Intelligent diagnosis method for low-voltage power distribution network line loss anomaly based on machine learning algorithm

Abstract

This invention discloses an intelligent diagnostic method for line loss anomalies in medium- and low-voltage distribution networks based on machine learning algorithms, belonging to the field of medium- and low-voltage distribution network technology. The method includes the following steps: collecting and preprocessing operational data of the medium- and low-voltage distribution network; extracting and selecting features from the operational data; training an intelligent diagnostic model for line loss anomalies in the medium- and low-voltage distribution network; diagnosing line losses in the medium- and low-voltage distribution network, locating the location of the anomalies, identifying the causes of the anomalies, and intelligently managing the line loss anomalies. This invention solves the problems of insufficient detection accuracy in existing methods, which prevents refined management of distribution network line losses and reduces grid operating efficiency. This invention can effectively improve the accuracy of line loss anomaly identification, reduce unnecessary energy waste, provide strong support for the sustainable development of the power system, and achieve refined management of distribution network line losses, thereby improving the overall operating efficiency of the power grid.

Description

Technical Field

[0001] This invention relates to the field of medium and low voltage distribution network technology, specifically to an intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms. Background Technology

[0002] Line loss refers to the loss of electrical energy during power transmission due to various factors. This loss is usually unavoidable, and its main causes include physical phenomena such as resistance heating, as well as the influence of users' electricity consumption behavior. In medium and low voltage distribution networks, abnormal line loss means that the actual line loss rate is much higher than the reasonable line loss level. This abnormal situation will not only significantly increase the operating cost of the power system, but may also reflect problems in the power grid at the technical level or in the management process.

[0003] Among the various factors contributing to abnormal line losses in the distribution network, physical issues such as excessively long conductors and voltage quality problems; management issues such as smart meter malfunctions; human factors such as illegal electricity use, theft, and leakage; and extreme conditions like severe weather can all lead to anomalies. Relying solely on manual statistics is not only time-consuming and labor-intensive but also ineffective. Traditional manual statistical methods are not only inefficient but also struggle to address complex anomaly diagnosis needs. For example, manual analysis typically relies on historical experience and simple threshold judgments, which often prove inadequate when facing dynamically changing electricity consumption environments and diverse anomaly types. Currently, most power companies use experience-based threshold setting methods to detect line loss anomalies, but this method has significant limitations: firstly, threshold design relies too heavily on human experience and lacks scientific basis; secondly, fixed thresholds cannot adapt to changes in electricity consumption characteristics in different regions and time periods, resulting in insufficient detection accuracy and failing to provide refined management of distribution network line losses, thus reducing the overall operational efficiency of the power grid. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms. This method can effectively improve the accuracy of line loss anomaly identification, reduce unnecessary energy waste, provide strong support for the sustainable development of the power system, realize refined management of distribution network line losses, improve the overall operating efficiency of the power grid, and solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] Intelligent diagnostic methods for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms include:

[0007] Collect and preprocess operational data from medium- and low-voltage power distribution networks;

[0008] Feature extraction is performed on the operation data of medium and low voltage distribution networks. Based on the target score value, the influence of the current feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is determined, and the operation feature data of medium and low voltage distribution networks are determined.

[0009] Based on machine learning algorithms, an intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks is constructed.

[0010] Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed and abnormal line losses are diagnosed. Combined with the line topology and equipment information, the location of abnormal line losses is located, and intelligent control is carried out on the abnormal line losses in medium and low voltage distribution networks.

[0011] Preferably, collect operational data from medium and low voltage distribution networks and perform the following operations:

[0012] Utilize Internet of Things (IoT) technology to install intelligent data acquisition devices in medium and low voltage power distribution networks;

[0013] The voltage status of the medium and low voltage distribution network is monitored and collected in real time based on intelligent data acquisition equipment to obtain voltage data of the medium and low voltage distribution network.

[0014] Based on intelligent data acquisition equipment, the current status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain current data of medium and low voltage distribution networks;

[0015] Based on intelligent data acquisition equipment, the power status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain power data of medium and low voltage distribution networks;

[0016] Based on intelligent data acquisition equipment, the load status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain load data of medium and low voltage distribution networks;

[0017] Among them, based on the voltage data, current data, power data and load data of the medium and low voltage distribution network, the operation data of the medium and low voltage distribution network based on the Internet of Things is determined.

[0018] Preferably, the operation data of medium and low voltage distribution networks are preprocessed, including:

[0019] Cleaning of IoT-based medium and low voltage power distribution network operation data;

[0020] Remove duplicate data, missing values, and outliers from the IoT-based operation data of medium and low voltage distribution networks that are not useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks;

[0021] For missing and outlier values ​​that are useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks, the median is used to fill in the missing values ​​and the average value is used to replace the outlier values.

[0022] Normalize the operation data of medium and low voltage distribution networks based on the Internet of Things;

[0023] The operation data of medium and low voltage distribution networks based on the Internet of Things (IoT) are converted into a unified format to remove the dimensional differences between the operation data of medium and low voltage distribution networks based on IoT, and to determine the standardized operation data of medium and low voltage distribution networks.

[0024] Preferably, feature extraction and selection are performed on the operation data of medium and low voltage distribution networks, including:

[0025] Based on principal component analysis, static and dynamic features related to line losses in medium and low voltage distribution networks are extracted from the operation data of medium and low voltage distribution networks.

