A method and system for automatically identifying operation risks of electric energy meters

Abstract

The present application relates to the technical field of risk identification of electric energy meters, and provides a method and system for automatically identifying the operation risks of electric energy meters. The method comprises: collecting historical electricity consumption data of the electric energy meter to be identified, analyzing the electricity consumption behavior, and constructing a first electricity consumption comparison space; constructing an electricity consumption comparison library, reading the comparison electricity consumption data set, and constructing a second electricity consumption comparison space; reading the real-time electricity consumption data of the electric energy meter to be identified, using two comparison spaces to identify abnormalities in the real-time data, and determining abnormal electricity consumption indicators; if the abnormal indicator is not zero, verifying it using an associated verification model, and if the verification fails, executing an electricity theft warning. The present application solves the technical problem that during the operation of the electric energy meter, it is impossible to timely and effectively identify abnormal electricity consumption behavior, resulting in a delay in warning of electricity theft behavior, and realizes the effective identification and verification of abnormal electricity consumption behavior through multi-dimensional comparison space and associated verification model, thereby improving the accuracy and timeliness of the identification of electricity theft risks of electric energy meters.

Description

Technical Field

[0001] The present application relates to the technical field of risk identification of electric energy meters, and in particular to a method and system for automatically identifying operating risks of electric energy meters. Background Art

[0002] With the rapid development of smart grids and the continuous upgrading of energy meter technology, energy meters are gradually widely used in residential and commercial areas, which can monitor users' electricity consumption in real time and record detailed electricity consumption data. However, with the popularization of smart energy meters, illegal electricity theft against energy meters is increasing, especially in residential areas. Users may circumvent the normal measurement of energy meters through various means, causing economic losses to power companies. In addition, traditional energy meter abnormality identification methods mainly rely on manual investigation or simple threshold alarms, which makes it difficult to accurately identify complex electricity theft behaviors, and there are problems such as identification lag and high misjudgment rate. Summary of the invention

[0003] The present application provides a method and system for automatically identifying the operating risks of an electric energy meter, aiming to solve the technical problem that during the operation of the electric energy meter, abnormal electricity usage behavior cannot be identified in a timely and effective manner, resulting in delayed early warning of electricity theft.

[0004] In view of the above problems, the present application provides a method and system for automatically identifying operation risks of an electric energy meter.

[0005] The first aspect disclosed in the present application provides a method for automatically identifying the operation risks of an electric energy meter, the method comprising: collecting historical electricity consumption data of the electric energy meter to be identified, analyzing the electricity consumption behavior based on the historical electricity consumption data, and constructing a first electricity consumption comparison space; constructing an electricity consumption comparison library, reading a comparison electricity consumption data set of the electricity consumption comparison library within a predetermined period, and constructing a second electricity consumption comparison space based on the comparison electricity consumption data set; reading the real-time electricity consumption data of the electric energy meter to be identified within a predetermined period, using the first electricity consumption comparison space and the second electricity consumption comparison space to respectively identify abnormalities in the real-time electricity consumption data, and determine an abnormal electricity consumption index; if the abnormal electricity consumption index is not zero, using an associated verification model to verify and judge the abnormal electricity consumption index, and if the verification fails, executing an electricity theft warning for the electric energy meter to be identified.

[0006] Another aspect disclosed in the present application provides an automatic identification system for operation risks of an electric energy meter, the system comprising: a first electricity usage comparison space construction module, the first electricity usage comparison space construction module is used to collect historical electricity usage data of the electric energy meter to be identified, analyze electricity usage behavior based on the historical electricity usage data, and construct a first electricity usage comparison space; a second electricity usage comparison space construction module, the second electricity usage comparison space construction module is used to construct an electricity usage comparison library, read a comparison electricity usage data set of the electricity usage comparison library within a predetermined period, and construct a second electricity usage comparison space based on the comparison electricity usage data set; an abnormality identification module, the abnormality identification module is used to read the real-time electricity usage data of the electric energy meter to be identified within a predetermined period, use the first electricity usage comparison space and the second electricity usage comparison space to perform abnormal identification on the real-time electricity usage data, and determine an abnormal electricity usage index; a verification and judgment module, the verification and judgment module is used to verify and judge the abnormal electricity usage index using an associated verification model if the abnormal electricity usage index is not zero, and if the verification fails, execute an electricity theft warning for the electric energy meter to be identified.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The above-mentioned method for automatically identifying the operation risk of an electric energy meter collects the historical electricity consumption data of the electric energy meter to be identified, analyzes the electricity consumption behavior, and constructs a first electricity consumption comparison space, thereby realizing in-depth mining of the user's long-term electricity consumption habits. At the same time, by constructing an electricity consumption comparison library, reading the comparison electricity consumption data set within a predetermined period, and generating a second electricity consumption comparison space, a more comprehensive comparison benchmark is provided. Subsequently, the real-time electricity consumption data of the electric energy meter to be identified is read, and the real-time data is respectively identified for abnormalities in combination with the first and second comparison spaces, and the abnormal electricity consumption indicators are accurately determined. If the abnormal indicator is detected to be non-zero, further verification and judgment are performed through the associated verification model to ensure the accuracy and reliability of the abnormal judgment. If the verification fails, an electricity theft warning is executed, which effectively improves the ability to timely detect and handle potential electricity theft behaviors.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 The figure is a flow chart of a method for automatically identifying operating risks of an electric energy meter in an embodiment.

