laplace kernel function to perform feature weighting on power grid analysis results;

[0031] The line loss prediction result is output based on the feature weighted result.

[0032] As a preferred solution of the method for identifying abnormal line loss in a distribution network of the present invention, the time resolution of the historical line loss data and the line loss prediction result is determined, and the historical line loss data and the line loss prediction result are aligned on the time axis;

[0033] Calculate the difference between the line loss prediction result and the historical line loss data.

[0034] As a preferred solution of the method for identifying abnormal line loss in a distribution network of the present invention, wherein: based on comparing the actual line loss value with the line loss threshold;

[0035] When the actual line loss value exceeds the line loss threshold, the monitoring system will issue an alarm;

[0036] The monitoring system will combine equipment status, historical operation data, and meteorological data to analyze the causes and areas of abnormalities, and send them to all operation and maintenance personnel;

[0037] The system records the time, location, difference size and cause of each abnormal event, analyzes common problems based on the recorded data, and proposes improvement measures.

[0038] In a second aspect, the present invention provides a distribution network line loss anomaly identification system, comprising:

[0039] Data collection and preprocessing module, which collects historical data, meteorological data and geographic location information, and performs data cleaning, fills in missing values, unifies data formats and normalizes data;

[0040] The feature extraction and fusion module extracts the features related to line loss from the preprocessed data and fuses these features to form a comprehensive feature vector;

[0041] The topology structure and analysis module builds the topology diagram of the distribution network based on the comprehensive feature vector, and uses the graph neural network to analyze the topology of the distribution network to obtain the power grid analysis results;

[0042] Line loss threshold setting module, which determines the line loss threshold through statistical analysis based on historical line loss data;

[0043] The line loss prediction and comparison module builds a line loss prediction model based on the power grid analysis results, outputs the line loss prediction results, and calculates the prediction results with the historical line loss data to obtain the actual line loss value;

[0044] Abnormal detection and response module: when the actual line loss value exceeds the line loss threshold, the monitoring system will immediately take measures to notify the operation and maintenance personnel and record and display the abnormal situation.

[0045] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the distribution network line loss anomaly identification method as described in the first aspect of the present invention is implemented.

[0046] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for identifying abnormal distribution network line losses as described in the first aspect of the present invention is implemented.

[0047] The beneficial effects of the present invention are as follows: by extracting features about line loss from preprocessed data and fusing these features to form a comprehensive feature vector, the features of multi-source data are effectively extracted and integrated, and the representativeness of the features is enhanced. This step utilizes nonlinear transformation to transform the fused feature vector into a comprehensive feature vector, which enhances the nonlinear fitting ability of the model, improves the feature discrimination and the generalization ability of the model; and by establishing a topological structure diagram of the distribution network based on the comprehensive feature vector, the topological structure of the distribution network is analyzed using a graph neural network to obtain the power grid analysis results. This step utilizes the powerful representation ability of the graph neural network to effectively capture the complex relationships between devices in the power grid and their dynamic changes, and improves the sensitivity and accuracy of line loss anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 This is a flow chart of the distribution network line loss abnormality identification method and system in Example 1.

[0050] Figure 2 This is a determination diagram of the distribution network line loss abnormality identification method and system in Example 1. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a method for identifying abnormal line loss in a distribution network, comprising the following steps:

[0053] S1. Collect historical data, meteorological data and geographic location information;

[0054] S1.1. The historical data includes historical monitoring operation data, user power consumption data, and distribution automation system data;

[0055] Historical monitoring operation data is the historical record of power grid operation, such as current, voltage, and power loss;

[0056] The user's electricity consumption data includes the user's electricity consumption, electricity consumption period, and electricity consumption mode information;

[0057] Distribution automation system data is data from the distribution automation system (DAS), including switch status, relay action records, etc.

[0058] S1.2. The meteorological data include environmental parameters such as temperature, humidity, wind speed, etc., which may affect the operation status of the power grid;

[0059] S1.3. The geographical location information includes the location information of each device in the distribution network, which is used to construct a geographical distribution map of the power grid.

[0060] Further explanation: By comprehensively collecting different types of data, we can ensure that there are abundant information sources in the analysis process, thereby improving the accuracy and comprehensiveness of line loss anomaly identification. Diversified data sources provide more basis for subsequent data analysis and help discover potential influencing factors.

[0061] S1.4. Preprocessing of the collected data includes data cleaning, filling missing values, unifying data formats and normalization.