[0026] Static characteristics include line parameters, equipment parameters, and topology.

[0027] Line parameters include line length, conductor type, cross-sectional area, and resistivity; equipment parameters include transformer capacity, model, and load rate; topology includes network structure and node connection relationships.

[0028] The dynamic characteristics include electrical measurement data, load characteristics, environmental factors, and time characteristics.

[0029] Electrical measurement data includes voltage, current, active power, reactive power, and power factor; load characteristics include load factor, load fluctuation, peak-to-valley difference, and load curve; environmental factors include temperature, humidity, and weather conditions; and time characteristics include hourly, daily, weekly, monthly, and seasonal dimensions.

[0030] Based on the random forest algorithm, the extracted static and dynamic features are selected, the influence of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is evaluated, the most valuable features for intelligent diagnosis of line loss anomalies in medium and low voltage distribution networks are screened out, and the operating characteristic data of medium and low voltage distribution networks are determined.

[0031] Preferably, the step of selecting extracted static and dynamic features based on the random forest algorithm and evaluating the impact of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks includes:

[0032] A feature set is established based on all dynamic and static features. The first change range of all dynamic features relative to historical dynamic features is obtained, and the second change range of all static features relative to historical static features is obtained. The ratio of the first change range to the second change range is used as the random sampling ratio of dynamic and static features.

[0033] According to the random sampling ratio, multiple feature groups are obtained by randomly sampling features with replacement from the feature set. Based on each feature group, a decision tree is constructed by combining network line loss anomaly information. When splitting the nodes of the decision tree, feature group features are randomly selected to determine the split point.

[0034] Each feature in the feature set is input into each decision tree in turn, and the score set of the current feature in all decision trees is determined based on the decision results;

[0035] The current feature is permuted to obtain a permuted feature value. The permuted feature value is then input into each decision tree in sequence to obtain a set of performance degradation magnitudes for all decision trees. The scoring weight for the current feature is determined based on the difference between the average degradation value of the set of performance degradation magnitudes and a preset reference degradation value.

[0036] The target score value for the current feature is obtained by weighting the average score value determined by the score set based on the score weights.

[0037] The degree of influence of the current feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is determined based on the target score value.

[0038] Preferably, an intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks is constructed, and the following operations are performed:

[0039] Collect historical data on the operation of medium and low voltage distribution networks, and divide the collected historical data to determine the training set and test set;

[0040] Based on machine learning algorithms, a training set is used to train the machine learning model, enabling the machine learning model to autonomously learn the intelligent diagnostic behavior of abnormal line losses in medium and low voltage distribution networks, and to determine the intelligent diagnostic model of abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms.

[0041] Based on the test set, the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms was tested. The accuracy, recall and F1 score were used to evaluate whether the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms can achieve the expected results.

[0042] Based on the evaluation results, the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms were adjusted. Through continuous iterative optimization, the optimal intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks was determined.

[0043] Preferably, the analysis includes the following:

[0044] Deploy the optimal intelligent diagnostic model for line loss anomalies in medium and low voltage distribution networks in a real intelligent diagnostic environment for line loss anomalies in medium and low voltage distribution networks.

[0045] The operational characteristic data of medium and low voltage distribution networks are input into the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks. Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed, and abnormal line losses in medium and low voltage distribution networks are diagnosed intelligently to determine the predicted values ​​of line losses in medium and low voltage distribution networks.

[0046] Preferably, abnormal diagnosis of line losses in medium and low voltage distribution networks includes:

[0047] The predicted line loss value is compared with the actual line loss value, the relative error is calculated, and the relative error is used to determine whether there is an abnormal line loss in the medium and low voltage distribution network.

[0048] in,

[0049] When the relative error is within 5%, the line loss of the medium and low voltage distribution network is considered to be in a normal state.

[0050] When the relative error exceeds 5%, the line loss of the medium and low voltage distribution network is determined to be in an abnormal state, triggering a line loss abnormality alarm and locating the abnormal line loss location.

[0051] Obtain topology and equipment information of medium and low voltage distribution network lines, and gradually narrow down the anomaly range to locate the specific line segment or equipment using the topology and equipment information;

[0052] Among them, obtaining the topology information of medium and low voltage distribution network lines includes the connection relationship of the lines, the length of the lines, the conductor type, the impedance parameters, and the location and status of the equipment;

[0053] A zonal analysis of the medium and low voltage distribution network is performed, dividing the medium and low voltage distribution network into several regions. Line loss anomalies are analyzed in each region, and the predicted and actual line loss values ​​of each region are calculated to identify the region with the largest deviation.

[0054] Check the status of equipment in areas with abnormal line loss, including: whether the transformer is overloaded or faulty, whether the switch is closed or opened normally, whether the capacitor is working normally, and determine whether there is any abnormality in the equipment through equipment monitoring data;

[0055] Analyze the line parameters and check the line parameters in areas with abnormal line loss, including: whether the line impedance is abnormal, whether there is three-phase imbalance or harmonic problems, and analyze the line operation status through electrical measurement data.

