Multi-feature electricity stealing detection method based on combination of software and hardware

Abstract

The invention discloses a multi-feature electricity larceny detection method based on combination of software and hardware in the technical field of power systems, and aims to improve the real-time performance, accuracy and economical efficiency of electricity larceny detection. According to the method, an electricity larceny detection device main body and an infrared meter reader are arranged on a power transmission line and a user electricity meter, and key data such as voltage, current, power factors and electricity consumption are collected in real time. And performing deep analysis on the data by using the Catch22 feature set and the supplementary features in combination with the convolutional neural network and the Transform network, thereby realizing accurate identification of the electricity stealing behavior. The modular design and low-cost material selection of the device enable the scheme to be easy to expand, economical and efficient. In addition, the scheme further comprises flexible setting of a data acquisition period and simple installation of the device, and the practicability of the device is further enhanced. Through the method, an electric power company can greatly improve the efficiency and accuracy of electricity stealing detection, power grid assets are effectively protected, and the safety and stability of electric power supply are ensured.

Description

Technical Field

[0001] The invention relates to the technical field of power systems, and in particular to a multi-feature electricity theft detection method based on the combination of software and hardware. Background Art

[0002] With the rapid development of industrialization and informatization, electric power resources have become an indispensable basic resource in modern society. The stable operation of the power system is crucial to ensuring the normal order of the national economy and social life. However, electricity theft, as an illegal act of obtaining electric power resources, not only causes serious damage to the economic interests of power companies, but also may threaten the safe and stable operation of the power grid. According to statistics, the economic losses caused by electricity theft in the world each year are as high as billions of dollars, especially in developing countries, where the problem of electricity theft is more serious.

[0003] Traditional methods of electricity theft detection mainly rely on manual on-site inspections and meter seal inspections, which are inefficient, costly, and inaccurate. With the development of smart grid technology, the popularity of smart meters and advanced metering systems (AMI) has provided new technical means for electricity theft detection. These technologies can collect and analyze power data in real time, providing a richer source of data for electricity theft detection. However, electricity theft detection in a smart grid environment also faces new challenges, including increased data complexity and detection difficulty.

[0004] In order to improve the efficiency and accuracy of electricity theft detection, a variety of new electricity theft detection technologies have emerged in recent years. These technologies are mainly divided into two categories: 1. Software detection methods: This type of method analyzes the periodic electricity consumption of users recorded by smart meters and uses big data analysis technology for detection. However, these methods usually only rely on electricity consumption data for analysis, which makes it difficult to achieve accurate detection and cannot provide an effective way to obtain evidence; 2. Hardware detection methods: This type of method analyzes the power network topology of a specific power consumption area and designs corresponding monitoring devices. These devices can monitor and measure the electrical status information of the installed nodes, and monitor and obtain evidence of electricity theft by comparing the difference between the measured values ​​and the user's meter. However, these methods have limited monitoring areas and high costs, making it difficult to achieve large-scale deployment.

[0005] In summary, the existing electricity theft detection technology still has deficiencies in terms of real-time performance, accuracy, economy, and evidence collection capabilities. Therefore, developing a new electricity theft detection method that can combine the advantages of software and hardware to improve the efficiency and accuracy of electricity theft detection is of great significance for ensuring the safe use of power resources and maintaining the stable operation of the power grid. Summary of the invention

[0006] The purpose of the present invention is to solve the problems raised in the above background technology. The present invention provides a multi-feature electricity theft detection method based on the combination of software and hardware.

[0007] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:

[0008] A multi-feature electricity theft detection method based on the combination of software and hardware, the method comprising the following steps:

[0009] S1. Hardware deployment and data collection: Through the main body of the electricity theft detection device and the infrared meter reader, multiple types of power status data such as voltage, current, power factor and power consumption of the three-phase power transmission line and the user's electricity meter are collected in real time;

[0010] S2, data processing, including normalization and missing value interpolation;

[0011] S3, data analysis and electricity theft identification, using the extracted multi-dimensional features to identify electricity theft through a deep learning model;

[0012] S4. Establishment of the device numbering system, including area code, line number, user serial number and phase identification;

[0013] S5. Set the data collection cycle according to actual needs and specifications, and ensure the timeliness and effectiveness of the data.