[0062] The data cleaning refers to removing noise and erroneous data, such as outliers and duplicate records;

[0063] Filling missing values ​​refers to filling missing values ​​in data by interpolation or other methods to avoid analysis bias caused by missing data;

[0064] The unified data formatting refers to converting data from different sources into a unified format to facilitate subsequent processing;

[0065] The normalization process refers to normalizing the data to eliminate the influence of the dimension so that data of different magnitudes can be compared on the same scale.

[0066] S2, extracting features about line loss from the preprocessed data;

[0067] The characteristics of the line loss include electrical parameters, operating conditions, environmental factors, geographical locations and historical data;

[0068] Electrical parameters include current, voltage, and power factor;

[0069] The operating status includes the operating time and switch status of the equipment;

[0070] Environmental factors include temperature, humidity, and wind speed;

[0071] Geographic location includes the location information of the equipment and its impact on line loss;

[0072] Historical data includes past line loss records and abnormal conditions.

[0073] S2.1, fusing the line loss features to obtain a fused feature vector;

[0074]

[0075] Among them, C represents the fused feature vector, N represents the total number of feature vectors, and F i represents the i-th eigenvector, and W represents the weight of the i-th eigenvector.

[0076] Further explanation: By merging multiple feature vectors into a comprehensive feature vector through weighted averaging, the importance and relationship of each feature can be better reflected. This not only reduces the feature dimension, but also enhances the representativeness of the feature, and improves the generalization ability and computational efficiency of the model.

[0077] S2.2, the fused feature vector is transformed into a comprehensive feature vector through nonlinear transformation, and its expression is:

[0078]

[0079] Among them, Z represents the comprehensive eigenvector, g represents the value after nonlinear transformation, and η represents the steepness of nonlinear transformation.

[0080] It is further shown that through nonlinear transformation, the features of different categories can be better separated, thereby improving the classification and prediction performance of the model.

[0081] S3. Establish the topological structure diagram of the distribution network based on the comprehensive feature vector. The specific steps are as follows:

[0082] S3.1, the system receives the comprehensive feature vector after nonlinear transformation;

[0083] S3.2. Establish a topology diagram based on the physical connection relationship between devices. The topology diagram includes nodes, edges, attributes, and partitions.

[0084] The nodes represent key equipment in the power grid, such as transformers, switches, and circuit breakers;

[0085] The edges represent connection relationships between devices, such as cables or wires;

[0086] The attributes indicate that each node and edge is accompanied by attribute information, such as device model, capacity, line length, etc.;

[0087] Zoning refers to dividing the power grid into different areas based on geographical distribution for better management and analysis.

[0088] S3.3. Define the propagation layer of the graph neural network. Each node receives messages from its neighbor nodes and fuses these messages with its own features. Based on the received messages and its own features, the node's hidden state is updated. Through multiple iterations of the propagation layer operation, the hidden state of each node is gradually updated. After each layer of propagation, the node's hidden state will contain more information from its neighbors.

[0089] In order to better capture the complex relationships in the graph, multiple layers of propagation are required. Each layer of propagation further expands the range of the node, enabling it to obtain information from more distant neighbors. Repeating the operation of the propagation layer, the final hidden state is obtained.

[0090] Based on the final hidden state, global pooling is performed to obtain the global feature vector, which is expressed as:

[0091]

[0092] Among them, G represents the global feature vector, M represents the total number of nodes, represents the hidden state of the jth node in the last layer;

[0093] The global eigenvector is transformed nonlinearly to obtain the power grid analysis results.

[0094] S4. According to the historical line loss data, the line loss threshold is determined through statistical analysis, and its expression is:

[0095]

[0096] Among them, T represents the line loss threshold, quantile represents the quantile used to calculate the corresponding given cumulative probability α, x represents the historical line loss data point, σ represents the standard deviation of the historical line loss data, dt represents the differential sign, α represents the pre-set threshold percentage, usually taking a value of 0.95 or higher, indicating the confidence level, μ represents the mean, and s represents the integral variable.

[0097] S4.1. Use the Laplace kernel function to perform feature weighting on the power grid analysis results. The specific steps are as follows:

[0098] For each time point t', calculate the Laplace kernel function value between it and the current time point t;

[0099] Multiply the power grid analysis result Y(t') at each time point t' by its corresponding kernel function value, and integrate and sum all time points;

[0100] The line loss prediction result is output based on the feature weighted result, and its expression is:

[0101]

[0102] Among them, A pred represents the line loss prediction result, Y(t') represents the power grid analysis result at time point t', β represents the constant for adjusting the steepness of the Laplace kernel function, t represents the time point, and dt' represents the differential on the time variable t'.

