Method and system for quickly identifying electricity stealing of metering box through intelligent measuring switch

Abstract

The invention provides a method and a system for quickly identifying electricity stealing of a metering box by an intelligent measuring switch, and relates to the technical field of electricity stealing detection. Setting a preliminary electricity stealing identification rule of the metering box; according to the electricity utilization data, under the metering box electricity stealing preliminary identification rule, metering box electricity stealing preliminary identification is carried out; when the metering box is preliminarily identified to have no electricity stealing phenomenon, discarding the collected electricity utilization data; when the metering box is preliminarily identified to have an electricity stealing phenomenon, performing data preprocessing on the electricity utilization data to obtain a two-dimensional signal diagram; a CNN-RF algorithm-based metering box electricity stealing accurate identification model is constructed; according to the two-dimensional signal diagram, through a CNN-RF algorithm-based metering box electricity larceny accurate identification model, metering box electricity larceny accurate identification is carried out; when the metering box is accurately identified that no electricity stealing phenomenon exists, discarding the collected electricity utilization data; and when the metering box is accurately identified to have an electricity stealing phenomenon, an early warning signal is sent out.

Description

Technical Field

[0001] The present invention relates to the technical field of power theft detection, and particularly to a method and system for quickly identifying power theft in a metering box by an intelligent metering switch. Background Art

[0002] With the continuous growth of power demand, power supply enterprises are facing increasingly severe contradictions between supply and demand and management pressures. Power theft not only infringes on the economic interests of power supply enterprises but also disrupts the normal power supply and consumption order of society. Power thieves usually install a switch between the power theft live wire in the metering box and the outgoing line of the electric meter, or insert a socket into the power theft line to connect part of the load to the power theft line. In this way, power thieves can turn off the switch or unplug the socket during power department inspections, making the electric meter seem to measure normally, thus avoiding manual verification.

[0003] Currently, power theft identification mainly relies on the comparative analysis of the power consumption information of intelligent electric meters and concentrators in the substation area. However, this method highly depends on manual verification in practical applications, resulting in low work efficiency, lagging discovery of power theft behaviors, and easy omission of intermittent power theft behaviors.

[0004] In addition, due to the diversity and complexity of user power consumption patterns, the detection method based solely on power consumption information comparison is prone to misjudgment and difficult to meet the requirements of efficient and accurate power theft identification. Summary of the Invention

[0005] In order to solve the technical problems in the prior art that power theft identification mainly relies on the comparative analysis of the power consumption information of intelligent electric meters and concentrators in the substation area, highly depends on manual verification in practical applications, resulting in low work efficiency, lagging discovery of power theft behaviors, and easy omission of intermittent power theft behaviors, and due to the diversity and complexity of user power consumption patterns, the detection method based solely on power consumption information comparison is prone to misjudgment and difficult to meet the requirements of efficient and accurate power theft identification, the present invention provides a method and system for quickly identifying power theft in a metering box by an intelligent metering switch.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect:

[0008] A method for quickly identifying power theft in a metering box by an intelligent metering switch provided by an embodiment of the present invention includes:

[0009] S1: Collect power consumption data through an intelligent metering switch;

[0010] S2: Set preliminary identification rules for power theft in the metering box;

[0011] S3: Based on the power consumption data, perform preliminary identification of power theft in the metering box under the preliminary identification rules for power theft in the metering box;

[0012] S4: When it is initially identified that there is no power theft in the metering box, discard the collected power consumption data and return to S1; when it is initially identified that there is power theft in the metering box, proceed to the next step;

[0013] S5: Perform data preprocessing on the power consumption data when it is initially identified that there is power theft in the metering box to obtain a two-dimensional signal diagram;

[0014] S6: Construct an accurate power theft identification model for the metering box based on the CNN-RF algorithm;

[0015] S7: According to the two-dimensional signal diagram, perform accurate power theft identification for the metering box through the accurate power theft identification model of the metering box based on the CNN-RF algorithm;

[0016] S8: When it is accurately identified that there is no power theft in the metering box, discard the collected power consumption data and return to S1; when it is accurately identified that there is power theft in the metering box, send out a warning signal.

[0017] Second aspect:

[0018] A system for an intelligent measurement switch to quickly identify power theft in a metering box provided by an embodiment of the present invention includes:

[0019] A processor;

[0020] A memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, the method for an intelligent measurement switch to quickly identify power theft in a metering box as described in the first aspect is implemented.

[0021] Third aspect:

[0022] A computer-readable storage medium provided by an embodiment of the present invention, on which a computer program is stored, and when the program is executed by a processor, the method for an intelligent measurement switch to quickly identify power theft in a metering box as described in the first aspect is implemented.

[0023] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0024] In the embodiments of the present invention, based on power consumption data, initial identification of electricity theft in the metering box is carried out under the initial identification rules of electricity theft in the metering box. Instead of solely relying on the comparative analysis of the electricity consumption information of the smart meter and the concentrator in the substation area, the high dependence on manual verification is effectively avoided in practical applications, thereby improving work efficiency, promptly detecting electricity theft behavior, and reducing the missed judgment rate of intermittent electricity theft behavior. By performing data preprocessing on the power consumption data when electricity theft is initially identified in the metering box, a two-dimensional signal diagram is obtained. According to the two-dimensional signal diagram, an accurate identification model for electricity theft in the metering box based on the CNN-RF algorithm is used to accurately identify electricity theft in the metering box. In the face of the diversity and complexity of user electricity consumption patterns, electricity theft identification is no longer solely dependent on the comparison of electricity consumption information, significantly reducing the misjudgment rate and meeting the requirements for efficient and accurate electricity theft identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0026] Figure 1 It is a schematic flowchart of a method for quickly identifying electricity theft in a metering box by an intelligent measurement switch provided by an embodiment of the present invention;

[0027] Figure 2 It is a schematic structural diagram of a system for quickly identifying electricity theft in a metering box by an intelligent measurement switch provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The following will describe the technical solutions in the present invention with reference to the drawings.