[0012] Figure 2 The present invention is an architecture diagram of an automatic identification system for operation risks of an electric energy meter in one embodiment.

[0013] Explanation of the accompanying drawings: first electricity usage comparison space construction module 1, second electricity usage comparison space construction module 2, abnormality identification module 3, verification and judgment module 4. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a method and system for automatically identifying the operating risks of an electric energy meter, thereby solving the technical problem that abnormal electricity usage behavior cannot be identified in a timely and effective manner during the operation of the electric energy meter, resulting in delayed early warning of electricity theft.

[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0016] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0017] Embodiment 1, as Figure 1 As shown, the present application provides a method for automatically identifying operation risks of an electric energy meter, the method comprising:

[0018] The historical electricity consumption data of the electric energy meter to be identified is collected, the electricity consumption behavior is analyzed based on the historical electricity consumption data, and a first electricity consumption comparison space is constructed.

[0019] In the embodiment of the present application, the system terminal obtains the electricity consumption data of the electric energy meter to be identified in the past period of time, which includes electricity consumption, electricity load curve, etc. Based on these historical data, the system terminal trains and constructs the first electricity consumption comparison space through the long short-term memory network for subsequent data comparison and analysis, helps predict the user's electricity consumption range, and serves as an important basis for evaluating whether future real-time data is abnormal.

[0020] Furthermore, the present application provides a method for analyzing electricity consumption behavior based on the historical electricity consumption data to construct a first electricity consumption comparison space, including:

[0021] Collect historical electricity consumption data of the electric energy meter to be identified, wherein the historical electricity consumption data includes historical electricity load curve, historical power consumption, historical equipment start and stop time and historical equipment start and stop frequency; arrange the historical electricity load curve and historical power consumption according to the collection timestamp to generate a historical load curve sequence and a historical power consumption sequence; train a long short-term memory network with the historical load curve sequence and the historical power consumption sequence to obtain a load curve prediction branch and a power consumption prediction branch, which are integrated to generate the first electricity consumption comparison space.

[0022] Preferably, the system terminal collects historical data from the electric energy meter to be identified, including but not limited to historical power load curve, historical power consumption, historical equipment start and stop time and historical equipment start and stop frequency. The historical power load curve represents the power load of the electric energy meter at each time point. For example, within a day or a week, the electric meter records the load changes every hour or every minute to form a complete load curve. The historical power consumption is the total power consumption data in the historical time period, reflecting the total power consumption in each time period. The historical equipment start and stop time records the time point when the power equipment is started and shut down. The historical equipment start and stop frequency represents the number of times the equipment is started and stopped within a period of time, providing information on the frequency of equipment use. The system terminal arranges the historical power load curve and the historical power consumption according to the collected timestamps to obtain a historical load curve sequence and a historical power consumption sequence. The load value and the power consumption value at each time point correspond to each other through the timestamp. Among them, the historical load curve sequence is the load curve data arranged in chronological order. The historical power consumption sequence is the power consumption data arranged in chronological order. These data sequences can be refined to granularity such as minutes and hours, depending on the data collection frequency of the electric meter. After obtaining the historical load curve sequence and the historical power consumption sequence, the system terminal uses the two time series to train the long short-term memory network, respectively, and constructs the load curve prediction branch and the power consumption prediction branch. Taking the load curve prediction branch as an example, the system terminal uses the long short-term memory network to construct a load curve prediction branch, including the input layer, LSTM layer, fully connected layer and output layer, and then inputs the generated historical load curve sequence into the initialized load curve prediction branch for forward propagation, passes through the LSTM unit layer by layer, captures the long-term and short-term dependencies of the historical load curve, and outputs the predicted future load curve through the fully connected layer. Subsequently, the mean square error loss function is used to calculate the error between the prediction result and the actual load curve to obtain the loss value, and then the gradient of the loss value to the weight of each layer is calculated layer by layer through back propagation, and the gradient descent method is used to optimize the parameters of the load curve prediction branch, and the weights of the LSTM layer and the fully connected layer are adjusted to minimize the loss function value. Repeat this process until the loss value converges or the maximum number of iterations is reached. After the training is completed, the data not used for training is input into the load curve prediction branch for testing to evaluate the accuracy of the load curve prediction branch in the load curve prediction task. If the prediction accuracy of the load curve prediction branch meets the requirements, the current load curve prediction branch is output as the final branch. Otherwise, the hyperparameters such as learning rate and training batch are adjusted to continue to optimize the performance of the load curve prediction branch. The trained load curve prediction branch can accurately predict future power loads and provide reliable support for subsequent abnormal power consumption identification. Similarly, the system terminal also uses the historical power consumption sequence training to construct a power consumption prediction branch in the same way as above. This branch can use the rules of historical power consumption to predict possible future power consumption.After obtaining the two prediction branches, the system terminal merges the two prediction branches to generate a first electricity usage comparison space for comparing future real-time electricity usage data, helping the system terminal to use it as a comparison reference when performing anomaly detection on real-time data in the future to determine whether there is abnormal electricity usage behavior.