[0103] It is further shown that the Laplace kernel function is very sensitive to local changes and can better capture short-term fluctuations in power grid analysis results. This local sensitivity helps to improve the accuracy of line loss prediction, especially when the power grid operation status changes rapidly.

[0104] S5. In order to ensure that historical line loss data and line loss prediction results can be accurately aligned and effectively compared, an appropriate time resolution needs to be determined;

[0105] The choice of time resolution is based on several factors,

[0106] Data collection frequency,The collection frequency of historical line loss data determines the available time granularity. For example, if historical line loss data is collected once every hour, the time resolution can be set to every hour.

[0107] Requirements of the prediction model,Line loss prediction models may require a specific temporal resolution to generate prediction results.,For example, some models may be more suitable for daily or weekly data.

[0108] According to the actual application requirements, a time resolution that can meet the accuracy requirements and ensure computational efficiency is selected.

[0109] S5.1. After determining the time resolution, the historical line loss data and the line loss prediction results need to be aligned on the time axis;

[0110] Data preprocessing: preprocess the historical line loss data and line loss prediction results to ensure that they have the same time resolution. This may include operations such as interpolation and resampling.

[0111] Timestamp matching: assign a corresponding timestamp to each data point to ensure that the timestamps of historical line loss data and line loss prediction results are consistent. For example, if the time resolution is hourly, each data point should have a timestamp accurate to the hour.

[0112] Data alignment: align historical line loss data and line loss prediction results according to timestamps. This ensures that data at the same time point can be directly compared.

[0113] S5.1. Calculate the difference between the line loss prediction result and the historical line loss data. The difference represents the actual line loss value. The expression is:

[0114] Actual Loss = A pred -Historical Loss;

[0115] Among them, Actual Loss represents the actual line loss value, and Historical Loss represents the historical line loss value;

[0116] Further explanation: By determining a uniform time resolution and aligning the data on the time axis, direct comparison of historical line loss data and line loss prediction results at the same time point is ensured. This precise alignment helps improve the accuracy of line loss analysis.

[0117] S6. The system continuously monitors the actual line loss value and compares it with the set line loss threshold;

[0118] When the actual line loss value exceeds the preset line loss threshold, the system will automatically trigger the alarm mechanism to ensure that the operation and maintenance personnel can receive the alarm information in time;

[0119] S6.1. The monitoring system will combine equipment status, historical operation data, and meteorological data to analyze the causes and areas of abnormalities, and send them to all operation and maintenance personnel;

[0120] The specific operations for analyzing abnormal causes and abnormal areas are as follows:

[0121] Equipment status, check the status of key equipment (such as transformers and switches) to identify whether there are faults or performance degradation; historical operation data, analyze historical operation data to find out whether there are records of similar abnormal situations to determine whether there are periodic or repetitive failure modes; meteorological data, consider environmental factors such as temperature, humidity, wind speed, etc., which may affect the operating status of the power grid;

[0122] Abnormal area location: through data analysis, determine the specific area where the abnormality occurs so as to quickly respond and handle it;

[0123] Send abnormal alerts and analysis results to all relevant operation and maintenance personnel to ensure they can take prompt action.

[0124] S6.2. The system will record the detailed information of each abnormal event, including time, location, difference size and the cause of the abnormality;

[0125] Regularly analyze recorded data to identify common problems and potential risk points;

[0126] Based on the results of data analysis, specific improvement measures are proposed to reduce the occurrence of similar abnormal events in the future. This may include equipment maintenance, system optimization, adjustment of operating procedures, etc.

[0127] This embodiment also provides a distribution network line loss abnormality identification method system, including:

[0128] Data collection and preprocessing module, which collects historical data, meteorological data and geographic location information, and performs data cleaning, fills in missing values, unifies data formats and normalizes data;

[0129] The feature extraction and fusion module extracts the features related to line loss from the preprocessed data and fuses these features to form a comprehensive feature vector;

[0130] The topology structure and analysis module builds the topology diagram of the distribution network based on the comprehensive feature vector, and uses the graph neural network to analyze the topology of the distribution network to obtain the power grid analysis results;

[0131] Line loss threshold setting module, which determines the line loss threshold through statistical analysis based on historical line loss data;

[0132] The line loss prediction and comparison module builds a line loss prediction model based on the power grid analysis results, outputs the line loss prediction results, and calculates the prediction results with the historical line loss data to obtain the actual line loss value;

[0133] Abnormal detection and response module: when the actual line loss value exceeds the line loss threshold, the monitoring system will immediately take measures to notify the operation and maintenance personnel and record and display the abnormal situation.

