Intelligent analysis method for line loss abnormity of transformer area

Abstract

The invention belongs to the field of transformer area line loss detection, and particularly discloses a transformer area line loss abnormity intelligent analysis method, which comprises the following steps: acquiring basic data; constructing and training an artificial intelligence line loss analysis model; performing anomaly detection and early warning; and a manager adjusts and optimizes a power supply strategy according to the pushed line loss analysis detection result. According to the method, the deep analysis of the line loss of the transformer area is carried out based on the artificial intelligence technology, the line loss analysis model is established by using the artificial intelligence technology, and the line loss problem is rapidly analyzed and positioned, so that the line loss analysis efficiency is improved, the manual intervention and the operation cost are reduced, and more accurate management and decision support are provided for power supply management; the labor cost of power supply line loss management can be reduced, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the field of line loss detection in a transformer area, and in particular relates to an intelligent analysis method for abnormal line loss in a transformer area. Background Art

[0002] The quality of the line loss index in the substation not only represents the waste of electric energy, but also directly reflects the operating status of a power supply station. The abnormal line loss problem in the substation seriously affects the economic operation and power supply quality of the power system. The traditional line loss analysis method is difficult to adapt to the increasingly complex power supply and use environment.

[0003] In order to make illegal profits, some unscrupulous users privately access the power supply, short-circuit the energy meter, modify the electricity stealing device, etc., which seriously infringes the legitimate rights and interests of the power supply company; the energy metering device is the key to accurately measure the user's electricity consumption, and its abnormality will directly lead to the distortion of the line loss in the substation. By regularly carrying out equipment status maintenance and timely eliminating equipment defects, the level of line loss can be effectively reduced. Due to the complex and diverse field environment and the large number of application scenarios, traditional detection methods are costly and require large investment. In recent years, as low-voltage collection data has been directly expanded from daily single-point indications to full data such as current, voltage, and power at 24-hour nodes, in-depth analysis of substation line loss has become possible. But at the same time, new problems are faced. For such a large amount of data, how to quickly mine and accurately locate anomalies has become a new problem that must be faced. With the development of smart grids and big data technologies, it is urgent to develop intelligent line loss anomaly detection methods to achieve real-time monitoring and accurate diagnosis of the operation status of substations. Summary of the invention

[0004] In order to address the shortcomings of the prior art, the present invention aims to provide an intelligent analysis method for line loss anomalies in a substation. The method conducts in-depth analysis of line losses in the substation based on artificial intelligence technology, establishes a line loss analysis model using artificial intelligence technology, and quickly analyzes and locates line loss problems, thereby improving the efficiency of line loss analysis, reducing manual intervention and operating costs, and providing more accurate management and decision-making support for power supply management, which can reduce the manpower cost of power supply line loss management and improve work efficiency.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: A method for intelligently analyzing abnormal line loss in a transformer area is implemented based on the following steps: S1. Collection of basic data: sort out the line loss related data of the power supply users in the power supply area, and clean, pre-process and merge the data to form a basic database; S2. Construction and training of artificial intelligence line loss analysis model: cluster analysis is performed on the collected basic data to form different power consumption models and establish a reference power consumption model library; neural networks are used to construct line loss analysis model data to express the logical relationship between data and line loss, and to achieve abnormal line loss detection, analysis and location; the constructed line loss analysis model is then trained to fit historical line loss data, and the model is optimized using intrinsic functions and optimization algorithms; S3. Abnormal detection and early warning: Use the artificial intelligence line loss analysis model to intelligently analyze line loss data and related data, timely discover abnormal data and trigger early warning notifications, warn and locate abnormal line losses, and form a detection report, and push the results to management personnel; S4. Decision support based on results: Managers can adjust and optimize power supply strategies based on the pushed line loss analysis and test results.

[0006] Furthermore, the substation line loss related data collected in step S1 include: substation basic data, electricity consumption and electricity fee data, meter reading operation data, line loss calculation data and meteorological environment data; the substation basic data include substation number, name, power supply range, capacity, voltage level, and topological structure; the electricity consumption and electricity fee data include the forward and reverse active / reactive electric energy of each substation gateway meter, branch box meter, and user electric energy meter, as well as the corresponding electricity fee information; the meter reading operation data include the operating conditions of the collector and the electric energy meter, and the operating conditions of the electric energy meter include the online rate, data reporting rate, and power outage events; the line loss calculation data include the power supply, electricity sales, loss, and line loss rate summarized monthly by substation; the meteorological environment data include the weather conditions, temperature and humidity of the substation location.

