Power consumer electricity stealing risk rating algorithm based on fuzzy analytic hierarchy process

Abstract

The invention discloses a power consumer electricity stealing risk rating algorithm based on a fuzzy analytic hierarchy process. The algorithm comprises the following steps: step 1, index selection; step 2, determining an index weight; step 3, index scoring; step 4, model input; step 5, model operation; 6, outputting the model; comprehensively considering the index weight and the user behavior characteristics, carrying out risk rating scoring on the user, and forming an electricity stealing risk rating result; and 7, model feedback correction. According to the invention, power marketing data can be effectively utilized, power user behavior characteristics can be efficiently analyzed, power stealing users can be accurately positioned, and power stealing behaviors can be furthest treated.

Description

Technical Field

[0001] The invention relates to an electricity theft risk rating algorithm for power users based on fuzzy analytic hierarchy process for use in the field of anti-electricity theft in power systems. Background Art

[0002] At present, although power supply companies have taken a series of measures in terms of policies and management to prevent electricity theft and realize the full collection of user power marketing data, the application of electricity information data mining in the field of anti-electricity theft is not yet widespread due to the limitations of the depth of data mining and algorithm application. The existing electricity theft identification methods have problems such as incomplete identification types and difficulty in adapting to changes in user behavior. Therefore, a more effective electricity theft risk rating algorithm is needed to solve these problems. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a power user electricity theft risk rating algorithm based on fuzzy analytic hierarchy process, so as to effectively utilize power marketing data, efficiently analyze the behavioral characteristics of power users, accurately locate electricity theft users and rectify electricity theft behavior to the greatest extent.

[0004] A technical solution to achieve the above purpose is: a power user electricity theft risk rating algorithm based on fuzzy analytic hierarchy process, including the following steps:

[0005] Step 1, indicator selection;

[0006] Step 2: Determine the indicator weights;

[0007] Step 3, indicator scoring:

[0008] Step 4, model input:

[0009] Step 5, model operation;

[0010] Step 6: Model output: Comprehensively consider the indicator weights and user behavior characteristics, and rate the user's risk to form a risk rating result for electricity theft;

[0011] Step 7: Model feedback correction.

[0012] Preferably, step 1 specifically includes: selecting power consumption, line loss rate, and abnormal power consumption records as indicators for model analysis.

[0013] As a preferred embodiment, step 2 is specifically as follows: based on the expert industry experience and professional knowledge, the importance of each indicator in assessing the risk of electricity theft is subjectively judged and scored, and by constructing a fuzzy judgment matrix, the expert scores are fuzzy processed and calculated to obtain objective and reasonable weight values.

[0014] Preferably, step 3 is specifically as follows: scoring the power consumption, line loss rate, and abnormal power consumption records according to actual data and preset standards.

[0015] Preferably, the scoring of the power consumption is performed in different score intervals according to the statistics of the user's power consumption mean and standard deviation, and a lower score is given if the power consumption deviates from the mean by more than a certain range.

[0016] Preferably, the line loss rate is scored based on industry standards and historical data, and the line loss rate is divided into different levels of normal, slightly abnormal, and seriously abnormal, and corresponding scores are assigned.

[0017] Preferably, the scoring of the abnormal electricity consumption records is combined with the scoring of the specific abnormality type and frequency, and lower scores are given to frequent abnormalities.

[0018] Preferably, step 4 specifically includes: inputting the user's power consumption, line loss data and abnormal records.

[0019] Preferably, step 5 specifically includes:

[0020] Extraction of electricity theft features: Analyze the changing trend, periodicity and seasonality of electricity consumption, establish a time series prediction model, predict the normal electricity consumption trend, compare it with the actual electricity consumption, and use the deviation beyond the prediction range as the electricity theft feature;

[0021] Line loss and electricity theft feature extraction: Analyze the value, change rate and direction of line loss rate. A sudden increase in line loss rate or a sustained high line loss rate is a sign of electricity theft.