[0056] If the equipment and line parameters are normal, there may be electricity theft. In this case, the difference between the user's electricity consumption and historical data is compared, and an electricity theft detection algorithm is used in conjunction with on-site inspections to check whether the user's electricity meter has been tampered with.

[0057] By using topology and equipment information, the scope of the anomaly is gradually narrowed down to pinpoint the specific line segment or equipment. When the line loss of a feeder is abnormal, the various branches under that feeder are analyzed. When the line loss of a transformer is abnormal, the power consumption of users under that transformer is checked.

[0058] Preferably, intelligent management and control of abnormal line losses in medium and low voltage distribution networks includes:

[0059] The intelligent diagnostic results of abnormal line losses in medium and low voltage distribution networks are analyzed. Combined with historical data and on-site inspections, the causes of abnormal line losses are identified, including equipment failure, line problems, electricity theft, operational issues, and management problems. Based on the causes of abnormal line losses, intelligent control measures for reducing line losses in medium and low voltage distribution networks are formulated, and intelligent control of abnormal line losses in medium and low voltage distribution networks is implemented.

[0060] In response to equipment failures, the measures include replacing or repairing faulty equipment, upgrading equipment, and conducting regular inspections and maintenance, including replacing overloaded transformers, repairing switches with poor contact, adopting high-efficiency and energy-saving equipment, and establishing an equipment inspection system to promptly identify and address potential failures.

[0061] For line problems, the measures include replacing aging lines, optimizing line layout, and strengthening joint maintenance, including using high conductivity, low loss conductors, reducing line length, lowering line impedance, and regularly checking line joints to ensure good contact.

[0062] In response to electricity theft, measures include strengthening electricity inspections, using electricity theft detection technologies and legal means, including real-time monitoring of users' electricity consumption behavior through smart meters and electricity information collection systems, using machine learning algorithms to detect abnormal electricity consumption behavior, severely cracking down on electricity theft, and increasing the cost of illegal activities.

[0063] In response to operational issues, adjustments are made to the three-phase balance, harmonic mitigation is implemented, and load management is carried out. This includes reducing three-phase imbalance by adjusting the load or installing three-phase balancing devices, reducing harmonic pollution by installing filters or reactive power compensation devices, optimizing load distribution, and avoiding local overload.

[0064] In response to management issues, measures include improving the accuracy of data collection and transmission, reducing measurement errors, establishing a line loss assessment mechanism to incentivize maintenance personnel to reduce losses and increase efficiency, and utilizing big data, artificial intelligence, and other technologies to achieve real-time monitoring and intelligent diagnosis of line losses.

[0065] Preferably, adjusting the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms according to the evaluation results includes:

[0066] The types and distributions of line loss anomalies are determined from the evaluation results, and the anomaly weights of the anomaly types are determined based on the types and distributions of line loss anomalies.

[0067] Relevant historical operation data related to anomaly types are obtained from the power grid operation history data, and the importance of the current relevant historical operation data is determined based on the anomaly weight of the anomaly type.

[0068] Based on the importance of each relevant historical operation data, weighted data extraction is performed on the historical power grid operation data to obtain a new training set and a new validation set.

[0069] The mean square error, error rate, and F1 score for the line loss anomaly assessment are determined from the assessment results. Based on the mean square error, error rate, and F1 score, the adjustment coefficients for the model structure weights are determined.

[0070] The latest weights for the model structure are determined based on the adjustment coefficients.

[0071] Based on the new training set, the new validation set, and the latest weights, the intelligent diagnostic model for line loss anomalies in low-voltage distribution networks is retrained and adjusted to obtain the latest intelligent diagnostic model for line loss anomalies.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] This invention collects voltage, current, power, and load data during the operation of medium- and low-voltage distribution networks to determine the operational data of these networks. The data is preprocessed, and feature extraction is performed to extract feature vectors related to line losses. A random forest algorithm is used to select the extracted feature vectors to determine the operational characteristics of the medium- and low-voltage distribution network. Based on machine learning algorithms, historical operational data of the medium- and low-voltage distribution network are used to train a machine learning model, and the model parameters are optimized to determine the optimal intelligent diagnostic model for abnormal line losses in the medium- and low-voltage distribution network. The intelligent diagnostic model for abnormal power grid line losses analyzes the operational characteristic data of medium and low voltage distribution networks, diagnoses abnormal line losses, and, by combining line topology and equipment information, locates the abnormal line loss position, identifies the cause of the abnormal line loss, and formulates intelligent control measures for abnormal line losses. This intelligent management and control of abnormal line losses in medium and low voltage distribution networks can effectively improve the accuracy of line loss anomaly identification, which not only helps optimize the operating efficiency of the power system but also reduces unnecessary energy waste, thus providing strong support for the sustainable development of the power system. It enables refined management of distribution network line losses and improves the overall operating efficiency of the power grid. Attached Figure Description

[0074] Figure 1This is a flowchart of the intelligent diagnosis method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms, as described in this invention. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] To address the problem of insufficient detection accuracy and the inability to perform refined management of distribution network line losses, which reduces the overall operating efficiency of the power grid, please refer to [link to relevant documentation]. Figure 1 This embodiment provides the following technical solution:

[0077] The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms includes the following steps:

[0078] S1. Data Acquisition and Preprocessing: Collect voltage, current, power, and load data during the operation of medium and low voltage distribution networks, determine the operation data of medium and low voltage distribution networks, and preprocess the operation data of medium and low voltage distribution networks.