[0023] A power usage comparison library is constructed, a comparison power usage data set of the power usage comparison library within a predetermined period is read, and a second power usage comparison space is constructed based on the comparison power usage data set.

[0024] In one embodiment, the system terminal analyzes the historical data of the electric energy meter to be identified, identifies the types of electric equipment belonging to these electric energy meters, and based on the obtained types of electric equipment, screens multiple electric energy meters in combination with data such as building specifications and characteristics, and then constructs an electric energy comparison library based on the screening results. This electric energy comparison library contains the historical electric energy data of multiple electric energy meters in a certain area. These data include the historical load curves, historical power consumption of the electric energy meters, and other parameters related to user behavior. Each electric energy meter in the electric energy comparison library has a similar power consumption environment or equipment configuration to ensure the accuracy of data comparison. Subsequently, in each predetermined period (for example, every day, every week, or every month), a comparison electric energy data set is extracted from the electric energy comparison library. The data set includes the electric energy data of all similar electric energy meters in the comparison library, such as load curves and power consumption. These data are used to compare with the real-time electric energy data of the electric energy meter to be identified to help determine whether the electric energy behavior is abnormal. Afterwards, based on the obtained comparison electric energy data set, the system terminal constructs a second electric energy comparison space. This comparison space is established by performing similarity comparison and deviation calculation on the electricity consumption data set, which can reflect the normal electricity consumption behavior pattern of similar electricity meters and provide a reliable benchmark for subsequent real-time data comparison. The construction of this second electricity consumption comparison space helps to more accurately identify whether the electricity meter to be identified has abnormal electricity consumption or potential electricity theft behavior.

[0025] Furthermore, the present application provides a method for constructing a power usage comparison library, including:

[0026] The load monitoring technology is used to identify the historical equipment start and stop time and the historical equipment start and stop frequency to determine the type of electrical equipment; the type of electrical equipment, building specification characteristics and historical segmented power consumption average are used as restriction conditions to perform similar screening on multiple electric energy meters in a predetermined area, and multiple similar electric energy meters that meet the predetermined similarity constraints are obtained to construct the power consumption comparison library.

[0027] Optionally, the system terminal uses load monitoring technology to identify the historical equipment start and stop time and historical equipment start and stop frequency of the electric energy meter to be identified, because each electric energy device will present different start and stop characteristics when in use. For example, air conditioners usually start and stop frequently when the temperature changes greatly, while the start and stop frequency of refrigerators is relatively stable. By identifying the start and stop time points and start and stop frequencies of these devices, the system terminal can identify the main types of electric energy devices used by the user by matching with the start and stop modes of common devices (such as air conditioners, refrigerators, washing machines, etc.). After identifying the type of equipment used by the user, it is combined with the building specification characteristics and the historical segmented power consumption average to construct a restriction condition for similar screening of multiple electric energy meters in a predetermined area. Among them, the building specifications include the type of building where the user is located (such as a single-family house, an apartment, etc.) and the building area. Different types of buildings have different configurations of electric energy devices due to different usage scenarios and space sizes. For example, a single-family house with a larger area may use more air conditioners and lighting equipment, while an apartment has relatively fewer equipment. By combining the equipment type with the building specification characteristics, the power usage habits of the user to whom the electric energy meter belongs can be more accurately identified. The historical segmented electricity consumption average refers to the average electricity consumption divided according to a specific time period (such as daytime, nighttime, weekdays or weekends). Different users have different electricity demand in different time periods. For example, residential users usually have higher electricity consumption at night, while office users use more electricity during the day. The system terminal will use the constructed restrictions to perform similar screening on multiple electricity meters in a predetermined area. Through this multi-dimensional screening, multiple electricity meters that are highly similar to the electricity meter to be identified in terms of equipment usage characteristics, building specifications, and electricity usage patterns can be obtained. These electricity meters will be collected to build a database that can reflect the electricity consumption behavior of similar user groups, namely the electricity comparison library. The electricity comparison library will be used for electricity data comparison and anomaly identification in subsequent steps, providing a reliable reference benchmark to detect whether there are abnormalities or potential electricity theft in the electricity consumption behavior of the electricity meter to be identified.