[0134] This embodiment also provides a computer device, which is applicable to the distribution network line loss abnormality identification method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the distribution network line loss abnormality identification method proposed in the above embodiment.

[0135] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0136] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for identifying abnormal line loss in a distribution network proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0137] In summary, the present invention achieves effective extraction and integration of multi-source data features and enhances the representativeness of features by: extracting features about line loss from preprocessed data and fusing these features to form a comprehensive feature vector. This step utilizes nonlinear transformation to convert the fused feature vector into a comprehensive feature vector, which enhances the nonlinear fitting ability of the model, improves the feature discrimination and the generalization ability of the model; and by establishing a topological structure diagram of the distribution network based on the comprehensive feature vector, using a graph neural network to analyze the topological structure of the distribution network, the power grid analysis results are obtained. This step utilizes the powerful representation ability of the graph neural network to effectively capture the complex relationships between devices in the power grid and their dynamic changes, thereby improving the sensitivity and accuracy of line loss anomaly detection.

[0138] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a method for identifying abnormal line loss in a distribution network is provided.

[0139] We selected a city's distribution network as the test object. The distribution network includes multiple substations, switch stations and distribution lines, covering an area of ​​about 50 square kilometers and serving about 100,000 users.

[0140] Data collection and preprocessing:

[0141] Historical data: Historical monitoring operation data, user power consumption data and distribution automation system data for the past year were collected from the power company's database.

[0142] Meteorological data: Temperature, humidity, and wind speed data for the past year were obtained from the local meteorological bureau.

[0143] Geographic location information: The location information of each device in the distribution network is obtained through the GIS system.

[0144] The collected data is preprocessed. The specific steps are as follows:

[0145] Data cleaning: Remove noise and outliers, such as extreme voltage fluctuations and current mutations.

[0146] Fill missing values: Use interpolation to fill missing data points.

[0147] Unified data format: Convert data from different sources into a unified CSV format.

[0148] Normalization: Normalize all data so that the data is in the range of [0,1].

[0149] Feature extraction and fusion:

[0150] Features related to line loss are extracted from the preprocessed data, such as electrical parameters such as current, voltage, and power factor, as well as environmental parameters such as temperature and humidity.

[0151] The extracted features are fused to form a comprehensive feature vector. During the fusion process, different weights are assigned to each feature according to its importance.

[0152] The fused feature vector is transformed into a comprehensive feature vector through nonlinear transformation to enhance the nonlinear fitting ability of the model.

[0153] Create a topology diagram:

[0154] The system receives a composite feature vector.

[0155] A topological structure diagram is established based on the connection relationship between devices. Nodes represent key equipment (such as transformers and switches), edges represent the connection relationship between devices, attributes represent device information, and partitions represent geographical distribution locations.

[0156] Define the propagation layer of the graph neural network, obtain the hidden state of the propagation layer, and repeat the operation of the propagation layer to obtain the final hidden state.

[0157] Global pooling is performed based on the final hidden state to obtain the global feature vector.

[0158] The global eigenvector is transformed nonlinearly to obtain the power grid analysis results.

[0159] Line loss threshold setting and prediction model construction:

[0160] Based on historical line loss data, the line loss threshold is determined through statistical analysis.

[0161] Normalize the power grid analysis results.

[0162] The Laplace kernel function is used to perform feature weighting on the normalized power grid analysis results.

[0163] The line loss prediction result is output based on the feature weighted result.

[0164] Actual line loss value calculation and monitoring:

[0165] Determine the temporal resolution of historical line loss data and line loss prediction results, and align the two on the time axis.

[0166] Calculate the difference between the line loss prediction result and the historical line loss data to obtain the actual line loss value.

[0167] When the actual line loss value exceeds the line loss threshold, the monitoring system will immediately issue an alarm, and combine the equipment status, historical operation data, and meteorological data to analyze the cause and abnormal area of ​​the abnormality and send it to all operation and maintenance personnel.