[0079] In this embodiment, the operating data of the medium and low voltage distribution network is determined, and the following operations are performed:

[0080] Utilize Internet of Things (IoT) technology to install intelligent data acquisition devices in medium and low voltage power distribution networks;

[0081] The voltage status of the medium and low voltage distribution network is monitored and collected in real time based on intelligent data acquisition equipment to obtain voltage data of the medium and low voltage distribution network.

[0082] Based on intelligent data acquisition equipment, the current status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain current data of medium and low voltage distribution networks;

[0083] Based on intelligent data acquisition equipment, the power status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain power data of medium and low voltage distribution networks;

[0084] Based on intelligent data acquisition equipment, the load status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain load data of medium and low voltage distribution networks;

[0085] Among them, based on the voltage data, current data, power data and load data of the medium and low voltage distribution network, the operation data of the medium and low voltage distribution network based on the Internet of Things is determined.

[0086] In this embodiment, the preprocessing of the medium- and low-voltage distribution network operation data includes:

[0087] Cleaning of IoT-based medium and low voltage power distribution network operation data;

[0088] Remove duplicate data, missing values, and outliers from the IoT-based operation data of medium and low voltage distribution networks that are not useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks;

[0089] For missing and outlier values ​​that are useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks, the median is used to fill in the missing values ​​and the average value is used to replace the outlier values.

[0090] Normalize the operation data of medium and low voltage distribution networks based on the Internet of Things;

[0091] The operation data of medium and low voltage distribution networks based on the Internet of Things (IoT) are converted into a unified format to remove the dimensional differences between the operation data of medium and low voltage distribution networks based on IoT, and to determine the standardized operation data of medium and low voltage distribution networks.

[0092] S2. Feature Extraction and Selection: Feature extraction is performed on the operation data of medium and low voltage distribution networks. Feature vectors related to line losses in medium and low voltage distribution networks are extracted, and the extracted feature vectors are selected based on the random forest algorithm to determine the operation feature data of medium and low voltage distribution networks.

[0093] In this embodiment, feature extraction and selection are performed on the operation data of medium and low voltage distribution networks, including:

[0094] Based on principal component analysis, static and dynamic features related to line losses in medium and low voltage distribution networks are extracted from the operation data of medium and low voltage distribution networks.

[0095] Static characteristics include line parameters, equipment parameters, and topology.

[0096] Line parameters include line length, conductor type, cross-sectional area, and resistivity; equipment parameters include transformer capacity, model, and load rate; topology includes network structure and node connection relationships.

[0097] The dynamic characteristics include electrical measurement data, load characteristics, environmental factors, and time characteristics.

[0098] Electrical measurement data includes voltage, current, active power, reactive power, and power factor; load characteristics include load factor, load fluctuation, peak-to-valley difference, and load curve; environmental factors include temperature, humidity, and weather conditions; and time characteristics include hourly, daily, weekly, monthly, and seasonal dimensions.

[0099] Based on the random forest algorithm, the extracted static and dynamic features are selected, the influence of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is evaluated, the most valuable features for intelligent diagnosis of line loss anomalies in medium and low voltage distribution networks are screened out, and the operating characteristic data of medium and low voltage distribution networks are determined.

[0100] In one embodiment, the selection of extracted static and dynamic features based on the random forest algorithm, and the evaluation of the impact of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks, includes:

[0101] A feature set is established based on all dynamic and static features. The first change range of all dynamic features relative to historical dynamic features is obtained, and the second change range of all static features relative to historical static features is obtained. The ratio of the first change range to the second change range is used as the random sampling ratio of dynamic and static features.

[0102] According to the random sampling ratio, multiple feature groups are obtained by randomly sampling features with replacement from the feature set. Based on each feature group, a decision tree is constructed by combining network line loss anomaly information. When splitting the nodes of the decision tree, feature group features are randomly selected to determine the split point.

[0103] Each feature in the feature set is input into each decision tree in turn, and the score set of the current feature in all decision trees is determined based on the decision results;

[0104] The current feature is permuted to obtain a permuted feature value. The permuted feature value is then input into each decision tree in sequence to obtain a set of performance degradation magnitudes for all decision trees. The scoring weight for the current feature is determined based on the difference between the average degradation value of the set of performance degradation magnitudes and a preset reference degradation value.

[0105] The target score value for the current feature is obtained by weighting the average score value determined by the score set based on the score weights.

[0106] The degree of influence of the current feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is determined based on the target score value.

[0107] In this embodiment, the higher the target score, the greater the degree of influence.

[0108] In this embodiment, randomly selecting feature groups to determine the split point when splitting nodes in the decision tree can increase the independence of the decision tree.

[0109] In this embodiment, the ratio of the first change amplitude to the second change amplitude is used as the random sampling ratio for dynamic and static features, ensuring the rationality of the allocation of the random extraction quantity of dynamic and static features, and providing a basis for ensuring the accuracy of the decision tree.

[0110] In this embodiment, when the average decrease value is less than the preset reference decrease value, the scoring weight is less than 1; when the average decrease value is greater than the preset reference decrease value, the scoring weight is greater than 1, and the greater the difference, the greater the scoring weight.