[0014] Furthermore, the hardware deployment and data collection steps include:

[0015] S1. Install the device body on the selected power transmission line;

[0016] S2. Install an infrared meter reader on the user's electric meter;

[0017] S3, reading the meter data through the infrared meter reader and transmitting the data to the device body;

[0018] S4. The device uploads the data to the cloud server.

[0019] Furthermore, the data processing step includes:

[0020] S1. Normalization, eliminating dimensional differences, using the formula

[0021] S2, missing value interpolation, using Lagrange interpolation method to fill the missing data.

[0022] Furthermore, the data analysis and electricity theft identification steps include:

[0023] S1, extract multidimensional features using Catch22 feature set and supplementary features;

[0024] S2, convert the features into a one-dimensional token sequence through a convolutional neural network;

[0025] S3. Use the Transformer network for deep learning processing to capture the complex relationships between data; S4. Determine whether the user has engaged in electricity theft.

[0026] Furthermore, the electricity theft detection device comprises:

[0027] The device body is used to periodically measure three-phase voltage, current, power factor and power consumption; the infrared meter reader is used to read the user's meter data.

[0028] Furthermore, the electricity theft detection device comprises:

[0029] Snap-on current sensor;

[0030] Clip-on voltage take-off;

[0031] Aluminum alloy housing.

[0032] Furthermore, the device numbering system includes:

[0033] The area code is used to identify the area or area group where the meter is located;

[0034] Line number, used to identify the specific line in the area;

[0035] User serial number, used to identify the access order of users on the same line;

[0036] Phase identification, used to identify whether the meter is three-phase or single-phase.

[0037] Furthermore, the data collection cycle setting method includes:

[0038] S1. Set the data collection cycle according to actual needs and specifications;

[0039] S2. Ensure the timeliness and validity of data;

[0040] S3. Audit to avoid errors.

[0041] Furthermore, the deep learning model includes:

[0042] Use multi-head attention mechanism to capture the attention distribution in different spaces in parallel;

[0043] Output sequence probability distribution for all categories;

[0044] The softmax function is used to determine whether the user steals electricity.

[0045] Furthermore, the method further comprises:

[0046] According to the electricity theft detection results, provide evidence collection and alarm for electricity theft.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. This solution uses automated hardware deployment, including the electricity theft detection device and infrared meter reader. This solution can realize real-time collection of power transmission line and user meter data, significantly improve the efficiency of electricity theft detection, and reduce dependence on traditional manual inspection.

[0049] 2. This solution adopts a point-to-point transmission method, which ensures the real-time and accuracy of the data transmission process from the infrared meter reader to the device body and then to the cloud server, providing a reliable basis for data analysis.

[0050] 3. This scheme uses the Catch22 feature set and three supplementary features (mean, variance, and slope) for multi-dimensional feature extraction, and combines the deep learning model of convolutional neural network and Transformer network to improve the precision and accuracy of electricity theft behavior identification.

[0051] 4. The design of the device body in this solution adopts cost-effective materials and structures, supports live installation, reduces deployment and maintenance costs, and at the same time, the low cost and easy installation characteristics of the infrared meter reader make the solution very economical.

[0052] 5. This solution supports the deployment of detection devices on a wide range of power transmission lines and user meters, effectively expanding the coverage of electricity theft detection and improving the overall security level of the power grid.

[0053] 6. The modular design of software and hardware in this solution enables the solution to be flexibly expanded and upgraded according to different application scenarios and requirements, enhancing the adaptability and long-term effectiveness of the system.

[0054] 7. The detection method in this scheme reduces the threat to the stability and reliability of the power grid by timely discovering and responding to electricity theft, and reduces the risk of safety accidents that may be caused by electricity theft.

[0055] 8. The data and analysis results provided in this solution can assist power companies in more effective grid management and planning, optimize resource allocation, and improve service quality and operational efficiency.

[0056] 9. This solution also includes evidence collection and alarm functions for electricity theft, which helps to quickly respond to electricity theft incidents and take timely measures to reduce the economic losses of power companies.