[0168] The system records the time, location, difference size and cause of each abnormal event, analyzes common problems based on the recorded data, and proposes improvement measures.

[0169] The details are shown in Table 1 below:

[0170] Table 1 Line loss technology comparison table

[0171]

[0172]

[0173] Through the analysis of the above test data, it can be clearly seen that the distribution network line loss anomaly identification method proposed in the present invention has significant advantages and innovations compared with the comparative technology.

[0174] Improved accuracy:

[0175] From the table, we can see that for test objects A, B, C and D, the predicted line loss rates of the comparison technology are 0.75%, 1.15%, 0.85% and 1.45% respectively, while the actual line loss values ​​are 0.76%, 1.18%, 0.87% and 1.48% respectively. The errors are 0.01%, 0.03%, 0.02% and 0.03% respectively.

[0176] The predicted line loss rates of the present invention are 0.74%, 1.14%, 0.84% ​​and 1.44% respectively, while the actual line loss values ​​are 0.75%, 1.16%, 0.85% and 1.46% respectively, and the errors are 0.01%, 0.02%, 0.01% and 0.02% respectively.

[0177] By comparison, it can be seen that the prediction error of the present invention is generally lower than that of the comparison technology, especially in test objects B and D, the error is reduced by 0.01% and 0.01% respectively. This shows that the method of the present invention can predict line loss more accurately and improve the prediction accuracy.

[0178] Real-time and responsiveness:

[0179] The present invention uses real-time monitoring and rapid response mechanisms. When the actual line loss value exceeds the line loss threshold, the monitoring system can immediately issue an alarm and take corresponding measures. For example, in test object D, the actual line loss value of the comparison technology is 1.48%, slightly lower than the threshold of 1.6%, but the actual line loss value of the present invention is 1.46%, which is also below the threshold, but the system can detect potential problems faster and issue early warnings.

[0180] Compared with other technologies, which usually rely on manual inspection and manual intervention, the response speed is slow and it is not possible to find and handle abnormal situations in time. The real-time monitoring and automatic response mechanism of the present invention greatly improves the response speed and efficiency of the system.

[0181] Multi-source data fusion and feature extraction:

[0182] The present invention forms a comprehensive feature vector by collecting and integrating historical data, meteorological data and geographical location information. This multi-source data fusion method can more comprehensively reflect the various factors affecting line loss, thereby improving the accuracy of prediction.

[0183] For example, in test object C, the actual line loss value of the prior art is 0.87%, while the actual line loss value of the present invention is 0.85%. Through comprehensive feature vector and graph neural network analysis, the present invention can better capture the changing trend of line loss, thereby maintaining high accuracy in more complex environments.

[0184] Comparative techniques usually rely on a single data source and fail to fully consider the impact of various factors, resulting in lower prediction accuracy in certain complex situations.

[0185] Topological structure analysis and graph neural networks:

[0186] The present invention establishes a topological structure diagram of the distribution network based on the comprehensive feature vector and uses a graph neural network for analysis. This method can effectively capture the complex relationship between devices in the power grid and their dynamic changes, and improve the sensitivity and accuracy of line loss anomaly detection.

[0187] For example, in test object B, the actual line loss value of the prior art is 1.18%, while the actual line loss value of the present invention is 1.16%. Through the analysis of graph neural network, the present invention can more accurately detect potential problems and give early warning, avoiding a wider range of abnormalities.

[0188] Contrast techniques usually use simple statistical methods or rule bases, which are difficult to handle complex topological structures and dynamic changes, and therefore have shortcomings in anomaly detection.

[0189] In summary, the distribution network line loss anomaly identification method proposed in the present invention significantly improves the accuracy, real-time performance and response speed of line loss prediction through multi-source data fusion, feature extraction, topology structure analysis and real-time monitoring. These innovative methods and technologies not only solve the deficiencies in the comparative technologies, but also provide a strong guarantee for the safe and stable operation of the distribution network. By comparing with the comparative technologies, the performance of the present invention is better than that of the comparative technologies in all aspects, demonstrating its significant innovation and advantages.