[0007] Furthermore, the establishment and optimization of the line loss artificial intelligence line loss analysis model in step S2 specifically includes: S2-1: Obtain the line loss data of the substation area and the state variables that affect the line loss, and perform data preprocessing; S2-2: Determine the prediction model parameters: Determine the topology of the neural network based on the BP neural network input and output data sets; S2-3: Initialization of PSO algorithm: Initialize the number, speed, position, inertia weight and maximum number of iterations of the particles according to the formula; S2-4: Evaluation: Based on the input and output data sets, the average absolute error between the actual area line loss data and the energy consumption prediction value of the neural network is used to update the individual and group optimal positions of the PSO algorithm in the hybrid prediction model according to the fitness function; S2-5: Iteration: Update the velocity and position of the particle, and re-constrain the particles that cross the velocity and position definition domain to the preset boundary; S2-6: Check: determine whether the current number of iterations reaches the preset value; if so, stop the iteration and output the optimal solution, which is the connection weight and threshold of the trained neural network; otherwise, return to step S2-4 and continue execution; S2-7: Substation line loss analysis: After the neural network structure parameter training is completed, the optimal connection weights and thresholds are assigned to the BP neural network, and then the substation line loss analysis is performed.

[0008] Furthermore, the number of particles during initialization in step S2-3 is determined by the topological structure of the BP neural network, that is, the number of connection weights and thresholds between layers.

[0009] Furthermore, step S2 also includes testing of the artificial intelligence line loss analysis model, specifically including: Data input testing: Verify whether the data can correctly receive and process various types of data; Data processing test: Verify whether the data processing module can correctly extract line loss related features and perform corresponding analysis; Model training and optimization test: Verify whether the model training and optimization modules can automatically adjust the model parameters according to the line loss data of the power supply station; Early warning function test: Verify whether the early warning module can trigger corresponding early warning prompts according to the prediction results and accurately report abnormal line loss conditions.

[0010] The data input test includes testing the processing capabilities of different data formats and data volumes.

[0011] The data processing test includes testing the ability to handle outliers, missing values ​​and large data sets.

[0012] The model training and optimization test includes testing of the training and verification effects on different line loss data sets.

[0013] The early warning function test includes testing of early warning effects under different early warning thresholds and data anomalies.

[0014] The beneficial effects of the present invention are: 1. The present invention uses advanced machine learning algorithms and artificial intelligence technology to automatically analyze data and automatically trigger early warning prompts, reducing manual intervention and operating costs, and providing more accurate management and decision support for power supply management. At the same time, through automated and intelligent processing methods, it can reduce the labor cost of line loss management in power supply stations and improve work efficiency.

[0015] 2. The power supply line loss analysis technology based on artificial intelligence of the present invention has outstanding advantages in comprehensive application of artificial intelligence technology, intelligent line loss analysis and management, ease of use and scalability, and can quickly identify the cause and location of line loss. Based on historical data and real-time data, and proposing optimization plans, it helps power supply management take measures in advance to reduce line loss rates, while reducing manual intervention and improving efficiency and accuracy.

[0016] 3. The present invention uses artificial intelligence technology to conduct in-depth analysis of power supply line loss data, quickly and accurately identify the causes of line loss, including abnormal connection, user electricity theft, photovoltaic abnormalities, etc., to help power supply stations take targeted measures to reduce line loss rates, help them optimize line loss management and operation strategies, reduce costs and improve efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0018] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not intended to limit the scope of use of the present invention.

[0019] like Figure 1 As shown, the present invention proposes an intelligent analysis method for abnormal line loss in the substation area, which uses advanced machine learning algorithms and artificial intelligence technology to automatically perform data analysis and automatically trigger early warning prompts, reducing manual intervention and operating costs, and providing more accurate management and decision support for power supply management. At the same time, through automated and intelligent processing methods, it can reduce the labor cost of line loss management in power supply stations and improve work efficiency. Specifically including: Step S1. Collection of basic data: sort out the line loss related data of the power supply substation and user substation, and clean, pre-process and merge the data to form a basic database.

[0020] The data collected by the present invention may be the user's data related to line loss in the substation for the past three years, including: basic substation data, electricity consumption and electricity fee data, meter reading operation data, line loss calculation data and meteorological environment data, etc. Among them, the basic substation data includes the substation number, name, power supply range, capacity, voltage level, topology structure, etc. The electricity consumption and electricity fee data include the forward and reverse active / reactive electric energy of each substation gateway meter, branch box meter, and user electric energy meter, as well as the corresponding electricity fee information; the time granularity is 15 minutes. The meter reading operation data includes the operating conditions of the collector and the electric energy meter, such as the online rate, data reporting rate, power outage events, etc. The line loss calculation data includes the power supply, electricity sales, loss of electricity, line loss rate, etc. summarized monthly by substation. The meteorological environment data includes the weather conditions, temperature and humidity of the substation location.