[0057] 10. The implementation of this plan has promoted the development of smart grid technology, especially the technological innovation in data collection, processing and analysis, laying the foundation for the intelligence and automation of future power grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of the steps of a multi-feature electricity theft detection method based on the combination of software and hardware in the present invention;

[0059] Figure 2 A schematic diagram of the steps of hardware deployment and data collection in the present invention;

[0060] Figure 3 A schematic diagram of the steps of data analysis and electricity theft identification in the present invention;

[0061] Figure 4 This is a working diagram of the electricity theft detection device in the present invention;

[0062] Figure 5 It is a flowchart of hardware deployment and data collection in the present invention;

[0063] Figure 6 A flow chart of data analysis and electricity theft identification in the present invention;

[0064] Figure 7 This is a flowchart of the pre-trained zone line Transformer in the present invention. DETAILED DESCRIPTION

[0065] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0066] This embodiment provides a multi-feature electricity theft detection method based on a combination of software and hardware, which is mainly used to solve the shortcomings of existing electricity theft detection technology in terms of real-time performance, accuracy, economy, and evidence collection ability. Figure 1-Figure 7 Make a detailed description:

[0067] A multi-feature electricity theft detection method based on the combination of software and hardware, the method comprising the following steps:

[0068] S1. Hardware deployment and data collection. Using the electricity theft detection device and infrared meter reader, we can realize the real-time collection of multiple types of power status data such as voltage, current, power factor and power consumption of three-phase power transmission lines and user meters. This real-time data collection mechanism not only improves the timeliness of electricity theft detection, but also provides rich and accurate raw data for subsequent data analysis.

[0069] S2. Data processing: normalize the collected data to eliminate the differences between different dimensions, and effectively fill in the missing parts of the data through missing value interpolation technology to ensure the integrity and consistency of the data. These preprocessing steps lay a solid foundation for subsequent data analysis;

[0070] S3, data analysis and electricity theft identification, using the extracted multi-dimensional features, through the deep learning model to identify electricity theft behavior, so as to accurately identify electricity theft behavior. This step is the core of the solution. It not only improves the accuracy of electricity theft detection, but also reduces manual intervention through intelligent analysis methods, thus improving the efficiency of the entire detection process;

[0071] S4. Establishment of device numbering system, including area code, line number, user serial number and phase identification, which provides a unique identification for each device, facilitates management and tracking, and also provides strong support for accurate matching and analysis of data;

[0072] S5. Setting of data collection cycle: According to actual needs and specifications, the data collection cycle can be flexibly set to ensure the timeliness and effectiveness of the data, so that the power theft detection can be more in line with the actual operating conditions and improve the adaptability and flexibility of the detection.

[0073] Through real-time data collection, precise data processing, in-depth data analysis and flexible data collection cycle setting, this solution not only improves the efficiency and accuracy of electricity theft detection, but also reduces costs and enhances the security and stability of the power grid. It has significant practical value and social significance.

[0074] In this solution, hardware deployment and data collection steps are the basis for realizing electricity theft detection, which specifically include the following key links:

[0075] S1. First, the electricity theft detection device is installed on the selected power transmission line. This step ensures that key power parameters can be obtained directly from the source of the power system. The installation location of the device is carefully selected to ensure that it can cover key power transmission nodes, thereby achieving effective monitoring of power flow.

[0076] S2. Subsequently, an infrared meter reader is configured on the user's electricity meter. The device can read the meter data in a non-invasive manner, avoiding direct contact with the user's electricity meter, while ensuring the safety and accuracy of data collection. The use of the infrared meter reader improves the efficiency of data collection and reduces interference with the user's daily electricity consumption activities;

[0077] S3. Next, the meter data is read through the infrared meter reader and the data is transmitted to the device body. This step utilizes the wireless transmission function of the infrared meter reader to achieve real-time data collection and transmission. During the data transmission process, encryption and verification technology is used to ensure the security and integrity of the data and prevent the data from being tampered with or lost during the transmission process.

[0078] S4. Finally, the device uploads the collected data to the cloud server. This step achieves fast data upload through a high-speed data transmission network. The powerful computing and storage capabilities of the cloud server provide strong support for subsequent data processing and analysis. At the same time, the centralized management of the cloud also facilitates the unified processing and analysis of large amounts of data, improving the efficiency and reliability of the entire electricity theft detection system.

[0079] The hardware deployment and data collection steps of this solution form a complete data collection and transmission process. From the installation of the main device of the power transmission line, to the configuration of the infrared meter reader of the user's meter, to the secure transmission and cloud storage of data, each link has been carefully designed and optimized to ensure the comprehensiveness, accuracy and real-time nature of data collection, laying a solid foundation for the accurate identification of electricity theft.