[0028] Furthermore, the present application provides a method for constructing a second electricity usage comparison space based on the compared electricity usage data set, including:

[0029] Select first electricity consumption data from the comparison electricity consumption data set, wherein the first electricity consumption data includes a first load curve and a first electricity consumption; perform a similarity comparison between the first load curve and other load curves in the comparison electricity consumption data set, and count the amount of data whose similarity is greater than a predetermined similarity threshold, which is recorded as a first criticality; if the first criticality is greater than the predetermined criticality threshold, perform a deviation calculation between the first electricity consumption and other electricity consumptions in the comparison electricity consumption data set, and count the amount of data whose deviation value is less than a predetermined deviation threshold, which is recorded as a second criticality; if the second criticality is greater than the predetermined criticality threshold, add the first electricity consumption data to the high-frequency electricity consumption data set, perform a mean operation on the high-frequency electricity consumption data set, and construct the second electricity consumption comparison space according to the operation result.

[0030] Preferably, the system terminal randomly selects a group of electricity consumption data from the comparison electricity consumption data set as the first electricity consumption data, including the first load curve and the first electricity consumption, which will serve as the basis for subsequent comparison and analysis. Subsequently, the first load curve is compared with the load curves of other electric energy meters in the comparison electricity consumption data set for similarity, and the similarity between the two curves is calculated. Commonly used methods include cosine similarity or Euclidean distance. Taking Euclidean distance as an example, the system terminal calculates the load difference between the first load curve and the comparison curve at each time point, and accumulates the squares of these differences, and then takes the square root of the accumulated result to obtain the Euclidean distance between the two curves. Afterwards, the calculated Euclidean distance is added to 1, and the inverse calculation is performed to obtain the similarity between the two curves, wherein the addition of 1 to the Euclidean distance is to avoid the situation where the denominator is 0 when the Euclidean distance is 0, and to ensure that the similarity ranges from 0 to 1. Repeating this process, the system terminal compares the first load curve with the remaining load curves for similarity, obtains multiple similarities, and then counts the number of load curves with similarities greater than the predetermined similarity threshold, and records the number as the first criticality. If the first criticality is greater than the predetermined critical threshold (that is, the number of similar curves is large enough), the system terminal will continue to the next step of analysis and enter the power consumption deviation calculation. If the first criticality does not reach the threshold, the power consumption data may not be representative enough and cannot proceed to the subsequent steps. For the electric energy meter that has passed the load curve similarity comparison, the system terminal calculates the deviation between the first power consumption and the power consumption of other electric energy meters in the comparison power consumption data set, calculates the deviation value between each comparison power consumption and the first power consumption, and counts the number of power consumption data with a deviation value less than the predetermined deviation threshold, and records the number as the second criticality. If the second criticality is greater than the predetermined critical threshold, that is, the number of power consumption data that meets the conditions is large enough, the system terminal regards the first power consumption data as having a high representativeness and adds it to the high-frequency power consumption data set. Otherwise, the data will be ignored. After all the power consumption data has been judged, the system terminal performs mean calculation on all the data that has been added to the high-frequency power consumption data set, calculates a comprehensive load curve that can represent the overall characteristics of these high-frequency power consumption data, and then generates a second power consumption comparison space based on the calculation results for comparison with the real-time data of the electric energy meter to be identified. Through this space, abnormal power consumption behaviors that differ greatly from normal power consumption patterns can be more accurately identified, providing a solid foundation for subsequent abnormal power consumption identification.

[0031] The real-time power consumption data of the electric energy meter to be identified within a predetermined period is read, and the real-time power consumption data is identified as abnormal using the first power consumption comparison space and the second power consumption comparison space respectively to determine the abnormal power consumption index.

[0032] In one embodiment, the system terminal reads the real-time power consumption data of the electric energy meter to be identified within a predetermined period, which data includes the real-time power load curve and power consumption. Subsequently, the first power consumption comparison space is used to predict the normal power consumption behavior of the electric energy meter to be identified in the current period, and the real-time power consumption data is compared with the prediction result, and the deviation between the real-time power load curve and the predicted load curve, as well as the difference between the real-time power consumption and the predicted power consumption are analyzed. If these differences exceed the predetermined deviation scalar, it is considered that the real-time data may be abnormal. At the same time, the real-time data is also compared with the second power consumption comparison space, and the deviation comparison is performed by calculating the real-time power consumption data and the mean operation result in the second comparison space to determine whether the real-time data deviates from the conventional power consumption mode. If the deviation exceeds the predetermined deviation scalar, it indicates that the power consumption behavior of the real-time data is significantly different from the conventional power consumption mode. Through the analysis of the above two comparison spaces, the system terminal performs an intersection operation on the identified anomalies and determines the abnormal power consumption index to deal with possible power consumption anomalies or power theft.