[0021] After pre-processing such as data cleaning and data integration, the above data is aligned according to timestamps and station IDs, eventually forming a database of about 65 million structured records with a total data volume of 1.2TB.

[0022] Step S2. Construction and training of artificial intelligence line loss analysis model: The analysis model is established on the basis of line loss data, taking into account force majeure factors such as data anomalies to construct a line loss artificial intelligence model.

[0023] Genetic algorithm and particle swarm optimization algorithm are typical swarm intelligence algorithms and are widely used in the fields of function optimization, artificial neural network learning and training, fuzzy system control, pattern recognition, etc. Since both particle swarm optimization algorithm and genetic algorithm have good global optimization and convergence performance and are easy to integrate with other algorithms, and artificial neural network has strong local search performance, the present invention is based on the artificial neural network line loss artificial intelligence model.

[0024] The present invention performs cluster analysis on the collected basic data to form different power consumption models and establish a reference power consumption model library; and uses a neural network to construct the line loss analysis model data, express the logical relationship between the data and the line loss, and realize abnormal line loss detection, analysis and positioning; and then trains the constructed line loss analysis model to fit the historical line loss data, and optimizes the model using intrinsic functions and optimization algorithms. Specifically, it includes: S2-1: Obtain the line loss data of the substation area and the state variables that affect the line loss, and perform data preprocessing; S2-2: Determine the prediction model parameters: Determine the topology of the neural network based on the BP neural network input and output data sets; S2-3: Initialization of PSO algorithm: Initialize the number, speed, position, inertia weight and maximum number of iterations of the particles according to the formula; S2-4: Evaluation: Based on the input and output data sets, the average absolute error between the actual area line loss data and the energy consumption prediction value of the neural network is used to update the individual and group optimal positions of the PSO algorithm in the hybrid prediction model according to the determined fitness function; S2-5: Iteration: Update the velocity and position of the particle, and re-constrain the particles that cross the velocity and position definition domain to the preset boundary; S2-6: Check: determine whether the current number of iterations reaches the preset value; if so, stop the iteration and output the optimal solution, which is the connection weight and threshold of the trained neural network; otherwise, return to step S2-4 and continue execution; S2-7: Substation line loss analysis: After the neural network structure parameter training is completed, the optimal connection weights and thresholds are assigned to the BP neural network, and then the substation line loss analysis is performed.

[0025] The present invention uses the PSO optimization algorithm to continuously optimize the connection weights and thresholds of each processing layer of the ANN until the fitness value of the PSO algorithm no longer changes meaningfully. Finally, the global optimal position Pbest obtained by the PSO algorithm is the connection weights and thresholds of each layer of the ANN that have been optimized. This process helps the BP neural network overcome its own defects such as slow convergence speed and easy to fall into the local optimal solution, give full play to the strong generalization ability and nonlinear mapping ability of the neural network, improve the convergence speed and self-learning and adaptive ability of the neural network, and effectively improve the accuracy of line loss calculation.

[0026] In addition, after the artificial intelligence line loss analysis model is established, the model needs to be tested, including: Data input testing: Verify that the data can correctly receive and process various types of data; including testing the processing capabilities of different data formats and data volumes.

[0027] Data processing test: Verify whether the data processing module can correctly extract line loss related features and perform corresponding analysis; including testing the ability to handle outliers, missing values ​​and large data sets.

[0028] Model training and optimization test: Verify whether the model training and optimization module can automatically adjust the model parameters according to the line loss data of the power supply station to improve the prediction accuracy; including testing the training and verification effects of different line loss data sets.

[0029] Early warning function test: Verify whether the early warning module can trigger corresponding early warning prompts according to the prediction results and accurately report abnormal line loss conditions; including testing the early warning effects under different early warning thresholds and data anomalies.

[0030] Step S3. Abnormal detection and early warning: Use the artificial intelligence line loss analysis model to intelligently analyze the line loss data and related data, timely detect abnormal data and trigger early warning notifications, warn and locate abnormal line losses, and form a detection report, and push the results to management personnel.

[0031] Step S4. Decision support of results: Provide technical support and decision support for management personnel based on the line loss analysis results. Provide decision support for power supply stations based on the line loss prediction results to adjust and optimize power supply strategies. In addition, visual interfaces and reports can be used to facilitate decision analysis and scheduling management for relevant personnel.