[0111] The beneficial effects of the above design scheme are as follows: By establishing a feature set based on all dynamic and static features, the first change range of all dynamic features relative to historical dynamic features is obtained, and the second change range of all static features relative to historical static features is obtained. The ratio of the first and second change ranges is used as the random sampling ratio for dynamic and static features to ensure the rationality of the random feature allocation, providing a foundation for ensuring the accuracy of the decision tree. Based on each feature group, a decision tree is constructed by combining network line loss anomaly information. When splitting nodes in the decision tree, features from the feature group are randomly selected to determine the split point, increasing the independence of the decision tree and providing a basis for multi-faceted feature scoring. Each feature in the feature set... The current feature is sequentially input into each decision tree, and the score set of the current feature in all decision trees is determined according to the decision results. The current feature is then subjected to feature value permutation to obtain a permuted feature value. The permuted feature value is sequentially input into each decision tree to obtain a set of performance degradation magnitudes for all decision trees. The score weight for the current feature is determined based on the difference between the average degradation value of the performance degradation magnitude set and a preset reference degradation value. The average score value determined by the score set is weighted based on the score weight to obtain the target score value of the current feature. By analyzing and integrating the two factors before and after permutation, it is ensured that the target score value of the corresponding feature can more accurately reflect the degree of influence on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks.

[0112] S3. Model Building and Training: Based on machine learning algorithms, the machine learning model is trained using historical data of medium and low voltage distribution networks, and the model parameters are optimized to determine the optimal intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks.

[0113] In this embodiment, the optimal intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks is determined, and the following operations are performed:

[0114] Collect historical data on the operation of medium and low voltage distribution networks, and divide the collected historical data to determine the training set and test set;

[0115] Based on machine learning algorithms, a training set is used to train the machine learning model, enabling the machine learning model to autonomously learn the intelligent diagnostic behavior of abnormal line losses in medium and low voltage distribution networks, and to determine the intelligent diagnostic model of abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms.

[0116] Based on the test set, the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms was tested. The accuracy, recall and F1 score were used to evaluate whether the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms can achieve the expected results.

[0117] Based on the evaluation results, the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms were adjusted. Through continuous iterative optimization, the optimal intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks was determined.

[0118] S4. Fault Prediction and Health Management: Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the system analyzes the operational characteristic data of medium and low voltage distribution networks, diagnoses abnormal line losses, locates the abnormal line loss locations, identifies the causes of abnormal line losses, formulates intelligent control measures for abnormal line losses, and implements intelligent control of abnormal line losses in medium and low voltage distribution networks to reduce line losses in medium and low voltage distribution networks.

[0119] In this embodiment, the operational characteristic data of medium and low voltage distribution networks are analyzed, including:

[0120] Deploy the optimal intelligent diagnostic model for line loss anomalies in medium and low voltage distribution networks in a real intelligent diagnostic environment for line loss anomalies in medium and low voltage distribution networks.

[0121] The operational characteristic data of medium and low voltage distribution networks are input into the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks. Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed, and abnormal line losses in medium and low voltage distribution networks are diagnosed intelligently to determine the predicted values ​​of line losses in medium and low voltage distribution networks.

[0122] In this embodiment, abnormal diagnosis of line losses in medium and low voltage distribution networks includes:

[0123] The predicted line loss value is compared with the actual line loss value, the relative error is calculated, and the relative error is used to determine whether there is an abnormal line loss in the medium and low voltage distribution network.

[0124] in,

[0125] When the relative error is within 5%, the line loss of the medium and low voltage distribution network is considered to be in a normal state.

[0126] When the relative error exceeds 5%, the line loss of the medium and low voltage distribution network is determined to be in an abnormal state, triggering a line loss abnormality alarm and locating the abnormal line loss location.

[0127] Obtain topology and equipment information of medium and low voltage distribution network lines, and gradually narrow down the anomaly range to locate the specific line segment or equipment using the topology and equipment information;

[0128] Among them, obtaining the topology information of medium and low voltage distribution network lines includes the connection relationship of the lines, the length of the lines, the conductor type, the impedance parameters, and the location and status of the equipment;

[0129] A zonal analysis of the medium and low voltage distribution network is performed, dividing the medium and low voltage distribution network into several regions. Line loss anomalies are analyzed in each region, and the predicted and actual line loss values ​​of each region are calculated to identify the region with the largest deviation.

[0130] Check the status of equipment in areas with abnormal line loss, including: whether the transformer is overloaded or faulty, whether the switch is closed or opened normally, whether the capacitor is working normally, and determine whether there is any abnormality in the equipment through equipment monitoring data;

[0131] Analyze the line parameters and check the line parameters in areas with abnormal line loss, including: whether the line impedance is abnormal, whether there is three-phase imbalance or harmonic problems, and analyze the line operation status through electrical measurement data.

[0132] If the equipment and line parameters are normal, there may be electricity theft. In this case, the difference between the user's electricity consumption and historical data is compared, and an electricity theft detection algorithm is used in conjunction with on-site inspections to check whether the user's electricity meter has been tampered with.

[0133] By using topology and equipment information, the scope of the anomaly is gradually narrowed down to pinpoint the specific line segment or equipment. When the line loss of a feeder is abnormal, the various branches under that feeder are analyzed. When the line loss of a transformer is abnormal, the power consumption of users under that transformer is checked.