[0029] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0030] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same. "Of", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same.

[0031] In the embodiments of the present invention, sometimes subscripts such as W 1 may be written in a non-subscript form such as W1. When the difference is not emphasized, their intended meanings are the same.

[0032] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0033] Refer to the attached Figure 1 illustrates a schematic flowchart of a method for a smart measurement switch to quickly identify electricity theft in a metering box provided by an embodiment of the present invention.

[0034] The embodiments of the present invention provide a method for a smart measurement switch to quickly identify electricity theft in a metering box. This method can be implemented by a device for a smart measurement switch to quickly identify electricity theft in a metering box, and this device for a smart measurement switch to quickly identify electricity theft in a metering box can be a terminal or a server. The processing flow of the method for a smart measurement switch to quickly identify electricity theft in a metering box can include the following steps:

[0035] S1: Collect power consumption data through a smart measurement switch.

[0036] Specifically, the smart measurement switch collects the power consumption data of the user's electricity meter through the RS-485 bus at a frequency of once per minute.

[0037] It should be noted that a smart measurement switch is an intelligent power device integrating power measurement, data collection, communication, and control functions, mainly used for real-time monitoring and management of the power consumption of power users. It can collect real-time data of power-consuming devices through built-in sensors and communication modules, and upload the data to the monitoring platform through methods such as the RS-485 bus, Ethernet, or wireless communication.

[0038] It should be noted that the RS-485 bus is a widely used multi-point communication standard in industry for realizing serial communication between devices. It adopts differential signal transmission mode and conducts data transmission through a pair of twisted pairs. It has the characteristics of strong anti-interference ability, long transmission distance, and supports connection of up to 32 devices. The RS-485 bus supports half-duplex communication, that is, data can only be transmitted unidirectionally at the same time, and is often used in scenarios such as industrial automation, building control, and intelligent meters that require reliable and long-distance communication.

[0039] Optionally, the electricity consumption data specifically includes: the active power of the intelligent measurement switch before meter reading, the active power of the intelligent measurement switch after meter reading, the active electric energy of the intelligent measurement switch after meter reading, the total active power of the single-phase electric energy meters connected under the intelligent measurement switch after meter reading, and the total active electric energy of the single-phase electric energy meters connected under the intelligent measurement switch after meter reading.

[0040] In the present invention, the intelligent measurement switch is adopted to obtain the electricity consumption data of the user's electric energy meter through the RS-485 bus at a frequency of once per minute, which can make full use of the advantages of the RS-485 bus such as strong anti-interference ability, long transmission distance, and support for multi-device connection, and realize stable and efficient acquisition of multi-user electricity consumption data.

[0041] S2: Set the preliminary identification rules for electricity theft in the metering box.

[0042] Optionally, there are two types of preliminary identification rules for electricity theft in the metering box. Those skilled in the art can choose any one of them according to actual needs for the preliminary identification of electricity theft in the metering box, and the present invention does not make any limitation here.

[0043] In the present invention, by setting two types of preliminary identification rules for electricity theft in the metering box, the identification strategy can be adjusted according to different application scenarios, data characteristics, or device conditions, so as to improve the applicability and accuracy of the preliminary identification of electricity theft. At the same time, it also provides technical guarantee for the popularization and optimization of the system in diverse applications.

[0044] The first type, the preliminary identification rules for electricity theft in the metering box specifically include:

[0045] S20A1: Calculate the active power fluctuation of the intelligent measurement switch during the meter reading process:

[0046]

[0047] Among them, represents the active power fluctuation of the intelligent measurement switch during the nth meter reading process, P b,n represents the active power of the intelligent measurement switch after the nth meter reading, P f,n represents the active power of the intelligent measurement switch before the nth meter reading.

[0048] S20A2: Calculate the difference between the active power of the intelligent measurement switch after meter reading and the total active power of the single-phase electric energy meters connected below:

[0049]

[0050] Wherein, represents the difference between the active power of the intelligent measurement switch after the nth meter reading and the total active power of the single-phase electric energy meters connected below, and P m,n represents the total active power of the single-phase electric energy meters connected below the intelligent measurement switch after the nth meter reading.

[0051] S20A3: Calculate the difference between the active electric energy of the intelligent measurement switch after meter reading and the total active electric energy of the single-phase electric energy meters connected below:

[0052] ΔE n = |E b,n - E m,n

[0053] Wherein, ΔE n represents the difference between the active electric energy of the intelligent measurement switch after the nth meter reading and the total active electric energy of the single-phase electric energy meters connected below, E b,n represents the active electric energy of the intelligent measurement switch after the nth meter reading, and E m,n represents the total active electric energy of the single-phase electric energy meters connected below the intelligent measurement switch after the nth meter reading.

[0054] S20A4: When the active power fluctuation of the intelligent measurement switch during meter reading is greater than the first electricity theft power threshold or the difference between the active power of the intelligent measurement switch after meter reading and the total active power of the single-phase electric energy meters connected below is less than the second electricity theft power threshold, the metering box is preliminarily identified as having no electricity theft phenomenon. When the power fluctuation of the intelligent measurement switch during meter reading is less than the first electricity theft power threshold and the difference between the active power of the intelligent measurement switch after meter reading and the total active power of the single-phase electric energy meters connected below is greater than the first electricity theft power threshold, proceed to the next step.

[0055] It should be noted that those skilled in the art can set the magnitudes of the first electricity theft power threshold and the second electricity theft power threshold according to actual needs, and the present invention does not make any limitations herein.

[0056] S20A5: Calculate the cumulative electricity theft amount:

[0057] E S = ΔE n - ΔE 1

[0058] Wherein, E S represents the cumulative electricity theft amount, and ΔE 1It represents the difference between the active electric energy of the intelligent measurement switch after the first meter reading and the total active electric energy of the single-phase electric energy meters connected below.