[0022] Extraction of electricity theft features from abnormal records: Correlation analysis is performed on different types of abnormal records to discover the electricity theft patterns hidden behind the abnormal records.

[0023] Preferably, step 7 is specifically as follows:

[0024] Indicator weight setting: fully consider the relationship between different indicators and the impact on the risk of electricity theft, and regularly re-evaluate and adjust the indicator weights;

[0025] Indicator scoring rules: Regularly evaluate the designated indicator scoring rules based on model output and customer actual situation to ensure the objectivity and consistency of the rating.

[0026] The power user electricity theft risk rating algorithm based on fuzzy analytic hierarchy process of the present invention is as follows:

[0027] Beneficial effects:

[0028] (1) The present invention adopts the fuzzy hierarchical analysis method, which can cope with the uncertainty and complexity of user data indicators and more accurately assess the risk of user electricity theft.

[0029] (2) Comprehensively consider multiple indicators such as power consumption, line loss rate and abnormal power consumption records to comprehensively assess the user's risk of power theft.

[0030] (3) Through a scientific weight determination and scoring mechanism, the accuracy and feasibility of electricity theft risk rating are improved. DETAILED DESCRIPTION

[0031] In order to better understand the technical solution of the present invention, the following is a detailed description through specific embodiments:

[0032] The power user electricity theft risk rating algorithm based on fuzzy analytic hierarchy process of the present invention comprises the following steps:

[0033] Step 1, indicator selection:

[0034] Electricity consumption: reflects the scale of electricity consumption of users and reveals potential anomalies through its changing trends. For example, a user with stable electricity consumption suddenly experiences a large increase or decrease, which may indicate electricity theft. At the same time, analyze the periodic and seasonal characteristics of electricity consumption to more comprehensively evaluate the electricity consumption behavior of users.

[0035] Line loss rate: It is an important indicator to measure the efficiency of power transmission. A high line loss rate may be caused by power theft. In addition to analyzing the numerical value of the line loss rate, we also pay attention to its change rate and direction. A sudden increase in line loss rate or a continuous high line loss rate may be a signal of power theft.

[0036] Abnormal electricity usage records: including overdue electricity bills, sudden increase in electricity consumption, abnormal electricity equipment, etc. By analyzing the peak and trough of electricity consumption of users in different time periods, and comparing them with normal electricity consumption patterns. If there is an electricity usage pattern that is inconsistent with the industry, season, user type, etc., it should be highly concerned. For example, a user's electricity consumption suddenly increased significantly in winter, and it was mainly concentrated at night, which was significantly different from the user's previous electricity usage pattern and the electricity usage pattern of other users in the same industry. This may be a manifestation of abnormal electricity usage records.

[0037] Step 2: Determine the indicator weights:

[0038] Based on the experts’ industry experience and expertise, they subjectively judge and score the importance of each indicator in assessing the risk of electricity theft. The experts will consider various factors, such as the degree of correlation between different types of electricity theft and each indicator, and the performance of each indicator in actual cases.

[0039] Construct a fuzzy judgment matrix, perform fuzzy processing and calculation on the expert scores to reduce the influence of subjective factors and obtain more objective and reasonable weight values. The fuzzy judgment matrix can handle the fuzzy relationship between indicators and make the determination of weights more scientific.

[0040] Step 3, indicator scoring:

[0041] Electricity consumption: Different score ranges are set based on the user's electricity consumption mean, standard deviation and other statistics. If the electricity consumption deviates from the mean by more than a certain range, a lower score will be given. At the same time, considering the changing trend of electricity consumption, if the electricity consumption shows an abnormal fluctuation trend over a period of time, a corresponding low score will also be given.

[0042] Line loss rate: Based on industry standards and historical data, the line loss rate is divided into normal, slightly abnormal, and severely abnormal levels, and the corresponding scores are assigned. For line loss rates that are close to or exceed the severely abnormal level, a lower score will be given.