[0134] In this embodiment, intelligent management and control of abnormal line losses in medium and low voltage distribution networks includes:

[0135] The intelligent diagnostic results of abnormal line losses in medium and low voltage distribution networks are analyzed. Combined with historical data and on-site inspections, the causes of abnormal line losses are identified, including equipment failure, line problems, electricity theft, operational issues, and management problems. Based on the causes of abnormal line losses, intelligent control measures for reducing line losses in medium and low voltage distribution networks are formulated, and intelligent control of abnormal line losses in medium and low voltage distribution networks is implemented.

[0136] In response to equipment failures, the measures include replacing or repairing faulty equipment, upgrading equipment, and conducting regular inspections and maintenance, including replacing overloaded transformers, repairing switches with poor contact, adopting high-efficiency and energy-saving equipment, and establishing an equipment inspection system to promptly identify and address potential failures.

[0137] For line problems, the measures include replacing aging lines, optimizing line layout, and strengthening joint maintenance, including using high conductivity, low loss conductors, reducing line length, lowering line impedance, and regularly checking line joints to ensure good contact.

[0138] In response to electricity theft, measures include strengthening electricity inspections, using electricity theft detection technologies and legal means, including real-time monitoring of users' electricity consumption behavior through smart meters and electricity information collection systems, using machine learning algorithms to detect abnormal electricity consumption behavior, severely cracking down on electricity theft, and increasing the cost of illegal activities.

[0139] In response to operational issues, adjustments are made to the three-phase balance, harmonic mitigation is implemented, and load management is carried out. This includes reducing three-phase imbalance by adjusting the load or installing three-phase balancing devices, reducing harmonic pollution by installing filters or reactive power compensation devices, optimizing load distribution, and avoiding local overload.

[0140] In response to management issues, measures include improving the accuracy of data collection and transmission, reducing measurement errors, establishing a line loss assessment mechanism to incentivize maintenance personnel to reduce losses and increase efficiency, and utilizing big data, artificial intelligence, and other technologies to achieve real-time monitoring and intelligent diagnosis of line losses.

[0141] In one embodiment, adjusting the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium- and low-voltage distribution networks based on machine learning algorithms, according to the evaluation results, includes:

[0142] The types and distributions of line loss anomalies are determined from the evaluation results, and the anomaly weights of the anomaly types are determined based on the types and distributions of line loss anomalies.

[0143] Relevant historical operation data related to anomaly types are obtained from the power grid operation history data, and the importance of the current relevant historical operation data is determined based on the anomaly weight of the anomaly type.

[0144] The formula for calculating the importance K of the current relevant historical operational data is as follows:

[0145]

[0146] Where C represents a constant with a value of 3, R represents the amount of relevant historical operational data, R0 represents the amount of reference data, n represents the number of anomaly types, and δ i G represents the anomaly weight of the i-th anomaly type. iThis represents the total amount of historical data related to the i-th anomaly type.

[0147] Based on the importance of each relevant historical operation data, weighted data extraction is performed on the historical power grid operation data to obtain a new training set and a new validation set.

[0148] The mean square error, error rate, and F1 score for the line loss anomaly assessment are determined from the assessment results. Based on the mean square error, error rate, and F1 score, the adjustment coefficients for the model structure weights are determined.

[0149] The formula for calculating the adjustment coefficient γ is as follows:

[0150]

[0151] Where F1 represents the F1 score, T represents the error rate, and σ 2 Mean squared error, where X represents the overall performance of the model, and e represents the natural constant, with a value of 2.72;

[0152] The latest weights for the model structure are determined based on the adjustment coefficients.

[0153] Based on the new training set, the new validation set, and the latest weights, the intelligent diagnostic model for line loss anomalies in low-voltage distribution networks is retrained and adjusted to obtain the latest intelligent diagnostic model for line loss anomalies.

[0154] In this embodiment, the denser the distribution of line loss anomaly types, the greater the corresponding anomaly weight.

[0155] In this embodiment, the greater the importance of the current relevant historical running data, the more important the data is for diagnosis, and the greater the extraction ratio when extracting the dataset and validation set.

[0156] In this embodiment, X is greater than If X is not greater than This indicates that no adjustment to the model structure weights is required.

[0157] In this embodiment, the larger the adjustment coefficient, the greater the adjustment range of the structural weights.

[0158] The beneficial effects of the above design scheme are as follows: By determining the type and distribution of line loss anomalies from the evaluation results, and determining the anomaly weights for each anomaly type based on these characteristics; by obtaining relevant historical operational data related to the anomaly type from the power grid's historical operational data, and determining the data importance of the current relevant historical operational data based on the anomaly weights for each anomaly type; by extracting weighted data from the power grid's historical operational data based on the data importance of each relevant historical operational data point, obtaining a new training set and a new validation set, increasing the proportion of operational data related to recurring line loss anomalies, and enabling the trained model to have a more accurate ability to identify and judge recurring line loss anomalies; then, by determining the mean square error, error rate, and F1 score of the line loss anomaly evaluation from the evaluation results, and by determining the adjustment coefficients for the model structure weights based on these mean square error, error rate, and F1 score, the model weights are adjusted based on the overall performance of the model, improving the performance of the latest intelligent diagnostic model for line loss anomalies, and providing a highly accurate model for intelligent diagnosis of line loss anomalies.