[0059] S20A6: Determine whether the cumulative power theft electricity amount is greater than the power theft electricity threshold. If so, the metering box is preliminarily identified as having a power theft phenomenon. Otherwise, the metering box is preliminarily identified as not having a power theft phenomenon.

[0060] In the present invention, by calculating the active power fluctuation, power difference, and electric energy difference, the power theft behavior is analyzed from different perspectives, comprehensively covering possible power theft modes, and the accuracy of preliminary identification is improved. The target parameters are automatically calculated based on the power consumption data collected by the measurement switch and the single-phase electric energy meter, realizing data-driven rule judgment, reducing the limitations of manual intervention and empirical rules, and improving the scientific nature of the identification rules.

[0061] Second, the preliminary identification rules for the metering box power theft specifically include:

[0062] S20B1: Calculate the SMA value of the power consumption data for the current hour:

[0063]

[0064] Among them, SMA n represents the SMA value of the power consumption data for the nth hour, n represents the number of hours, and x i represents the power consumption data for the ith hour.

[0065] It should be noted that SMA (Simple Moving Average) is a commonly used time series analysis method for smoothing data to extract its long-term trend. It calculates the average value of data points within a specific time window, reducing the impact of short-term fluctuations on the data, and thus more clearly reflecting the change trend of the data. In power theft detection, SMA is often used to analyze user power consumption data, and by smoothing the power consumption characteristics within a certain time window, abnormal fluctuations or abnormal trends are identified.

[0066] S20B2: Calculate the first target SMA difference based on the SMA difference between the power consumption data for two adjacent hours:

[0067]

[0068] Among them, ΔSMA 1 represents the first target SMA difference, max represents maximizing, and SMA i-1 represents the SMA value of the power consumption data for the (i - 1)th hour.

[0069] S20B3: Based on the first target SMA difference, use the hourly status algorithm to determine whether the electricity consumption data for the current hour is abnormal. If so, the metering box is initially identified as having a power theft phenomenon. Otherwise, proceed to the next step:

[0070]

[0071] Among them, S n represents the decision condition of the hourly status algorithm for the nth hour. Among them, when S n = 0, the electricity consumption data for the nth hour is normal. When S n = 1, the electricity consumption data for the nth hour is abnormal. SMA n-1 represents the SMA value of the electricity consumption data for the (n - 1)th hour.

[0072] S20B4: Based on the SMA difference of the electricity consumption data at the same moment on two adjacent days for the current hour, calculate the second target SMA difference:

[0073]

[0074] Among them, ΔSMA 2 represents the second target SMA difference, m represents the number of days, SMA j,n represents the SMA value of the electricity consumption data for the nth hour on the jth day, and SMA j-1,n represents the SMA value of the electricity consumption data for the nth hour on the (j - 1)th day.

[0075] S20B5: Based on the second target SMA difference, use the daily status algorithm to determine whether the electricity consumption data for the current hour on the current day is abnormal. If so, the metering box is initially identified as having a power theft phenomenon. Otherwise, the metering box is initially identified as not having a power theft phenomenon:

[0076]

[0077] Among them, S m,n represents the decision condition of the daily status algorithm for the nth hour on the mth day. Among them, when S m = 0, the electricity consumption data for the nth hour on the mth day is normal. When S m = 1, the electricity consumption data for the nth hour on the mth day is abnormal. SMA m,n represents the SMA value of the electricity consumption data for the nth hour on the mth day, and SMA m-1,n represents the SMA value of the electricity consumption data for the nth hour on the (m - 1)th day.

[0078] In the present invention, by smoothing the electricity consumption data through simple moving average (SMA), the influence of short-term fluctuations can be effectively reduced, highlighting the long-term trend of the data, and providing a clearer reference benchmark for subsequent anomaly detection. By analyzing the SMA difference of electricity consumption data for two adjacent hours, possible abnormal electricity consumption behaviors of users in the short term can be quickly identified. By analyzing the SMA difference of electricity consumption data at the same moment on two adjacent days, anomalies in the long-term trend can be detected. Combining short-term fluctuations and long-term trends, various electricity theft behaviors can be identified, including sudden electricity theft and periodic electricity theft, further enhancing the reliability of the detection rules.

[0079] S3: According to the electricity consumption data, under the preliminary identification rule of electricity theft in the metering box, conduct a preliminary identification of electricity theft in the metering box.

[0080] In the present invention, by using the preliminary identification rule of electricity theft in the metering box according to the electricity consumption data for preliminary identification of electricity theft, it is possible to quickly determine whether there is an electricity theft behavior through efficient rule screening, reducing the possibility of false alarms and missed reports.

[0081] S4: When it is preliminarily identified that there is no electricity theft in the metering box, discard the collected electricity consumption data and return to S1. When it is preliminarily identified that there is electricity theft in the metering box, proceed to the next step.

[0082] Specifically, when it is preliminarily identified that there is no electricity theft in the metering box, discard the collected electricity consumption data and return to S1 to collect a new round of electricity consumption data. When it is preliminarily identified that there is electricity theft in the metering box, enter S5.

[0083] In the present invention, when it is preliminarily identified that there is no electricity theft in the metering box, discarding the electricity consumption data and returning for re-collection can avoid the storage and processing of invalid data, reducing system resource consumption. And when it is preliminarily identified that there is electricity theft in the metering box and proceeding to the next step for further analysis, it can ensure that the system focuses its computing and detection resources on high-risk targets, thereby improving the overall detection efficiency and accuracy, and at the same time achieving a rapid response to electricity theft behaviors.

[0084] S5: Perform data preprocessing on the electricity consumption data when it is preliminarily identified that there is electricity theft in the metering box to obtain a two-dimensional signal diagram.