[0043] Abnormal electricity usage records: Scoring is based on the specific abnormal type and frequency. For example, users who frequently have high electricity consumption at night and overdue electricity bills may receive a lower score.

[0044] Step 4, model input:

[0045] Electricity consumption: Input the user's electricity consumption data, including electricity consumption data for different time periods such as daily, weekly, and monthly, and analyze their electricity consumption patterns and scale as an important basis for evaluating their electricity consumption behavior.

[0046] Line loss data: Analyze line loss data, not only focusing on the value of the line loss rate, but also analyzing the changes in the line loss rate in different time periods, identifying line loss anomalies, and providing a basis for power theft risk rating. Line loss anomalies may indicate potential power theft.

[0047] Abnormal records: Summarize the user's historical abnormal electricity usage records, including the specific number and time of overdue electricity bills, the magnitude and frequency of sudden increases in electricity consumption, the type and time of abnormalities in electrical equipment, etc.

[0048] Step 5, model calculation:

[0049] Electricity theft feature extraction: When extracting electricity theft features from electricity consumption data, in addition to focusing on the absolute value changes in electricity consumption, the changing trend, periodicity, and seasonality of electricity consumption are also analyzed simultaneously. For example, by establishing a time series prediction model, such as the ARIMA model, the normal trend of electricity consumption is predicted, and then compared with the actual electricity consumption, and the deviation beyond the prediction range is used as the electricity theft feature. At the same time, the periodic characteristics of electricity consumption are analyzed. For example, if the electricity consumption of some users fluctuates abnormally during a specific period of time each week or month, it may be a sign of electricity theft.

[0050] Line loss and electricity theft feature extraction: For line loss data, in addition to analyzing the numerical value of the line loss rate, we also analyze the rate and direction of change of the line loss rate. A sudden increase in the line loss rate or a sustained high line loss rate may be a signal of electricity theft. In addition, we also pay attention to whether the trend of the line loss rate changes is consistent with the trend of the user's electricity consumption. If there is a significant inconsistency between the two, it may also indicate the existence of electricity theft.

[0051] Extraction of electricity theft features from abnormal records: Correlation analysis is performed on different types of abnormal records. For example, frequent overdue electricity bills and sudden increases in electricity consumption may indicate that the user is stealing electricity. By establishing an association rule mining model for abnormal events, the electricity theft patterns hidden behind the abnormal records can be discovered. For example, it is found that some users frequently have overdue electricity bills and sudden increases in electricity consumption in a certain season or time period, which may be a potential electricity theft pattern.

[0052] Step 6, model output:

[0053] Indicator scoring: Comprehensively consider the indicator weights and user behavior characteristics, and conduct risk rating and scoring of users to form the risk rating results of electricity theft. In the scoring process, accurate scoring will be conducted based on the specific situation of the user, combined with the indicator weights and scoring rules, to obtain objective risk rating results of electricity theft.

[0054] Step 7, model feedback correction:

[0055] Indicator weight setting: Fully consider the relationship between different indicators and the degree of impact on the risk of electricity theft. For example, there may be a certain correlation between electricity consumption and line loss rate, and repeated calculations need to be avoided when setting weights. At the same time, with the changes in the electricity market and the development of technology, the indicator weights should be re-evaluated and adjusted regularly to ensure the timeliness and accuracy of the model. For example, if the power system is upgraded and the normal range of the line loss rate changes, the corresponding indicator weights also need to be adjusted.

[0056] Indicator scoring rules: Develop detailed indicator scoring rules, including scoring standards and methods, to ensure the objectivity and consistency of the rating. The scoring rules will clearly specify the scores of each indicator in different situations, such as the specific degree of deviation of electricity consumption from the mean corresponding to different score levels.