[0080] In this solution, the data processing step is the key link to ensure the accuracy of electricity theft detection, which includes the following detailed operations:

[0081] S1. First, normalization is performed to eliminate dimensional differences. This step is crucial because it ensures that power data from different sources and different units can be compared and analyzed under a unified standard. Specifically, the following formula is used for normalization: This formula maps all data values ​​to between 0 and 1, so that data with different parameters can be compared on the same scale, thus providing a consistent basis for subsequent feature extraction and data analysis;

[0082] S2. Subsequently, in order to deal with the problem of missing data that may occur during the data collection process, the Lagrange interpolation method is used to fill the missing values. The Lagrange interpolation method is an effective mathematical tool that estimates unknown or missing data points by using known data points. This method takes into account the local characteristics of the data and can accurately predict missing values, thereby ensuring the integrity and continuity of the data. In this way, even when the data is incomplete, effective data analysis can be performed, which improves the reliability and accuracy of electricity theft detection.

[0083] The data processing steps of this scheme ensure the consistency and integrity of the data through normalization and missing value interpolation. These processes not only provide a high-quality data basis for data analysis, but also optimize the data processing process through mathematical methods, so that the electricity theft detection system can more accurately identify and analyze potential electricity theft behaviors. The implementation of these steps significantly improves the efficiency and effectiveness of the entire electricity theft detection process and provides strong data support for the safe operation of the power system.

[0084] In this solution, data analysis and electricity theft identification steps are the core links to achieve accurate electricity theft detection. The specific steps are as follows:

[0085] S1. First, use the Catch22 feature set and supplementary features to extract multidimensional features. This step is the starting point of data analysis. By extracting key information from the collected power data, including voltage, current, power factor, power consumption, etc., as well as their statistical characteristics such as mean, variance and slope, these features together constitute a comprehensive multidimensional feature set, providing rich input data for subsequent deep learning analysis;

[0086] S2. Subsequently, the extracted multi-dimensional features are converted into a one-dimensional token sequence through a convolutional neural network. This step is the key link in data preprocessing. The convolutional neural network can automatically learn the spatial hierarchical structure in the data and effectively encode the multi-dimensional features into a one-dimensional token sequence, making the data more suitable for subsequent deep learning model processing;

[0087] S3. Next, we use the Transformer network for deep learning processing to capture the complex associations between data. The Transformer network is an advanced deep learning model that can handle long-distance dependencies in sequence data through the self-attention mechanism. In this solution, the Transformer network is used to analyze token sequences and identify potential complex patterns and associations in the data, thus providing strong support for the identification of electricity theft.

[0088] S4. Finally, determine whether the user has committed electricity theft. After in-depth analysis of the above steps, the model will give the final electricity theft detection results, including whether electricity theft is detected and the type and degree of electricity theft. These results can be directly used for decision support of power companies, helping them to take timely measures to prevent and reduce the occurrence of electricity theft.

[0089] The data analysis and electricity theft identification steps of this solution achieve efficient and accurate identification of electricity theft through a series of carefully designed data processing and analysis processes. From multi-dimensional feature extraction to the application of deep learning models, each step aims to improve the performance and reliability of electricity theft detection and ensure the safe and stable operation of the power system.

[0090] In this solution, the design and function of the electricity theft detection device are key components to achieve effective electricity theft monitoring, including the following two core components:

[0091] First, the device body is designed to periodically measure three-phase voltage, current, power factor and power consumption. This component is the data acquisition center of the power theft detection system. Through high-precision sensors and measurement technology, it can monitor the operating status of the power system in real time. The measurement function of the device body not only covers the basic parameters of the power system, but also can adjust the measurement frequency and accuracy as needed to ensure the real-time and accuracy of the data, providing a solid data foundation for the analysis of power theft behavior;

[0092] Next, the infrared meter reader is integrated to read the user's electricity meter data. The infrared meter reader uses infrared communication technology to wirelessly read and record the detailed electricity consumption data of the meter without physical contact with the meter. This not only improves the convenience of data collection, but also avoids interference with the user's electricity consumption. Through the infrared meter reader, the system can directly obtain the user's electricity consumption information, including electricity consumption, electricity consumption time and electricity consumption pattern, etc. This information is crucial for analyzing and identifying electricity theft.