[0085] In the present invention, performing data preprocessing on the electricity consumption data preliminarily identified as having electricity theft and generating a two-dimensional signal diagram can effectively reduce the noise interference of the original data, providing high-quality data input for accurate identification of electricity theft, thereby improving the identification efficiency and accuracy of the model.

[0086] In a possible implementation manner, S5 specifically includes sub-steps S501 to S503:

[0087] S501: Normalize the electricity consumption data when power theft is initially identified in the metering box:

[0088]

[0089] Among them, R z represents the normalized data value of the z-th electricity consumption data, x z represents the original data value of the z-th electricity consumption data, x min represents the minimum data value, x max represents the maximum data value.

[0090] S502: Determine the symmetric mapping angles of each electricity consumption data:

[0091]

[0092] Among them, represents the forward mapping angle of the z-th electricity consumption data, represents the initial angle, δ represents the rotation range, l represents the time lag coefficient, R z+l represents the normalized data value of the (i + l)-th electricity consumption data, represents the reverse mapping angle of the z-th electricity consumption data.

[0093] S503: Generate a two-dimensional signal diagram with the normalized data value as the radius and the forward mapping angle and reverse mapping angle as the mapping angles.

[0094] In the present invention, through normalization, the original data in different ranges is mapped to the [0, 1] interval, making the data have a consistent scale. The normalized data can better highlight the relative changes, facilitating the model to capture the characteristics of abnormal electricity consumption behavior. Combining the normalized values, the forward and reverse mapping angles are generated, making full use of the time series characteristics of the electricity consumption data and enhancing the ability to capture the time pattern of power theft behavior. Using the normalized value as the radius and the forward and reverse mapping angles as coordinates to generate a two-dimensional signal diagram helps to capture the change law of the electricity consumption data through the spatial distribution.

[0095] S6: Construct an accurate power theft identification model for the metering box based on the CNN-RF algorithm.

[0096] It should be noted that the CNN-RF algorithm is a hybrid algorithm that combines a Convolutional Neural Network (CNN) and a Random Forest (RF), aiming to utilize the advantages of both for efficient and accurate classification tasks. The CNN is responsible for performing multi-level feature extraction on the input data (such as images or signals), gradually refining the depth information from low-level local features to high-level semantic features. The RF, based on the high-dimensional features extracted by the CNN, uses multiple decision trees for classification and outputs the final classification result through a voting mechanism. This algorithm not only has the powerful feature extraction ability of the CNN but also combines the efficient and robust classification performance of the RF, making it particularly suitable for task scenarios that require complex feature analysis and high accuracy, such as electricity theft detection, fault diagnosis, and target recognition.

[0097] Among them, the Convolutional Neural Network (CNN) is a deep learning model designed specifically for processing images and time-series data, and is good at automatically extracting the local features and spatial relationships of data. The CNN consists of a convolutional layer (for feature extraction), a pooling layer (for dimensionality reduction and enhancing feature robustness), and a fully connected layer (for final classification or regression). The convolutional layer scans the input data with a sliding convolutional kernel to extract features such as edges, textures, and patterns. The pooling layer further compresses the feature map, reducing the computational complexity and preventing overfitting. The CNN is widely used in fields such as image classification, object detection, and natural language processing, and is favored for its automated feature extraction ability and powerful modeling ability.

[0098] Among them, the Random Forest (RF) is a machine learning algorithm based on the idea of ensemble learning, consisting of multiple independently constructed decision trees, each of which is generated through random sampling and feature selection. The RF outputs the final classification result (or regression prediction value) through the voting mechanism of "a small number of related but not completely identical" decision trees, thereby reducing the overfitting risk of a single decision tree and improving the robustness of the model. The RF combines efficiency, ease of use, and noise resistance, and is suitable for high-dimensional feature data and imbalanced datasets, and is commonly used in scenarios such as classification, regression, and feature selection, such as credit assessment, medical diagnosis, and electricity theft detection.

[0099] Optionally, the CNN-RF algorithm includes an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer, and an RF classification layer. The convolutional layer includes a first convolutional layer, a second convolutional layer, and a third convolutional layer. The pooling layer includes a first pooling layer, a second pooling layer, and a third pooling layer.

[0100] In the present invention, a hybrid algorithm (CNN-RF) combining a convolutional neural network (CNN) and a random forest (RF) is adopted, which gives full play to the advantages of the CNN in automatic feature extraction and high-dimensional feature expression, as well as the efficiency and robustness of the RF in the classification stage, enabling the algorithm to accurately capture complex patterns and abnormal features in the electricity consumption data of the metering box, while reducing the risk of overfitting and noise interference. Through the deep features extracted by the CNN and the integrated decision-making mechanism of the RF, the present invention realizes efficient and accurate electricity theft identification, improving the identification accuracy, adaptability and overall reliability of the system.

[0101] S7: According to the two-dimensional signal graph, perform accurate electricity theft identification of the metering box through the accurate electricity theft identification model of the metering box based on the CNN-RF algorithm.

[0102] In the present invention, by inputting the generated two-dimensional signal graph into the accurate electricity theft identification model of the metering box based on the CNN-RF algorithm, the deep feature extraction ability of the CNN for image and time-series data and the efficient classification performance of the RF can be fully utilized, enabling meaningful features to be automatically extracted from a large amount of data, reducing human intervention, and improving the accuracy and robustness of electricity theft identification, effectively identifying potential electricity theft behaviors.

[0103] In a possible implementation manner, S7 specifically includes sub-steps S701 to S710:

[0104] S701: In the input layer, input the two-dimensional signal graph.

[0105] S702: In the first convolutional layer, perform feature extraction on the two-dimensional signal graph to obtain the first local feature map.

[0106] It should be noted that the first convolutional layer is a convolutional layer containing 8 7×7 convolutional kernels.

[0107] S703: In the first pooling layer, perform a max-pooling operation on the first local feature map to obtain the first pooled feature map.