[0057] In order to verify the accuracy of the actual use of the user electricity theft risk rating algorithm model, we conducted the following algorithm model verification. From the output results of the algorithm model for electricity users in a certain district in Shanghai, screen out users who are judged to have a high suspected risk of electricity theft. Dispatch inspectors to conduct on-site inspections of these highly suspected users. Focus on electricity meters, lines, electrical equipment, etc., and comprehensively collect detailed information and strong evidence of actual local electricity consumption. The results of the on-site inspection are reviewed one by one with the prediction results of the algorithm model for the electricity theft risk rating of electricity users in a certain district in Shanghai. The comparison factors include actual electricity consumption, line loss conditions, electricity consumption patterns, etc., to determine whether they are consistent with the electricity theft risk characteristics predicted by the model.

[0058] After comparative analysis, it was found that the on-site verification results were basically consistent with the risk rating results of electricity theft by the algorithm model. Most of the electricity users in a certain district of Shanghai who were identified as having high risk of electricity theft by the model were indeed confirmed to have different degrees of electricity theft or abnormal electricity usage during the on-site verification.

[0059] This verification strongly proves that the user electricity theft risk rating algorithm model constructed in this study is accurate and reliable in Songjiang, Shanghai, and can effectively identify local potential electricity theft risk users, providing solid technical support and scientific decision-making basis for the anti-electricity theft work of Shanghai Songjiang Power Supply Company.

[0060] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.

Claims

1. The risk rating algorithm of electricity theft by power users based on fuzzy analytic hierarchy process is characterized by: The following steps are involved: Step 1, indicator selection; Step 2: Determine the indicator weights; Step 3, indicator scoring: Step 4, model input: Step 5, model operation; Step 6: Model output: Comprehensively consider the indicator weights and user behavior characteristics, and rate the user's risk to form a risk rating result for electricity theft; Step 7: Model feedback correction.

2. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 1 is specifically as follows: selecting power consumption, line loss rate, and abnormal power consumption records as indicators for model analysis.

3. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 2 is as follows: Based on the experts' industry experience and professional knowledge, the importance of each indicator in assessing the risk of electricity theft is subjectively judged and scored. By constructing a fuzzy judgment matrix, the expert scores are fuzzy processed and calculated to obtain objective and reasonable weight values.

4. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 3 is as follows: according to the actual data and preset standards, the power consumption, line loss rate and abnormal power consumption records are scored.

5. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 4 is characterized in that: The electricity consumption score is set in different score ranges according to the statistics of the user's electricity consumption mean and standard deviation. A lower score is given when the electricity consumption deviates from the mean by more than a certain range.

6. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 4 is characterized in that: The line loss rate is scored based on industry standards and historical data, and the line loss rate is divided into different levels: normal, slightly abnormal, and severely abnormal, and corresponding scores are assigned.

7. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 4 is characterized in that: The abnormal electricity consumption records are scored in combination with the specific abnormality type and frequency, with lower scores being given to frequent abnormalities.

8. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 4 specifically includes: inputting the user's power consumption, line loss data and abnormal records.

9. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 5 specifically includes: Extraction of electricity theft features: Analyze the changing trend, periodicity and seasonality of electricity consumption, establish a time series prediction model, predict the normal electricity consumption trend, compare it with the actual electricity consumption, and use the deviation beyond the prediction range as the electricity theft feature; Line loss and electricity theft feature extraction: Analyze the value, change rate and direction of line loss rate. A sudden increase in line loss rate or a sustained high line loss rate is a sign of electricity theft. Extraction of electricity theft features from abnormal records: Correlation analysis is performed on different types of abnormal records to discover the electricity theft patterns hidden behind the abnormal records.

10. The electricity theft risk rating algorithm based on fuzzy analytic hierarchy process according to claim 1 is characterized in that: Step 7 is as follows: Indicator weight setting: fully consider the relationship between different indicators and the impact on the risk of electricity theft, and regularly re-evaluate and adjust the indicator weights; Indicator scoring rules: Regularly evaluate the designated indicator scoring rules based on model output and customer actual situation to ensure the objectivity and consistency of the rating.