[0033] Furthermore, the present application provides using the first power usage comparison space and the second power usage comparison space to respectively identify abnormalities in the real-time power usage data and determine abnormal power usage indicators, including:

[0034] The electricity consumption behavior within a predetermined period is predicted through the first electricity consumption comparison space, and a predicted load curve and a predicted power consumption are output; real-time electricity consumption data is read, wherein the real-time electricity consumption data includes a real-time electricity load curve and a real-time power consumption; deviation analysis is performed on the predicted load curve and the real-time electricity load curve, the predicted power consumption and the real-time power consumption, respectively, to determine a first curve deviation and a first power consumption deviation; the first curve deviation and the first power consumption deviation are judged respectively according to a predetermined deviation scalar, to determine a first abnormal electricity consumption index; the real-time electricity load curve and the real-time power consumption are input into the second electricity consumption comparison space for deviation comparison, and a second abnormal electricity consumption index is determined according to the second curve deviation and the second power consumption deviation; an intersection operation is performed on the first abnormal electricity consumption index and the second abnormal electricity consumption index, and the abnormal electricity consumption index is output.

[0035] Optionally, the system terminal uses the load curve prediction branch and the power consumption prediction branch in the first power consumption comparison space to predict the power consumption behavior within a predetermined period, obtains the predicted load curve and the predicted power consumption, and then reads the real-time power consumption data from the electric energy meter in real time, including: the real-time power load curve and the real-time power consumption. Subsequently, the predicted load curve and the real-time load curve, the predicted power consumption and the real-time power consumption are compared one by one, and the first curve deviation and the first power consumption deviation are calculated. The first curve deviation is the difference between the two load curves at each time point, which is determined by calculating the difference between the two curves. The first power consumption deviation is the difference between the predicted power consumption and the real-time power consumption, which is the direct difference between the two values. Afterwards, the calculated first curve deviation and the first power consumption deviation are compared with the corresponding preset deviation scalar. If the deviation exceeds the preset deviation scalar, it is determined that there is an abnormal situation in the first comparison space, and a first abnormal power consumption indicator is generated. In addition, the system terminal also inputs the real-time power load curve and the real-time power consumption into the second power comparison space for deviation comparison, and uses the same method as above to calculate the second curve deviation and the second power consumption deviation based on the mean operation result in the second power comparison space, and judges whether the power consumption behavior of the electric energy meter is consistent with the normal power consumption mode in the comparison space. If the deviation exceeds the preset deviation scalar, it is determined that there is an abnormal situation, and a second abnormal power consumption index is generated. Then, the intersection operation is performed on the first abnormal power consumption index and the second abnormal power consumption index, and the abnormal moments existing in both abnormal power consumption indicators are screened out, and the load data and power consumption data corresponding to these abnormal moments are stored in the abnormal power consumption index as the final evaluation of the abnormal degree of power consumption of the electric energy meter in the current cycle. Through this complete process, it is possible to combine the historical power consumption data of the electric energy meter itself and the data of similar electric energy meters to accurately judge whether there is abnormal power consumption behavior, and ensure the accuracy and comprehensiveness of the identification.

[0036] If the abnormal power consumption index is not zero, the abnormal power consumption index is verified and judged using the associated verification model. If the verification fails, the power theft warning of the electric energy meter to be identified is executed.

[0037] In one embodiment, when it is detected that the abnormal power consumption index is not zero, it means that there may be abnormalities in the power consumption behavior. At this time, the system terminal will verify these anomalies through the association verification model. The association verification model is built based on historical power consumption behavior and equipment correlation. It can analyze whether the power consumption anomaly of the electric energy meter is associated with the normal power consumption pattern. If the deviation between the load curve and power consumption generated by the association verification model and the load data and power consumption data in the abnormal power consumption index exceeds the preset threshold, the system terminal will determine that there is a potential risk of power theft. At this time, an electricity theft warning will be issued to remind managers that there may be power theft and further investigation is required.

[0038] Furthermore, the present application provides a method for constructing an association verification model, including:

[0039] Based on electricity consumption related indicators, query the electricity consumption monitoring log to obtain a sample electricity consumption related indicator set, a sample electricity load curve set and a sample electricity consumption set; use the sample electricity consumption related indicator set and the sample electricity load curve set to train to obtain a first correlation analysis channel, and use the sample electricity consumption related indicator set and the sample electricity consumption set to train to obtain a second correlation analysis channel; integrate the first correlation analysis channel and the second correlation analysis channel to generate the correlation verification model.

[0040] Preferably, the system terminal queries relevant data from the electricity consumption monitoring log based on the electricity consumption related indicators. The electricity consumption related indicators include water consumption, gas consumption and ambient temperature, etc. These indicators can reflect the user's electricity consumption behavior. For example, in some cases, the increase in water consumption may be accompanied by the opening of the electric water heater, the change in gas consumption may be related to the use of gas equipment, and the change in ambient temperature will affect the use of air conditioning and heating equipment. The system terminal obtains these related indicator data from the monitoring log as a sample electricity consumption related indicator set, and simultaneously extracts the electricity load curve and the electricity consumption of the corresponding time period as a sample electricity load curve set and a sample electricity consumption set. Subsequently, the same training method as above is used to train the LSTM model based on the sample electricity consumption related indicator set and the sample electricity load curve set to construct a first association analysis channel. Similarly, the system terminal also uses the sample electricity consumption related indicator set and the sample electricity consumption set to construct a second association analysis channel. After obtaining the first association analysis channel and the second association analysis channel, the system terminal integrates the first association analysis channel and the second association analysis channel to form a complete association verification model. The model can simultaneously analyze the impact of related indicators on the power load curve and power consumption, ensuring that the verification of abnormal power consumption behavior is more comprehensive and accurate, thereby determining whether the anomaly is reasonable or a potential power theft behavior.