[0108] It should be noted that the first pooling layer is a pooling layer containing a 2×2 max-pooling window with a stride of 2.

[0109] S704: In the second convolutional layer, perform feature extraction on the first pooled feature map to obtain the second local feature map.

[0110] It should be noted that the second convolutional layer is a convolutional layer containing 12 5×5 convolutional kernels.

[0111] S705: In the second pooling layer, perform a max-pooling operation on the second local feature map to obtain the second pooled feature map.

[0112] It should be noted that the second pooling layer is a pooling layer with a 2×2 maximum pooling window and a stride of 2.

[0113] S706: In the third convolutional layer, feature extraction is performed on the second pooled feature map to obtain a third local feature map.

[0114] It should be noted that the third convolutional layer is a convolutional layer containing 16 3×3 convolutional kernels.

[0115] S707: In the third pooling layer, a maximum pooling operation is performed on the third local feature map to obtain a third pooled feature map.

[0116] It should be noted that the third pooling layer is a pooling layer with a 2×2 maximum pooling window and a stride of 2.

[0117] S708: The first pooled feature map, the second pooled feature map, and the third pooled feature map are weighted and fused to obtain a target fused feature map.

[0118] Optionally, the target feature map is specifically:

[0119]

[0120] where H represents the target feature map, W 1 represents the weight coefficient of the first pooled feature map, A represents the activation function, F 1 represents the first pooled feature map, represents the concatenation operation, W 2 represents the weight coefficient of the second pooled feature map, F 2 represents the second pooled feature map, W 3 represents the weight coefficient of the third pooled feature map, F 3 represents the third pooled feature map.

[0121] S709: In the fully connected layer, the target fused feature map is flattened into a target feature vector.

[0122] S710: In the RF classification layer, based on the target feature vector, accurate identification of electricity theft in metering boxes is performed through the RF algorithm.

[0123] In the present invention, through multiple convolutional layers, the model can gradually extract local features of electricity consumption data from low levels to high levels. Each pooling layer is used to reduce the dimension of the output of the convolutional layer, compress the feature map, enhance the robustness of the features, and reduce the computational complexity. Weighted fusion of multiple pooled feature maps can effectively combine the features extracted from different pooling layers and improve the recognition ability for complex data patterns. The target fusion feature map is transformed into a vector format that can be input to a random forest classifier (RF) through a flattening operation, ensuring that the features can be effectively used for classification. Through the RF classifier and the ensemble method combining multiple decision trees, the risk of overfitting of a single tree can be reduced and the robustness of the model can be improved.

[0124] In a possible implementation manner, S710 specifically includes sub-steps S7101 to S7106:

[0125] S7101: Construct an initial forest composed of multiple decision trees, and the decision tree includes multiple nodes.

[0126] S7102: Calculate the information gain and Gini index of the current node of each decision tree:

[0127]

[0128] Among them, Gain() represents the information gain, M represents the number of target feature vectors in the current node, a represents the attribute of the target feature vector, Ent() represents the entropy of the target feature vector, V represents the number of child nodes generated after node division, || represents taking the absolute value, M v represents the number of target feature vectors in the v-th child node after node division, K represents the number of categories, p k represents the probability that the target feature vector belongs to the k-th category, and Gini() represents the Gini index.

[0129] S7103: Combine the information gain and Gini index of the current node to dynamically determine the node division rule:

[0130]

[0131] Among them, O represents the node division rule, min represents taking the minimum value, W{} represents the comprehensive function, ρ represents the weight coefficient of the Gini index, τ represents the weight coefficient of the information gain, and R represents the domain of the weight coefficient.

[0132] S7104: According to the node division rule, use the target feature vector to perform node division on the current node.

[0133] S7105: Stop node division when the number of target feature vectors in the current node is less than the preset number, and obtain multiple accurate recognition results output by multiple decision trees.

[0134] S7106: Based on each precise recognition result, through a voting mechanism, determine the precise recognition result with the most votes in the end as the precise recognition result of the metering box.

[0135] In the present invention, by combining information gain and Gini index, dynamically selecting the partitioning rule, it can more accurately judge which features contribute more to classification, thereby optimizing node partitioning and improving the accuracy of the model. By dynamically adjusting the weight coefficients of the Gini index and information gain, the partitioning strategy of the decision tree can be flexibly optimized according to actual application requirements, enabling the model to better adapt to specific recognition tasks. By having multiple decision trees vote to determine the final classification result, it ensures that the model is more robust and accurate, and improves the stability of the final result.

[0136] S8: When it is precisely recognized that there is no power theft in the metering box, discard the collected power consumption data and return to S1. When it is precisely recognized that there is power theft in the metering box, send out a warning signal.

[0137] Specifically, when it is preliminarily recognized that there is no power theft in the metering box, discard the collected power consumption data and return to S1 to collect a new round of power consumption data. When it is preliminarily recognized that there is power theft in the metering box, the system will send out a warning signal in various ways, including sending real-time alarm information to the power monitoring center, notifying relevant management personnel via text message or email, generating an alarm record on the monitoring platform, and triggering an audible and visual alarm to prompt on-site personnel.

[0138] In the present invention, when it is precisely recognized that there is power theft in the metering box, sending out a warning signal in a timely manner can quickly respond to potential power theft behaviors, effectively reducing power losses and equipment damage. In addition, by sending real-time alarm information to the power monitoring center and relevant management personnel, and triggering an audible and visual alarm on-site, a multi-level and rapid alarm feedback mechanism can be realized to ensure that power theft incidents are handled in a timely manner. On the other hand, when it is preliminarily recognized as no power theft, discarding invalid data and re-collecting it can avoid wasting system resources by invalid information, improving the efficiency and accuracy of the system.

[0139] In a possible implementation manner, the method for the intelligent measurement switch to quickly identify power theft in the metering box further includes:

[0140] S9: With the goal of minimizing the mean square error loss function, optimize the precise power theft recognition model of the metering box based on the CNN-RF algorithm through the particle swarm optimization algorithm, and determine the optimal parameter set of the model.

