Electric energy meter fault prediction method based on decision tree algorithm

Abstract

The invention discloses an electric energy meter fault prediction method based on a decision tree algorithm. The electric energy meter fault prediction method comprises the following steps: acquiring relevant data of a user by using an acquirer and eliminating wrong data; analyzing whether electric energy meters which have abnormal environments and abnormal quality exist in specific power transformer users and low-voltage users within statistical time buckets; counting whether metering and electricity utilization abnormal events which meet requirements of fault judgment exist for the special power transformer users, and counting whether metering abnormal events which meet requirements of fault judgment exist for the low-voltage users; establishing an electric energy meter fault prediction model by using the decision tree algorithm according to manufacturers corresponding to the electric energy meter and information of failed batches and on the basis of electric energy meter metering fault information of a marketing system; judging whether the result is accurate or not; if the result is accurate, finishing the process; and if the result is inaccurate, continuing judging faults. By the electric energy meter fault prediction method based on the decision tree algorithm, failed electric energy meters can be found out accurately in time, so that workload of field investigation and metering fault treatment of an examination department is reduced.

Description

technical field [0001] The invention relates to a method based on a decision tree algorithm to predict the probability of failure of an electric energy meter during a period of time (month, season, half a year, year) when metering and power consumption are abnormal, and belongs to the field of electricity marketing and metering technology in the electric power industry. Background technique [0002] At present, since Anhui Provincial Electric Power Company has installed more than 25 million electric energy meters on site, and cannot analyze the operation status of electric energy meters in operation on site, it is found that electric energy meters with a high probability of failure can only be used in accordance with DL / T448-2000 " According to the requirements of "Technical Management Regulations for Electric Energy Metering Devices" and "Electrical Energy Meter On-Site State Inspection Program", regular sampling inspections should be carried out on 2% of Class I, II, and II...

AI Technical Summary

Problems solved by technology

[0002] At present, since Anhui Provincial Electric Power Company has installed more than 25 million electric energy meters on site, and cannot analyze the operation status of electric energy meters in operation on site, it is found that electric energy meters with a high probability of failure can only be used in accordance with DL / T448-2000 " According to the requirements of "Technical Management Regulations for Electric Energy Metering Devices" and "Electrical Energy Meter On-Site State Inspection Program", regular sampling inspections should be carried out on 2% of Class I, II, and III electric energy meters, and rotations will be arranged for unqualified sampling inspections, while 98% of Class IV electric energy meters , Class V electric energy meters do not conduct on-site sampling inspections, and lack practical means to analyze and predict the operating conditions of electric energy meters. At the same time, due to the existence of a large number of random sampling work, it will also cause a large amount of meter waste, and there are also some real faults The label was not replaced in time because it was not found in the sample

Therefore, the existing management regulations and status inspection schemes cannot meet the current business status of Anhui Electric Power Company, which restricts the development of metering business. It is urgent to apply new technical means to solve the above problems

Images