[0041] Furthermore, the present application provides inputting the abnormal power consumption indicator into a correlation verification model for verification and judgment, including:

[0042] Collect the real-time power consumption related data of the electric energy meter to be identified within a predetermined period, input the real-time power consumption related data into the related verification model, and output the related load curve and the related power consumption; judge the abnormal power consumption index based on the related load curve and the related power consumption, and generate a power theft warning signal if the deviation exceeds a preset threshold.

[0043] Optionally, the system terminal collects real-time electricity consumption related data within a predetermined period from the electric energy meter to be identified. These related data can help identify the relationship between the user's electricity consumption behavior and external conditions, so as to perform more accurate abnormal identification. Subsequently, the collected real-time electricity consumption related data is input into the trained related verification model to generate a related load curve and related power consumption. The related load curve is the first related analysis channel in the model that predicts the normal electricity load curve of the user in the current period based on related indicators such as water consumption, gas consumption and ambient temperature. The related power consumption is the total power consumption in the current period predicted by the second related analysis channel based on the related indicators. After that, the load data and power consumption data in the abnormal power consumption indicator are compared with the related load curve and related power consumption output by the model, and the deviation at the same time is calculated. If any one of the load curve deviation and power consumption deviation is detected to exceed the preset threshold, it means that there is potential abnormal power consumption behavior, which cannot be explained by normal power consumption related indicators. In this case, the system terminal will generate a power theft warning signal, indicating that there may be power theft behavior, triggering further investigation. Through this process, the system terminal uses the correlation verification model to analyze electricity-related data to ensure that early warnings are issued in a timely manner when abnormal electricity consumption behavior is inconsistent with normal electricity consumption patterns, thereby improving the accuracy and pertinence of anomaly detection.

[0044] In summary, the embodiments of the present application have at least the following technical effects:

[0045] The embodiment of the present application collects the historical electricity consumption data of the electric energy meter, analyzes the electricity consumption behavior, constructs a first electricity consumption comparison space, and further constructs an electricity consumption comparison library to establish a second electricity consumption comparison space based on the comparison electricity consumption data set. During operation, the real-time electricity consumption data of the electric energy meter is read, and the real-time data is abnormally identified using the first and second electricity consumption comparison spaces to determine the abnormal electricity consumption index. If the abnormal electricity consumption index is not zero, it is verified and judged by the associated verification model, and a power theft warning signal is generated if the verification fails. In order to more accurately identify abnormal behavior, the associated verification model judges the rationality of the real-time electricity consumption data through multiple associated analysis channels, and finally generates a judgment result. These technical effects jointly solve the technical problem that the abnormal electricity consumption behavior cannot be identified in a timely and effective manner during the operation of the electric energy meter, resulting in a delayed warning of the electricity theft behavior, and realizes the effective identification and verification of abnormal electricity consumption behavior through multi-dimensional comparison space and associated verification model, and improves the accuracy and timeliness of the identification of the risk of electricity theft of the electric energy meter.

[0046] Embodiment 2, based on the same inventive concept as the method for automatically identifying the operation risk of an electric energy meter in the above embodiment, Figure 2As shown, the present application provides an automatic identification system for operation risks of an electric energy meter, and the system includes: a first electricity usage comparison space construction module 1: the first electricity usage comparison space construction module 1 is used to collect historical electricity usage data of the electric energy meter to be identified, analyze electricity usage behavior based on the historical electricity usage data, and construct a first electricity usage comparison space; a second electricity usage comparison space construction module 2: the second electricity usage comparison space construction module 2 is used to construct an electricity usage comparison library, read the comparison electricity usage data set of the electricity usage comparison library within a predetermined period, and construct a second electricity usage comparison space based on the comparison electricity usage data set; anomaly identification module 3: the anomaly identification module 3 is used to read the real-time electricity usage data of the electric energy meter to be identified within a predetermined period, use the first electricity usage comparison space and the second electricity usage comparison space to perform anomaly identification on the real-time electricity usage data respectively, and determine the abnormal electricity usage index; verification and judgment module 4: the verification and judgment module 4 is used to verify and judge the abnormal electricity usage index using an associated verification model if the abnormal electricity usage index is not zero, and if the verification fails, execute the electricity theft warning of the electric energy meter to be identified.

[0047] Furthermore, the first electricity usage comparison space construction module 1 is also used to execute the following method:

[0048] Collect historical electricity consumption data of the electric energy meter to be identified, wherein the historical electricity consumption data includes historical electricity load curve, historical power consumption, historical equipment start and stop time and historical equipment start and stop frequency; arrange the historical electricity load curve and historical power consumption according to the collection timestamp to generate a historical load curve sequence and a historical power consumption sequence; train a long short-term memory network with the historical load curve sequence and the historical power consumption sequence to obtain a load curve prediction branch and a power consumption prediction branch, which are integrated to generate the first electricity consumption comparison space.