[0141] It should be noted that the Particle Swarm Optimization (PSO) is a global optimization algorithm based on swarm intelligence, which finds the optimal solution to a problem by simulating the foraging behavior of bird flocks or the swimming behavior of fish schools in nature. Each "particle" represents a potential solution. The particles move in the search space, and their positions and velocities are guided by their own historical optimal solutions (personal best) and the overall optimal solution of the group (global best). By continuously updating their velocities and positions, the particles gradually approach the optimal solution. PSO has the characteristics of simple algorithm, few parameters, and strong global search ability, and is widely used in function optimization, neural network training, path planning, and other complex optimization problems.

[0142] In the present invention, the Particle Swarm Optimization (PSO) is used to minimize the mean square error loss function and optimize the parameters of the electricity theft accurate identification model based on the CNN-RF algorithm, which can improve the accuracy and efficiency of parameter selection by simulating swarm intelligence to search for the optimal solution.

[0143] In a possible implementation manner, S9 specifically includes sub-steps S901 and S902:

[0144] S901: With the goal of minimizing the mean square error loss function, construct an objective function:

[0145]

[0146] where min represents taking the minimum value, MSE represents the objective function, θ represents the set of parameters of the electricity theft accurate identification model of the metering box based on the CNN-RF algorithm, Q represents the total number of initial images of the head-mounted earphone injection mold, Y q represents the true accurate identification result of the qth electricity consumption data, represents the actual accurate identification result of the qth electricity consumption data.

[0147] It should be noted that the mean square error loss function is an index used to measure the error between the model prediction result and the true result, and evaluates the model performance by calculating the average value of the square of the difference between the predicted value and the true value. This function has a higher penalty for larger prediction errors, so it can more sensitively reflect the prediction accuracy of the model. The smaller the mean square error, the closer the predicted value is to the true value, and the better the model performance. It is widely used in regression problems and model optimization.

[0148] S902: According to the objective function, optimize the electricity theft accurate identification model of the metering box based on the CNN-RF algorithm through the particle swarm optimization algorithm, and determine the optimal set of model parameters.

[0149] In a possible implementation manner, S902 specifically includes sub-steps S9021 to S9026:

[0150] S9021: Initialize parameters and set the maximum number of iterations of the particle swarm optimization algorithm.

[0151] S9022: Generate an initial particle swarm. The initial particle swarm includes multiple particles, and each particle represents a set of parameters of a feasible metering box electricity theft accurate identification model based on the CNN-RF algorithm.

[0152] S9023: Calculate the non-linear inertia weight of each particle:

[0153]

[0154] where ω(t) represents the inertia weight at the t-th iteration, ω max represents the initial maximum value of the inertia weight, ω min represents the initial minimum value of the inertia weight, T max represents the maximum number of iterations, and e represents the exponential function.

[0155] S9024: Use the objective function as the fitness function, calculate the fitness value of each particle, and take the particle with the maximum fitness value as the current particle.

[0156] S9025: Update the velocity and position of the current particle according to the non-linear inertia weight:

[0157] v b (t + 1) = ω(t)·v b (t) + c 1 r 1 (P b -X b ) + c 2 r 2 (G - X b )

[0158] X b (t + 1) = X b (t) + X b (t + 1)

[0159] where v b (t + 1) represents the velocity of the b-th particle at the (t + 1)-th iteration, v b (t) represents the velocity of the b-th particle at the t-th iteration, c 1 represents the individual learning factor, r 1 represents the first random number, r 2 represents the second random number, P b represents the historical best position of the b-th particle, X b represents the position of the b-th particle at the t-th iteration, c 2represents the swarm learning factor, G represents the global optimal position, and X b (t + 1) represents the position of the b-th particle at the (t + 1)-th iteration.

[0160] S9026: Determine whether the maximum number of iterations has been reached. If so, output the optimal parameter set of the metering box electricity theft precise identification model based on the CNN-RF algorithm. Otherwise, return to step S9025.

[0161] In the present invention, by minimizing the mean square error loss function (MSE), the gap between the model prediction value and the true value can be quantified, ensuring that the model can gradually reduce the prediction error through the optimization process and improve the prediction accuracy of the model. PSO realizes global search by simulating swarm intelligence, can find the optimal solution in a wide solution space, is especially suitable for complex high-dimensional problems, and can find the global optimal parameter set, thereby improving the generalization ability of the model. By dynamically adjusting the inertia weight of the particles, the search process can widely explore the solution space in the initial stage and focus on local optimization in the later stage, further improving the search efficiency and accuracy.

[0162] Refer to the attached drawings of the specification Figure 2 which shows the structural schematic diagram of a system for quickly identifying electricity theft in a metering box by an intelligent measurement switch provided by the present invention.

[0163] The present invention also provides a system 20 for quickly identifying electricity theft in a metering box by an intelligent measurement switch, which is applied to the method for quickly identifying electricity theft in a metering box by an intelligent measurement switch described above, and includes:

[0164] A processor 201;

[0165] A memory 202, on which computer-readable instructions are stored. When the computer-readable instructions are executed by the processor 201, the method for quickly identifying electricity theft in a metering box by an intelligent measurement switch as described in the method embodiment is realized.

[0166] The system 20 for quickly identifying electricity theft in a metering box provided by the present invention can execute the method for quickly identifying electricity theft in a metering box by an intelligent measurement switch described above and achieve the same or similar technical effects. To avoid repetition, the present invention will not be elaborated further.