[0032] The present invention realizes accurate positioning of abnormal line losses in substations through accurate analysis of the artificial intelligence line loss analysis model, guides the management of power supply line losses, and improves the qualified rate of line losses. It can timely discover users who default on electricity theft and reduce power losses; it can also improve the power supply quality of power supply stations, reduce the risk of power grid operation, reduce power failures and power outages caused by line losses, and reduce related maintenance and loss costs. By analyzing the abnormalities of connection and collection discovered, work order processing and on-site maintenance are carried out in a timely manner to improve the business management capabilities of power supply stations. At the same time, the improvement of the qualified rate of line losses can improve power supply reliability, ensure users' electricity needs, reduce energy waste, and help reduce carbon dioxide emissions and environmental pollution.

[0033] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for intelligent analysis of abnormal line loss in a transformer area, characterized by: The method is implemented based on the following steps: S1. Collection of basic data: sort out the line loss data of power supply and user areas, clean, pre-process and merge the data to form a basic database; S2. Construction and training of artificial intelligence line loss analysis model: cluster analysis is performed on the collected basic data to form different power consumption models and establish a reference power consumption model library; neural networks are used to construct line loss analysis model data to express the logical relationship between data and line loss, and to achieve abnormal line loss detection, analysis and location; the constructed line loss analysis model is then trained to fit historical line loss data, and the model is optimized using intrinsic functions and optimization algorithms; S3. Abnormal detection and early warning: Use the artificial intelligence line loss analysis model to intelligently analyze line loss data and related data, timely discover abnormal data and trigger early warning notifications, warn and locate abnormal line losses, and form a detection report, and push the results to management personnel; S4. Decision support based on results: Managers can adjust and optimize power supply strategies based on the pushed line loss analysis and test results.

2. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 1, characterized in that: The data related to the line loss in the substation area collected in step S1 include: basic data of the substation area, electricity consumption and electricity fee data, meter reading operation data, line loss calculation data and meteorological environment data.

3. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 2 is characterized in that: The basic data of the substation area includes the substation area number, name, power supply range, capacity, voltage level, and topological structure; The electricity charge data includes the forward and reverse active / reactive electric energy of each area gateway meter, branch box meter, and user electric energy meter, as well as the corresponding electricity charge information; The meter reading operation data includes the operation conditions of the collector and the electric energy meter, and the operation conditions of the electric energy meter include the online rate, the data reporting rate, and the power outage event; The line loss calculation data includes the power supply, power sales, power loss and line loss rate summarized monthly by substation area; The meteorological environment data include weather conditions, temperature and humidity at the station location.

4. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 1, characterized in that: The establishment and optimization of the line loss artificial intelligence line loss analysis model in step S2 specifically includes: S2-1: Obtain the line loss data of the substation area and the state variables that affect the line loss, and perform data preprocessing; S2-2: Determine the prediction model parameters: Determine the topology of the neural network based on the BP neural network input and output data sets; S2-3: Initialization of PSO algorithm: Initialize the number, speed, position, inertia weight and maximum number of iterations of the particles according to the formula; S2-4: Evaluation: Based on the input and output data sets, the average absolute error between the actual area line loss data and the energy consumption prediction value of the neural network is used to update the individual and group optimal positions of the PSO algorithm in the hybrid prediction model according to the fitness function; S2-5: Iteration: Update the velocity and position of the particle, and re-constrain the particles that cross the velocity and position definition domain to the preset boundary; S2-6: Check: determine whether the current number of iterations reaches the preset value; if so, stop the iteration and output the optimal solution, which is the connection weight and threshold of the trained neural network; otherwise, return to step S2-4 and continue execution; S2-7: Substation line loss analysis: After the neural network structure parameter training is completed, the optimal connection weights and thresholds are assigned to the BP neural network, and then the substation line loss analysis is performed.

5. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 4 is characterized in that: The number of particles during initialization in step S2-3 is determined by the topological structure of the BP neural network, that is, the number of connection weights and thresholds between layers.

6. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 4, characterized in that: Step S2 also includes testing of the artificial intelligence line loss analysis model, specifically including: Data input testing: Verify whether the data can correctly receive and process various types of data; Data processing test: Verify whether the data processing module can correctly extract line loss related features and perform corresponding analysis; Model training and optimization test: Verify whether the model training and optimization modules can automatically adjust the model parameters according to the line loss data of the power supply station; Early warning function test: Verify whether the early warning module can trigger corresponding early warning prompts according to the prediction results and accurately report abnormal line loss conditions.

7. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 6, characterized in that: The data input test includes testing the processing capabilities of different data formats and data volumes.

8. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 6, characterized in that: The data processing test includes testing the ability to handle outliers, missing values ​​and large data sets.

9. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 6, characterized in that: The model training and optimization test includes testing of the training and verification effects on different line loss data sets.

10. The method for intelligent analysis of abnormal line loss in a transformer area according to claim 6, characterized in that: The early warning function test includes testing of early warning effects under different early warning thresholds and data anomalies.

Images