[0093] In general, the electricity theft detection device realizes comprehensive monitoring of the power system and user meters through the coordinated work of the device body and the infrared meter reader. The periodic measurement of the device body provides a macro perspective of the operation of the power system, while the meter data reading of the infrared meter reader provides micro details of the user's electricity usage behavior. The combination of the two enables the electricity theft detection system to capture data from multiple dimensions, providing comprehensive and accurate information for subsequent data analysis and electricity theft identification. This comprehensive monitoring mechanism significantly improves the efficiency and accuracy of electricity theft detection, and provides a strong guarantee for the safe operation of the power system.

[0094] In this solution, the composition of the electricity theft detection device is an important basis for achieving efficient electricity theft monitoring, and is specifically composed of the following key components:

[0095] First, the snap-on current sensor included in the device is a key component for measuring current. This sensor is compact in design and easy to install. It can be quickly fixed on the power line through the snap-on mechanism to achieve accurate current measurement. Its high sensitivity and fast response characteristics ensure the accuracy and real-time nature of current data, providing important current information for power theft detection.

[0096] Next, the clamp-type voltage sensor is another core component of the device, which is used to obtain voltage data in a contactless manner. This design allows the device to measure voltage without being directly connected to the power supply, reducing safety risks during installation and avoiding potential damage to power lines. The stability and durability of the clamp-type voltage sensor ensure the continuity and reliability of voltage data.

[0097] Finally, the entire electricity theft detection device is encapsulated in an aluminum alloy housing, which not only provides physical protection to prevent the device from being damaged by environmental factors such as moisture, dust and impact, but also has excellent electromagnetic compatibility and heat dissipation performance, ensuring the long-term stable operation of the internal components of the device. The choice of aluminum alloy material also takes into account weight and cost-effectiveness, making the device both lightweight and economical.

[0098] In general, the electricity theft detection device realizes an efficient, stable and durable electricity theft monitoring solution through the carefully designed snap-on current sensor, clamp-type voltage collector and aluminum alloy casing. The coordinated work of these components not only improves the performance and reliability of electricity theft detection, but also reduces the overall cost and installation difficulty of the device through optimized design, providing strong support for the safe operation and maintenance of the power system.

[0099] In this solution, the device numbering system is an important part of ensuring the efficient operation of the electricity theft detection system and accurate data management, which includes the following key elements:

[0100] First, the area code is used to identify the area or area group where the meter is located. This coding system gives each area or area group a unique identifier, which facilitates the rapid identification and location of meters within a wide geographical range, thereby optimizing the data classification and retrieval process.

[0101] Next, the line number is used to identify specific lines within the area. By assigning a unique number to each line, it is possible to accurately trace and associate it to a specific power transmission path, which is crucial for analyzing power flow and identifying abnormal patterns.

[0102] In addition, the introduction of user serial numbers is used to identify the order in which users access the same line. This serial number not only helps record the access point of each user, but also ensures that it can be accurately associated with a specific user during data analysis, thus providing the necessary user-level information for identifying electricity theft.

[0103] Finally, the phase identification is designed to identify whether the meter is three-phase or single-phase. This identification is directly related to the type of meter and the power parameters it measures, and is crucial to ensuring the accuracy of the data and the applicability of subsequent analysis.

[0104] In general, the device numbering system provides a comprehensive, orderly and efficient identification and classification system for electricity theft detection devices through the organic combination of area codes, line numbers, user serial numbers and phase identification. This not only improves the efficiency of data management, but also enhances the accuracy and response speed of electricity theft detection through precise identification and tracking, providing a strong guarantee for the safe and stable operation of the power system.

[0105] In this solution, the data collection cycle setting method is a key step to ensure the efficient and accurate operation of the electricity theft detection system, which specifically includes the following consecutive operation links:

[0106] S1. First, set the data collection cycle according to actual needs and specifications. This step involves detailed planning of power system monitoring, including determining the frequency and time interval of data collection. By considering the operating characteristics of the power grid, the user's power consumption pattern, and the possible time of power theft, the data collection cycle can be reasonably set to meet the dual requirements of real-time performance and accuracy.