[0167] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0168] In the embodiments of the present invention, based on electricity consumption data, preliminary identification of electricity theft in the metering box is carried out under the preliminary identification rules of electricity theft in the metering box, rather than simply relying on the comparative analysis of the electricity consumption information of the smart meter and the concentrator in the substation area. In practical applications, the high dependence on manual verification is effectively avoided, thereby improving work efficiency, promptly discovering electricity theft behavior, and reducing the misjudgment rate of intermittent electricity theft behavior. By performing data preprocessing on the electricity consumption data when electricity theft is preliminarily identified in the metering box, a two-dimensional signal diagram is obtained, and based on the two-dimensional signal diagram, an accurate identification model of electricity theft in the metering box based on the CNN-RF algorithm is used to accurately identify electricity theft in the metering box. In the face of the diversity and complexity of user electricity consumption patterns, electricity theft identification is no longer solely dependent on the comparison of electricity consumption information, significantly reducing the misjudgment rate and meeting the requirements for efficient and accurate electricity theft identification.

[0169] It should be understood that the processor in the embodiments of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0170] It should also be understood that the memory in the embodiments of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0171] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0172] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0173] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0174] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0175] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0176] Those skilled in the art can clearly understand that for the sake of convenience and brevity of description, the specific working processes of the above-described devices, apparatuses, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0177] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0178] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0179] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0180] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0181] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, it implements the method for quickly identifying electricity theft in a metering box as described in the method embodiment.

[0182] The computer-readable storage medium provided by the present invention can implement the steps and effects of the method for quickly identifying electricity theft in a metering box in the above method embodiment. To avoid repetition, the present invention will not elaborate further.

[0183] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0184] In the embodiment of the present invention, based on electricity consumption data, under the preliminary identification rule of electricity theft in a metering box, the preliminary identification of electricity theft in the metering box is carried out, and it no longer simply relies on the comparative analysis of the electricity consumption information of the intelligent electricity meter and the concentrator in the area. In practical applications, it effectively avoids the high dependence on manual verification, thereby improving work efficiency, promptly discovering electricity theft behavior, and reducing the missed judgment rate of intermittent electricity theft behavior. By performing data preprocessing on the electricity consumption data when electricity theft is preliminarily identified in the metering box, a two-dimensional signal diagram is obtained, and based on the two-dimensional signal diagram, through the accurate identification model of electricity theft in the metering box based on the CNN-RF algorithm, the accurate identification of electricity theft in the metering box is carried out. In the face of the diversity and complexity of user electricity consumption patterns, it no longer solely relies on the comparison of electricity consumption information for electricity theft identification, significantly reducing the misjudgment rate and being able to meet the requirements of efficient and accurate electricity theft identification.

[0185] As described above, only the specific implementation manners of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0186] The following points need to be explained:

[0187] (1) The accompanying drawings of the embodiments of the present invention only relate to the structures involved in the embodiments of the present invention, and other structures can refer to the general design.

[0188] (2) For clarity, in the accompanying drawings used to describe the embodiments of the present invention, the thickness of layers or regions is enlarged or reduced, that is, these drawings are not drawn to actual scale. It can be understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element can be "directly" on or under the other element or there can be intervening elements.

[0189] (3) Without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0190] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for quickly identifying electricity theft in a meter box using an intelligent measuring switch, characterized in that: include: S1: Collect electricity consumption data through intelligent measuring switches; S2: Setting the preliminary identification rules for electricity theft in the meter box; S3: performing preliminary identification of electricity theft in the meter box according to the electricity consumption data and the preliminary identification rule of electricity theft in the meter box; S4: When the meter box is preliminarily identified as not having electricity theft, the collected electricity consumption data is discarded and the process returns to S1; when the meter box is preliminarily identified as having electricity theft, the process proceeds to the next step; S5: preprocessing the electricity consumption data when the meter box is initially identified to have electricity theft, and obtaining a two-dimensional signal graph; S6: Construct an accurate identification model for electricity theft in meter boxes based on CNN-RF algorithm; S7: According to the two-dimensional signal graph, the meter box electricity theft accurate identification model based on the CNN-RF algorithm is used to accurately identify the meter box electricity theft; S8: When the meter box is accurately identified as not having electricity theft, the collected electricity usage data is discarded and the process returns to S1; when the meter box is accurately identified as having electricity theft, an early warning signal is issued.

2. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1 is characterized in that: The electricity consumption data specifically includes: the active power of the smart measuring switch before meter reading, the active power of the smart measuring switch after meter reading, the active electric energy of the smart measuring switch after meter reading, the total active power of the single-phase electric energy meter under the smart measuring switch after meter reading, and the total active electric energy of the single-phase electric energy meter under the smart measuring switch after meter reading.

3. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1 is characterized in that: The preliminary identification rules for electricity theft in the meter box specifically include: S20A1: Calculate the active power fluctuation of the intelligent measuring switch during meter reading: in, represents the active power fluctuation of the intelligent measuring switch during the nth meter reading process, P b,n Represents the active power of the intelligent measuring switch after the nth meter reading, P f,n Indicates the active power of the intelligent measuring switch before the nth meter reading; S20A2: Calculate the difference between the active power of the intelligent measuring switch after meter reading and the total active power of the single-phase electric energy meter connected to it: in, It represents the difference between the active power of the intelligent measuring switch after the nth meter reading and the total active power of the single-phase electric energy meter hanging below, P m,n It indicates the total active power of the single-phase electric energy meter under the intelligent measuring switch after the nth meter reading; S20A3: Calculate the difference between the active energy of the intelligent measuring switch after meter reading and the total active energy of the downstream single-phase energy meter: ΔE n =|E b,n -E m,n Among them, ΔE n It represents the difference between the active energy of the intelligent measuring switch after the nth meter reading and the total active energy of the single-phase energy meter hanging below, E b,n It represents the active energy of the intelligent measuring switch after the nth meter reading, E m,n It indicates the total active energy of the single-phase energy meter under the intelligent measuring switch after the nth meter reading; S20A4: When the active power fluctuation of the intelligent measuring switch during the meter reading process is greater than the first power theft threshold or the difference between the active power of the intelligent measuring switch after the meter reading and the total active power of the single-phase electric energy meter hanging below is less than the second power theft threshold, the meter box is preliminarily identified as not having power theft; when the power fluctuation of the intelligent measuring switch during the meter reading process is less than the first power theft threshold and the difference between the active power of the intelligent measuring switch after the meter reading and the total active power of the single-phase electric energy meter hanging below is greater than the first power theft threshold, proceed to the next step; S20A5: Calculate the accumulated electricity consumption: E S =ΔE n -ΔE1 Among them, E S Indicates the accumulated amount of electricity theft, ΔE1 indicates the difference between the active energy of the intelligent measuring switch after the first meter reading and the total active energy of the single-phase energy meter hanging below; S20A6: Determine whether the accumulated electricity theft amount is greater than the electricity theft amount threshold; if so, the meter box is preliminarily identified as having electricity theft; otherwise, the meter box is preliminarily identified as not having electricity theft.

4. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1 is characterized in that: The preliminary identification rules for electricity theft in the meter box specifically include: S20B1: Calculate the SMA value of the electricity consumption data for the current hour: Among them, SMA n Indicates the SMA value of the electricity consumption data in the nth hour, where n represents the number of hours and x represents the number of hours. i represents the electricity consumption data for the i-th hour; S20B2: Based on the SMA difference of the electricity consumption data of two consecutive hours, calculate the first target SMA difference: Among them, ΔSMA1 represents the first target SMA difference, max represents maximization, SMA i-1 Represents the SMA value of the electricity consumption data in the i-1th hour; S20B3: Based on the first target SMA difference, determine whether the electricity consumption data of the current hour is abnormal through the hourly status algorithm; if so, the meter box is preliminarily identified to have electricity theft; otherwise, proceed to the next step: Among them, S n represents the decision condition of the hourly state algorithm for the nth hour, where when S n = 0, the electricity consumption data of the nth hour does not have any abnormality. n = 1, the electricity consumption data of the nth hour is abnormal, SMA n-1 Represents the SMA value of the electricity consumption data for the n-1th hour; S20B4: Calculate the second target SMA difference based on the SMA difference of the electricity consumption data of the current hour at the same time on two consecutive days: Among them, ΔSMA2 represents the second target SMA difference, m represents the number of days, SMA j,n It represents the SMA value of the electricity consumption data at the nth hour on the jth day. j-1,n It represents the SMA value of the electricity consumption data at the nth hour on the j-1th day; S20B5: According to the second target SMA difference, through the day status algorithm, determine whether the electricity consumption data of the current hour on the day is abnormal; if so, the meter box is preliminarily identified as having electricity theft; otherwise, the meter box is preliminarily identified as not having electricity theft: Among them, S m,n represents the decision condition of the day-state algorithm for the nth hour on the mth day, where when S m = 0, the electricity consumption data of the nth hour on the mth day does not have any abnormality. m = 1, the electricity consumption data of the nth hour on the mth day is abnormal, SMA m,n It represents the SMA value of the electricity consumption data at the nth hour on the mth day. m-1,n Represents the SMA value of the electricity consumption data at the nth hour on the m-1th day.

5. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1 is characterized in that: The S5 specifically includes: S501: normalizing the electricity consumption data when the meter box is initially identified to have electricity theft; S502: Determine the symmetric mapping angle of each power consumption data; S503: Using the normalized data value as the radius, and using the forward mapping angle and the reverse mapping angle as mapping angles, to generate a two-dimensional signal graph.

6. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1 is characterized in that: The CNN-RF algorithm includes an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer and an RF classification layer, the convolutional layer includes a first convolutional layer, a second convolutional layer and a third convolutional layer, and the pooling layer includes a first pooling layer, a second pooling layer and a third pooling layer.

7. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 6 is characterized in that: The S7 specifically includes: S701: In the input layer, input the two-dimensional signal graph; S702: In the first convolutional layer, extract features from the two-dimensional signal graph to obtain a first local feature graph; S703: In the first pooling layer, performing a maximum pooling operation on the first local feature map to obtain a first pooling feature map; S704: In the second convolutional layer, extract features from the first pooled feature map to obtain a second local feature map; S705: In the second pooling layer, performing a maximum pooling operation on the second local feature map to obtain a second pooling feature map; S706: In the third convolutional layer, extract features from the second pooled feature map to obtain a third local feature map; S707: In the third pooling layer, performing a maximum pooling operation on the third local feature map to obtain a third pooling feature map; S708: performing weighted fusion on the first pooling feature map, the second pooling feature map, and the third pooling feature map to obtain a target fused feature map; S709: In the fully connected layer, flatten the target fusion feature map into a target feature vector; S710: In the RF classification layer, based on the target feature vector, the electricity theft in the meter box is accurately identified by using an RF algorithm.

8. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 7 is characterized in that: The S710 specifically includes: S7101: construct an initial forest consisting of multiple decision trees, where the decision trees include multiple nodes; S7102: Calculate the information gain and Gini index of the current node of each decision tree; S7103: Dynamically determine the node division rule based on the information gain and Gini index of the current node; S7104: Perform node division on the current node using the target feature vector according to the node division rule; S7105: When the number of target feature vectors in the current node is less than a preset number, stop node division to obtain multiple accurate recognition results output by multiple decision trees; S7106: According to each of the precise identification results, through a voting mechanism, determine the precise identification result with the most votes as the precise identification result of the meter box.

9. The method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to claim 1, characterized in that: Also includes: S9: With the goal of minimizing the mean square error loss function, the meter box electricity theft accurate identification model based on the CNN-RF algorithm is optimized by the particle swarm optimization algorithm to determine the optimal parameter set of the model.

10. A system for quickly identifying electricity theft in a meter box using an intelligent measuring switch, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for quickly identifying electricity theft in a meter box by an intelligent measuring switch according to any one of claims 1 to 9 is implemented.

Images