[0190] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for identifying abnormal line loss in a distribution network, characterized in that: include, Collect historical data, meteorological data and geographic location information, and perform pre-processing; Extract features from the preprocessed data and fuse the extracted features to form a comprehensive feature vector; The topological structure diagram of the distribution network is established based on the comprehensive feature vector, and the topological structure of the distribution network is analyzed using the graph neural network to obtain the power grid analysis results; Establish line loss thresholds based on historical line loss data; Build a line loss prediction model based on the power grid analysis results, output the line loss prediction results, calculate the line loss prediction results with the historical line loss data, and obtain the actual line loss value; When the actual line loss value exceeds the line loss threshold, the monitoring system will take immediate action.

2. The method for identifying abnormal line loss in a distribution network according to claim 1, characterized in that: Collect historical data, meteorological data and geographic location information, and perform preprocessing, including the following steps: The historical data includes historical monitoring operation data, user power consumption data, and distribution automation system data; The meteorological data include temperature, humidity, and wind speed; The geographical location information includes location information of each device in the distribution network; The preprocessing of the collected data includes data cleaning, filling missing values, unifying data formats and normalization.

3. The method for identifying abnormal line loss in a distribution network according to claim 2, characterized in that: The preprocessed data is subjected to feature extraction, and the extracted features are fused to form a comprehensive feature vector, including the following steps: Extracting features about line loss from the preprocessed data; The line loss features are fused to obtain a fused feature vector; The fused feature vector is transformed into a comprehensive feature vector through nonlinear transformation.

4. The method for identifying abnormal line loss in a distribution network according to claim 3, characterized in that: The topological structure diagram of the distribution network is established based on the comprehensive feature vector. The specific steps are as follows: The system receives a comprehensive feature vector; Establish a topological structure graph containing nodes, edges, attributes, and partitions based on the connection relationship between devices; Define the propagation layer of the graph neural network, where each node receives information from its neighbor nodes and fuses this information with its own features; By iterating the operation of the propagation layer multiple times, the hidden state of each node is gradually updated; Repeat the operation of the propagation layer to obtain the final hidden state; Perform global pooling based on the final hidden state to obtain the global feature vector; The global eigenvector is transformed nonlinearly to obtain the power grid analysis results.

5. The method for identifying abnormal line loss in a distribution network according to claim 4, characterized in that: The line loss threshold is set based on the historical line loss data, the line loss prediction model is constructed according to the power grid analysis results, and the line loss prediction results are output, including the following steps: Determine the line loss threshold through statistical analysis based on historical line loss data; Use Laplace kernel function to perform feature weighting on power grid analysis results; The line loss prediction result is output based on the feature weighted result.

6. The method for identifying abnormal line loss in a distribution network according to claim 5, characterized in that: Determine the time resolution of historical line loss data and line loss prediction results, and align the historical line loss data with the line loss prediction results on the time axis; Calculate the difference between the line loss prediction result and the historical line loss data.

7. The method for identifying abnormal line loss in a distribution network according to claim 6, characterized in that: When the actual line loss value exceeds the line loss threshold, the monitoring system will immediately take measures, including the following steps: Based on the comparison between the actual line loss value and the line loss threshold; When the actual line loss value exceeds the line loss threshold, the monitoring system will issue an alarm; The monitoring system will combine equipment status, historical operation data, and meteorological data to analyze the causes and areas of abnormalities, and send them to all operation and maintenance personnel; The system records the time, location, difference size and cause of each abnormal event, analyzes common problems based on the recorded data, and proposes improvement measures.

8. A distribution network line loss anomaly identification system, based on the distribution network line loss anomaly identification method according to any one of claims 1 to 7, characterized in that: include, Data collection and preprocessing module, which collects historical data, meteorological data and geographic location information, and performs data cleaning, fills in missing values, unifies data formats and normalizes data; The feature extraction and fusion module extracts the features related to line loss from the preprocessed data and fuses these features to form a comprehensive feature vector; The topology structure and analysis module builds the topology diagram of the distribution network based on the comprehensive feature vector, and uses the graph neural network to analyze the topology of the distribution network to obtain the power grid analysis results; Line loss threshold setting module, which determines the line loss threshold through statistical analysis based on historical line loss data; The line loss prediction and comparison module builds a line loss prediction model based on the power grid analysis results, outputs the line loss prediction results, and calculates the prediction results with the historical line loss data to obtain the actual line loss value; Abnormal detection and response module: when the actual line loss value exceeds the line loss threshold, the monitoring system will immediately take measures to notify the operation and maintenance personnel and record and display the abnormal situation.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for identifying abnormal line loss in a distribution network according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for identifying abnormal line loss in a distribution network according to any one of claims 1 to 7 are implemented.

Images