[0159] In summary, based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed to diagnose abnormal line losses. Combining line topology and equipment information, the location of abnormal line losses is determined, the causes are identified, and intelligent control measures for abnormal line losses are formulated. This intelligent management of abnormal line losses in medium and low voltage distribution networks effectively improves the accuracy of line loss anomaly identification. This not only helps optimize the operating efficiency of the power system but also reduces unnecessary energy waste, thus providing strong support for the sustainable development of the power system. It enables refined management of distribution network line losses and improves the overall operating efficiency of the power grid.

[0160] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0161] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent diagnosis of low-voltage power distribution network line loss anomaly based on a machine learning algorithm, characterized in that, include: Collect and preprocess operational data from medium- and low-voltage power distribution networks; Feature extraction is performed on the operation data of medium and low voltage distribution networks. Based on the target score value, the influence of the current feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is determined, and the operation feature data of medium and low voltage distribution networks are determined. Based on machine learning algorithms, an intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks is constructed. Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed and abnormal line losses are diagnosed. Combined with the line topology and equipment information, the location of abnormal line losses is located, and intelligent control is carried out on the abnormal line losses in medium and low voltage distribution networks. Construct an intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks and perform the following operations: Collect historical data on the operation of medium and low voltage distribution networks, and divide the collected historical data to determine the training set and test set; Based on machine learning algorithms, a training set is used to train the machine learning model, enabling the machine learning model to autonomously learn the intelligent diagnostic behavior of abnormal line losses in medium and low voltage distribution networks, and to determine the intelligent diagnostic model of abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms. Based on the test set, the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms was tested. The accuracy, recall and F1 score were used to evaluate whether the intelligent diagnostic model for abnormal line loss in medium and low voltage distribution networks based on machine learning algorithms can achieve the expected results. Based on the evaluation results, the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms were adjusted. Through continuous iterative optimization, the optimal intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks was determined. Based on the evaluation results, the parameters and structure of the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms were adjusted, including: The types and distributions of line loss anomalies are determined from the evaluation results, and the anomaly weights of the anomaly types are determined based on the types and distributions of line loss anomalies. Relevant historical operation data related to anomaly types are obtained from the power grid operation history data, and the importance of the current relevant historical operation data is determined based on the anomaly weight of the anomaly type. The importance of current relevant historical operational data The calculation formula is as follows: ； in, This represents a constant with a value of 3. This represents the amount of relevant historical operational data currently available. This represents the amount of reference data, and n represents the number of exception types. This represents the anomaly weight of the i-th anomaly type. This represents the total amount of historical data related to the i-th anomaly type. Based on the importance of each relevant historical operation data, weighted data extraction is performed on the historical power grid operation data to obtain a new training set and a new validation set. The mean square error, error rate, and F1 score for the line loss anomaly assessment are determined from the assessment results. Based on the mean square error, error rate, and F1 score, the adjustment coefficients for the model structure weights are determined. The adjustment coefficient The calculation formula is as follows: ； in, Indicates the F1 score. Indicates the error rate. Mean square error, This represents the overall performance value of the model, where e represents the natural constant, with a value of 2.

72. The latest weights for the model structure are determined based on the adjustment coefficients. Based on the new training set, the new validation set, and the latest weights, the intelligent diagnostic model for line loss anomalies in low-voltage distribution networks is retrained and adjusted to obtain the latest intelligent diagnostic model for line loss anomalies.

2. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 1, characterized in that, Collect operational data from medium and low voltage power distribution networks and perform the following operations: Utilize Internet of Things (IoT) technology to install intelligent data acquisition devices in medium and low voltage power distribution networks; The voltage status of the medium and low voltage distribution network is monitored and collected in real time based on intelligent data acquisition equipment to obtain voltage data of the medium and low voltage distribution network. Based on intelligent data acquisition equipment, the current status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain current data of medium and low voltage distribution networks; Based on intelligent data acquisition equipment, the power status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain power data of medium and low voltage distribution networks; Based on intelligent data acquisition equipment, the load status of medium and low voltage distribution networks is monitored and collected in real time during operation to obtain load data of medium and low voltage distribution networks; Among them, based on the voltage data, current data, power data and load data of the medium and low voltage distribution network, the operation data of the medium and low voltage distribution network based on the Internet of Things is determined.

3. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 2, characterized in that, Preprocessing of operational data from medium- and low-voltage distribution networks includes: Cleaning of IoT-based medium and low voltage power distribution network operation data; Remove duplicate data, missing values, and outliers from the IoT-based operation data of medium and low voltage distribution networks that are not useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks; For missing and outlier values ​​that are useful for intelligent diagnosis of abnormal line losses in medium and low voltage distribution networks, the median is used to fill in the missing values ​​and the average value is used to replace the outlier values. Normalize the operation data of medium and low voltage distribution networks based on the Internet of Things; The operation data of medium and low voltage distribution networks based on the Internet of Things (IoT) are converted into a unified format to remove the dimensional differences between the operation data of medium and low voltage distribution networks based on IoT, and to determine the standardized operation data of medium and low voltage distribution networks.

4. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 3, characterized in that, Feature extraction is performed on the operation data of medium and low voltage distribution networks, including: Based on principal component analysis, static and dynamic features related to line losses in medium and low voltage distribution networks are extracted from the operation data of medium and low voltage distribution networks. Static characteristics include line parameters, equipment parameters, and topology. Line parameters include line length, conductor type, cross-sectional area, and resistivity; equipment parameters include transformer capacity, model, and load rate; topology includes network structure and node connection relationships. The dynamic characteristics include electrical measurement data, load characteristics, environmental factors, and time characteristics. Electrical measurement data includes voltage, current, active power, reactive power, and power factor; load characteristics include load factor, load fluctuation, peak-to-valley difference, and load curve; environmental factors include temperature, humidity, and weather conditions; and time characteristics include hourly, daily, weekly, monthly, and seasonal dimensions. Based on the random forest algorithm, the extracted static and dynamic features are selected, the influence of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is evaluated, the most valuable features for intelligent diagnosis of line loss anomalies in medium and low voltage distribution networks are screened out, and the operating characteristic data of medium and low voltage distribution networks are determined.

5. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 4, characterized in that, The random forest algorithm is used to select extracted static and dynamic features and evaluate the impact of each feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks, including: A feature set is established based on all dynamic and static features. The first change range of all dynamic features relative to historical dynamic features is obtained, and the second change range of all static features relative to historical static features is obtained. The ratio of the first change range to the second change range is used as the random sampling ratio of dynamic and static features. According to the random sampling ratio, multiple feature groups are obtained by randomly sampling features with replacement from the feature set. Based on each feature group, a decision tree is constructed by combining network line loss anomaly information. When splitting the nodes of the decision tree, feature group features are randomly selected to determine the split point. Each feature in the feature set is input into each decision tree in turn, and the score set of the current feature in all decision trees is determined based on the decision results; The current feature is permuted to obtain a permuted feature value. The permuted feature value is then input into each decision tree in sequence to obtain a set of performance degradation magnitudes for all decision trees. The scoring weight for the current feature is determined based on the difference between the average degradation value of the set of performance degradation magnitudes and a preset reference degradation value. The target score value for the current feature is obtained by weighting the average score value determined by the score set based on the score weights. The degree of influence of the current feature on the intelligent diagnosis and prediction results of line loss anomalies in medium and low voltage distribution networks is determined based on the target score value.

6. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 5, characterized in that, Analysis of operational characteristic data of medium and low voltage distribution networks, including: Deploy the optimal intelligent diagnostic model for line loss anomalies in medium and low voltage distribution networks in a real intelligent diagnostic environment for line loss anomalies in medium and low voltage distribution networks. The operational characteristic data of medium and low voltage distribution networks are input into the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks. Based on the intelligent diagnostic model for abnormal line losses in medium and low voltage distribution networks, the operational characteristic data of medium and low voltage distribution networks are analyzed, and abnormal line losses in medium and low voltage distribution networks are diagnosed intelligently to determine the predicted values ​​of line losses in medium and low voltage distribution networks.

7. The intelligent diagnostic method for abnormal line losses in medium and low voltage distribution networks based on machine learning algorithms as described in claim 6, characterized in that, Anomaly diagnosis of line losses in medium and low voltage distribution networks includes: The predicted line loss value is compared with the actual line loss value, the relative error is calculated, and the relative error is used to determine whether there is an abnormal line loss in the medium and low voltage distribution network. Wherein, relative error = ; When the relative error is within 5%, the line loss of the medium and low voltage distribution network is considered to be in a normal state. When the relative error exceeds 5%, the line loss of the medium and low voltage distribution network is determined to be in an abnormal state, triggering a line loss abnormality alarm and locating the abnormal line loss location. Obtain topology and equipment information of medium and low voltage distribution network lines, and gradually narrow down the anomaly range to locate the specific line segment or equipment using the topology and equipment information; Among them, obtaining the topology information of medium and low voltage distribution network lines includes the connection relationship of the lines, the length of the lines, the conductor type, the impedance parameters, and the location and status of the equipment; A zonal analysis of the medium and low voltage distribution network is performed, dividing the medium and low voltage distribution network into several regions. Line loss anomalies are analyzed in each region, and the predicted and actual line loss values ​​of each region are calculated to identify the region with the largest deviation. Check the status of equipment in areas with abnormal line loss, including: whether the transformer is overloaded or faulty, whether the switch is closed or opened normally, whether the capacitor is working normally, and determine whether there is any abnormality in the equipment through equipment monitoring data; Analyze the line parameters and check the line parameters in areas with abnormal line loss, including: whether the line impedance is abnormal, whether there is three-phase imbalance or harmonic problems, and analyze the line operation status through electrical measurement data. If the equipment and line parameters are normal, there may be electricity theft. In this case, the difference between the user's electricity consumption and historical data is compared, and an electricity theft detection algorithm is used in conjunction with on-site inspections to check whether the user's electricity meter has been tampered with. By using topology and equipment information, the scope of the anomaly is gradually narrowed down to pinpoint the specific line segment or equipment. When the line loss of a feeder is abnormal, the various branches under that feeder are analyzed. When the line loss of a transformer is abnormal, the power consumption of users under that transformer is checked.

Images