[0107] S2. Subsequently, ensure the timeliness and effectiveness of the data. After the data collection cycle is determined, the system will collect data according to the established schedule. This step requires that data collection must not only be timely to capture immediate changes in the power system, but also ensure the quality and integrity of the data for effective analysis. Through real-time monitoring and quality control mechanisms, it can be ensured that the collected data has practical significance and analytical value for power theft detection;

[0108] S3. Finally, audit to avoid errors. After data collection is completed, strict data audit is an important step to avoid errors and improve data quality. This step includes verifying and proofreading the collected data, checking the consistency and accuracy of the data, and identifying and correcting possible errors or anomalies. Through this audit process, the reliability of the data can be further improved, providing a solid foundation for subsequent data analysis and electricity theft identification.

[0109] In summary, the data collection cycle setting method builds a systematic data collection and quality management process through continuous planning, execution and review steps. This process not only improves the efficiency and accuracy of electricity theft detection, but also provides strong data support for the safety and stability of the power system by ensuring the timeliness and effectiveness of the data.

[0110] In this solution, the deep learning model is the core part of realizing accurate electricity theft identification, which includes the following key steps:

[0111] First, the multi-head attention mechanism is used to capture the attention distribution in different spaces in parallel. This step is one of the characteristics of the deep learning model. Through the multi-head attention mechanism, the model can focus on multiple important features in the data at the same time, effectively capturing and processing complex data patterns. This parallel processing method enhances the model's sensitivity to subtle differences in data and provides rich information for the identification of electricity theft.

[0112] Then, the probability distribution of all categories of the output sequence is analyzed through the multi-head attention mechanism. After that, the model will classify each data point and generate a probability distribution, indicating the possibility that the data point belongs to different categories. This probability distribution provides a quantitative basis for the final decision, allowing the model to more objectively and accurately evaluate the risk of electricity theft for each data point.

[0113] Finally, the softmax function is used to determine whether the user has stolen electricity. After obtaining the probability distribution, the model will use the softmax function to make the final judgment on the classification result. The Softmax function can convert the original score output by the model into a probability value, so that the output value of each category is between 0 and 1, and the sum of the output values ​​of all categories is 1. In this way, the model can determine the category to which each data point is most likely to belong, that is, to determine whether the user has stolen electricity.

[0114] In summary, the deep learning model builds an efficient and accurate electricity theft identification process through the multi-head attention mechanism, the output of probability distribution and the application of softmax function. This process not only improves the accuracy and efficiency of electricity theft detection, but also provides strong technical support for the safety management of power systems through quantitative analysis methods.

[0115] In this solution, the multi-feature electricity theft detection method based on the combination of software and hardware further includes the functions of providing evidence collection and alarm of electricity theft behavior according to the electricity theft detection results. The specific steps are as follows:

[0116] First, when the deep learning model completes the identification and analysis of electricity theft, the system will automatically record and save the relevant detection results. This step ensures that all identified electricity theft incidents have detailed data records, providing a basis for subsequent evidence collection work;

[0117] The system will then collect evidence of the electricity theft detected. In this step, the system will collect and organize all relevant data and evidence, including abnormal changes in power parameters, deviations in user power usage patterns, and other possible abnormal behaviors. These evidences will be recorded and stored in detail so that they can be provided to relevant departments for further investigation and processing when necessary.

[0118] Finally, the system will automatically trigger the alarm mechanism according to the set thresholds and rules. When the detected electricity theft exceeds the preset threshold or meets the specific alarm conditions, the system will immediately send an alarm signal. This alarm signal can be a real-time notification, an automatically sent report, or a linkage response with other security systems to ensure that relevant personnel can be informed of the electricity theft incident in a timely manner and take appropriate measures.

[0119] In summary, this solution combines the power theft detection results with evidence collection and alarm functions to form a complete power theft detection and response process. This process not only improves the accuracy and efficiency of power theft detection, but also provides strong support for the safety management of the power system through timely evidence collection and alarm, ensuring the safety and reliable supply of power resources.

[0120] The implementation steps of the multi-feature electricity theft detection method based on the combination of software and hardware are as follows:

[0121] 1. Hardware deployment and data collection:

[0122] First, the electricity theft detection device is installed on the selected power transmission line. The device is responsible for periodically measuring key power parameters such as three-phase voltage, current, power factor and power consumption.

[0123] Next, an infrared meter reader is configured on the user's electricity meter, which is used to read the meter data contactlessly;

[0124] The meter data read is wirelessly transmitted to the device body through the infrared meter reader, and then the device body uploads the data to the cloud server for storage and analysis through a wired or wireless network.