[0049] Furthermore, the second electricity usage comparison space construction module 2 is also used to execute the following method:

[0050] The load monitoring technology is used to identify the historical equipment start and stop time and the historical equipment start and stop frequency to determine the type of electrical equipment; the type of electrical equipment, building specification characteristics and historical segmented power consumption average are used as restriction conditions to perform similar screening on multiple electric energy meters in a predetermined area, and multiple similar electric energy meters that meet the predetermined similarity constraints are obtained to construct the power consumption comparison library.

[0051] Furthermore, the second electricity usage comparison space construction module 2 is also used to execute the following method:

[0052] Select first electricity consumption data from the comparison electricity consumption data set, wherein the first electricity consumption data includes a first load curve and a first electricity consumption; perform a similarity comparison between the first load curve and other load curves in the comparison electricity consumption data set, and count the amount of data whose similarity is greater than a predetermined similarity threshold, which is recorded as a first criticality; if the first criticality is greater than the predetermined criticality threshold, perform a deviation calculation between the first electricity consumption and other electricity consumptions in the comparison electricity consumption data set, and count the amount of data whose deviation value is less than a predetermined deviation threshold, which is recorded as a second criticality; if the second criticality is greater than the predetermined criticality threshold, add the first electricity consumption data to the high-frequency electricity consumption data set, perform a mean operation on the high-frequency electricity consumption data set, and construct the second electricity consumption comparison space according to the operation result.

[0053] Furthermore, the abnormality identification module 3 is also used to execute the following method:

[0054] The electricity consumption behavior within a predetermined period is predicted through the first electricity consumption comparison space, and a predicted load curve and a predicted power consumption are output; real-time electricity consumption data is read, wherein the real-time electricity consumption data includes a real-time electricity load curve and a real-time power consumption; deviation analysis is performed on the predicted load curve and the real-time electricity load curve, the predicted power consumption and the real-time power consumption, respectively, to determine a first curve deviation and a first power consumption deviation; the first curve deviation and the first power consumption deviation are judged respectively according to a predetermined deviation scalar, to determine a first abnormal electricity consumption index; the real-time electricity load curve and the real-time power consumption are input into the second electricity consumption comparison space for deviation comparison, and a second abnormal electricity consumption index is determined according to the second curve deviation and the second power consumption deviation; an intersection operation is performed on the first abnormal electricity consumption index and the second abnormal electricity consumption index, and the abnormal electricity consumption index is output.

[0055] Furthermore, the verification and judgment module 4 is also used to execute the following method:

[0056] Based on electricity consumption related indicators, query the electricity consumption monitoring log to obtain a sample electricity consumption related indicator set, a sample electricity load curve set and a sample electricity consumption set; use the sample electricity consumption related indicator set and the sample electricity load curve set to train to obtain a first correlation analysis channel, and use the sample electricity consumption related indicator set and the sample electricity consumption set to train to obtain a second correlation analysis channel; integrate the first correlation analysis channel and the second correlation analysis channel to generate the correlation verification model.

[0057] Furthermore, the verification and judgment module 4 is also used to execute the following method:

[0058] Collect the real-time power consumption related data of the electric energy meter to be identified within a predetermined period, input the real-time power consumption related data into the related verification model, and output the related load curve and the related power consumption; judge the abnormal power consumption index based on the related load curve and the related power consumption, and generate a power theft warning signal if the deviation exceeds a preset threshold.

[0059] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0061] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for automatically identifying the operation risk of an electric energy meter, characterized in that: include: Collecting historical electricity consumption data of the electric energy meter to be identified, performing electricity consumption behavior analysis based on the historical electricity consumption data, and constructing a first electricity consumption comparison space; Building a power usage comparison library, reading a comparison power usage data set of the power usage comparison library within a predetermined period, and building a second power usage comparison space based on the comparison power usage data set; Reading the real-time power consumption data of the electric energy meter to be identified within a predetermined period, using the first power consumption comparison space and the second power consumption comparison space to perform abnormality identification on the real-time power consumption data, and determining an abnormal power consumption index; If the abnormal power consumption index is not zero, the abnormal power consumption index is verified and judged by using the associated verification model, and if the verification fails, the power theft warning of the electric energy meter to be identified is executed; Building a power usage comparison library with multiple power meters that have similar power usage environments or equipment configurations as the power meter to be identified; Using the first power usage comparison space and the second power usage comparison space to respectively identify abnormalities in the real-time power usage data and determine abnormal power usage indicators includes: Predicting the power consumption behavior of the space within a predetermined period through the first power consumption comparison, and outputting a predicted load curve and predicted power consumption; Reading real-time power consumption data, wherein the real-time power consumption data includes a real-time power load curve and real-time power consumption; Perform deviation analysis on the predicted load curve and the real-time power load curve, the predicted power consumption and the real-time power consumption respectively, and determine a first curve deviation and a first power consumption deviation; According to a predetermined deviation scalar, the first curve deviation and the first power consumption deviation are judged respectively to determine a first abnormal power consumption index; Inputting the real-time power load curve and the real-time power consumption into the second power consumption comparison space for deviation comparison, and determining a second abnormal power consumption index according to the second curve deviation and the second power consumption deviation; Performing an intersection operation on the first abnormal power consumption index and the second abnormal power consumption index, and outputting the abnormal power consumption index; The correlation verification model is built based on historical power consumption behavior and equipment correlation, and is used to analyze whether the power consumption anomaly of the energy meter is associated with the normal power consumption pattern; Inputting the abnormal power consumption index into the associated verification model for verification and judgment, including: Collecting real-time power consumption associated data of the electric energy meter to be identified within a predetermined period, inputting the real-time power consumption associated data into the associated verification model, and outputting an associated load curve and associated power consumption; The abnormal power consumption index is judged based on the associated load curve and the associated power consumption, and if the deviation exceeds a preset threshold, a power theft warning signal is generated.