[0125] 2. Data processing:

[0126] On the cloud server, the collected data is normalized using the formula To eliminate dimensional differences and ensure that data are compared under the same standard;

[0127] The missing values ​​in the data are filled using Lagrange interpolation method to ensure the integrity and continuity of the data.

[0128] 3. Data analysis and electricity theft identification:

[0129] Use Catch22 feature set and supplementary features (mean, variance, slope) to extract multidimensional features from the processed data;

[0130] The multi-dimensional features are converted into one-dimensional token sequences through convolutional neural networks to provide input for deep learning models;

[0131] Use a deep learning model based on the Transformer network to analyze token sequences, capture complex relationships between data, and identify characteristic patterns of electricity theft behavior;

[0132] The model outputs the probability distribution of each user's electricity theft behavior and uses the softmax function to determine whether the user has stolen electricity.

[0133] 4. Establishment of device numbering system:

[0134] Establish a unique numbering system for each electricity theft detection device, including area code, line number, user serial number and phase identification, to facilitate equipment management and data tracking.

[0135] 5. Setting of data collection cycle:

[0136] Set the data collection cycle according to actual needs and specifications to ensure the timeliness and effectiveness of the data;

[0137] Regularly review data collection cycles to avoid errors and adapt to changes in grid operation.

[0138] 6. Evidence collection and reporting of electricity theft:

[0139] Based on the detection results of the deep learning model, the system automatically records and saves detailed data and evidence of electricity theft;

[0140] When electricity theft is detected, the system triggers an alarm mechanism and promptly notifies relevant personnel to take action through real-time notification, automatic reporting or other linkage responses.

[0141] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-feature electricity theft detection method based on the combination of software and hardware, characterized in that: The method comprises the following steps: S1. Hardware deployment and data collection: Through the main body of the electricity theft detection device and the infrared meter reader, multiple types of power status data such as voltage, current, power factor and power consumption of the three-phase power transmission line and the user's electricity meter are collected in real time; S2, data processing, including normalization and missing value interpolation; S3, data analysis and electricity theft identification, using the extracted multi-dimensional features to identify electricity theft through a deep learning model; S4. Establishment of the device numbering system, including area code, line number, user serial number and phase identification; S5. Set the data collection cycle according to actual needs and specifications, and ensure the timeliness and effectiveness of the data.

2. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The hardware deployment and data collection steps include: S1. Install the device body on the selected power transmission line; S2. Install an infrared meter reader on the user's electric meter; S3, reading the meter data through the infrared meter reader and transmitting the data to the device body; S4. The device uploads the data to the cloud server.

3. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The data processing steps include: S1. Normalization, eliminating dimensional differences, using the formula S2, missing value interpolation, using Lagrange interpolation method to fill the missing data.

4. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The data analysis and electricity theft identification steps include: S1, extract multidimensional features using Catch22 feature set and supplementary features; S2, convert the features into a one-dimensional token sequence through a convolutional neural network; S3, using Transformer network for deep learning processing to capture complex relationships between data; S4. Determine whether the user has engaged in electricity theft.

5. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The electricity theft detection device comprises: The device body is used to periodically measure three-phase voltage, current, power factor and power consumption; Infrared meter reader, used to read user's electricity meter data.

6. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The electricity theft detection device comprises: Snap-on current sensor; Clip-on voltage take-off; Aluminum alloy housing.

7. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The device numbering system described includes: The area code is used to identify the area or area group where the meter is located; Line number, used to identify the specific line in the area; User serial number, used to identify the access order of users on the same line; Phase identification, used to identify whether the meter is three-phase or single-phase.

8. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The data acquisition cycle setting method comprises: S1. Set the data collection cycle according to actual needs and specifications; S2. Ensure the timeliness and validity of data; S3. Audit to avoid errors.

9. The method for detecting electricity theft based on multiple characteristics of software and hardware according to claim 1 is characterized by: The deep learning model includes: Use multi-head attention mechanism to capture the attention distribution in different spaces in parallel; Output sequence probability distribution for all categories; The softmax function is used to determine whether the user steals electricity.

10. The preparation method of the multi-feature electricity theft detection method based on the combination of software and hardware according to claim 1, characterized in that: The method further comprises: According to the electricity theft detection results, provide evidence collection and alarm for electricity theft.

Images