2. The method for automatically identifying the operation risk of an electric energy meter according to claim 1, characterized in that: Performing power consumption behavior analysis based on the historical power consumption data to construct a first power consumption comparison space includes: Collect historical electricity consumption data of the electric energy meter to be identified, wherein the historical electricity consumption data includes historical electricity load curve, historical electricity consumption, historical equipment start and stop time and historical equipment start and stop frequency; Arrange the historical power load curve and the historical power consumption according to the collection timestamp to generate a historical load curve sequence and a historical power consumption sequence; The long short-term memory network is trained with the historical load curve sequence and the historical power consumption sequence to obtain a load curve prediction branch and a power consumption prediction branch, which are fused to generate the first power consumption comparison space.

3. The method for automatically identifying the operation risk of an electric energy meter according to claim 2, characterized in that: Build an electricity comparison library, including: Using load monitoring technology to identify the historical equipment start and stop time and historical equipment start and stop frequency, and determine the type of electrical equipment; Taking the type of electrical equipment, building specification characteristics and historical segmented power consumption average as restriction conditions, multiple electric energy meters in a predetermined area are screened for the same type, and multiple similar electric energy meters that meet the predetermined similarity constraints are obtained to construct the power consumption comparison library.

4. The method for automatically identifying the operation risk of an electric energy meter according to claim 1, characterized in that: Constructing a second electricity usage comparison space based on the compared electricity usage data set includes: Selecting first electricity consumption data from the compared electricity consumption data set, wherein the first electricity consumption data includes a first load curve and a first electricity consumption; Performing a similarity comparison between the first load curve and other load curves in the comparison power consumption data set, and counting the amount of data whose similarity is greater than a predetermined similarity threshold, which is recorded as a first criticality; If the first criticality is greater than a predetermined critical threshold, a deviation calculation is performed between the first power consumption and other power consumptions in the compared power consumption data set, and a data volume with a deviation value less than a predetermined deviation threshold is counted and recorded as a second criticality; If the second criticality is greater than a predetermined critical threshold, the first electricity consumption data is added to a high-frequency electricity consumption data set, and a mean operation is performed on the high-frequency electricity consumption data set, and the second electricity consumption comparison space is constructed according to the operation result.

5. The method for automatically identifying the operation risk of an electric energy meter according to claim 1, characterized in that: Build an associated validation model, including: Based on the electricity consumption related indicators, query the electricity consumption monitoring log to obtain a sample electricity consumption related indicator set, a sample electricity load curve set and a sample electricity consumption set; The first correlation analysis channel is obtained by training the sample electricity consumption related indicator set and the sample electricity consumption load curve set, and the second correlation analysis channel is obtained by training the sample electricity consumption related indicator set and the sample electricity consumption set; The first association analysis channel and the second association analysis channel are integrated to generate the association verification model.

6. An automatic identification system for electric energy meter operation risks, characterized in that: The steps for implementing the method for automatically identifying the operation risk of an electric energy meter as claimed in any one of claims 1 to 5 include: A first electricity usage comparison space construction module: collects historical electricity usage data of the electric energy meter to be identified, performs electricity usage behavior analysis based on the historical electricity usage data, and constructs a first electricity usage comparison space; A second electricity usage comparison space construction module: constructs an electricity usage comparison library, reads a comparison electricity usage data set from the electricity usage comparison library within a predetermined period, and constructs a second electricity usage comparison space based on the comparison electricity usage data set; Anomaly identification module: reads the real-time power consumption data of the electric energy meter to be identified within a predetermined period, uses the first power consumption comparison space and the second power consumption comparison space to perform abnormality identification on the real-time power consumption data, and determines the abnormal power consumption index; Verification and judgment module: if the abnormal power consumption index is not zero, the abnormal power consumption index is verified and judged using the associated verification model; if the verification fails, the power theft warning of the electric energy meter to be identified is executed.

Images