High-order Statistics Analysis And Family Feature Identification Of Electric Energy Meter Verification Data

Electricity billing is an important part of my country's electric power economy.Electric energy meters are used as electric energy measuring instruments in the electricity trade between users and power supply companies.The result of their measurement is an important basis for electricity trade settlement.The accurate metering of electric energy meters is conducive to maintaining the interests of both power supply and consumer,reducing economic disputes and ensuring the fairness of transactions.However,there are many manufacturers of electric energy meters on the market,and the measurement accuracy of electric energy meters differs due to differences in the selection of internal electronic components,circuit design,and manufacturing technology of the meters in the process of producing electric energy meters.In this context,this article uses the verification data of the electric energy meter as the basis to study the family characteristics of the electric energy meters produced by different manufacturers,and provides decision support for the power supply enterprises to purchase accurate and stable electric energy meters in the future.Electric energy meter measurement data and subsequent verification data provide the basis for data correction and provide data reference for manufacturers to improve energy meters.First of all,in view of the large amount of data and uneven distribution of electric energy meter verification data,according to the numerical characteristics of the verification data,the overall level and distribution probability density of the metering results of different manufacturers,the fusion clustering algorithm is used to divide it into different categories and obtain different The measurement error division of the electric energy meter produced by the manufacturer.Secondly,high-order statistics are introduced to conduct in-depth exploration of the overall sample of electric energy verification data and the samples in different categories after clustering,and obtain the specific distribution characteristics of measurement errors of different manufacturers' electric energy meters under different loads,such as error size,dispersion,symmetry,and The difference in extreme values describes the family characteristics of electric energy meters produced by different manufacturers.Then,according to the different sizes of the energy meter verification data statistics of different manufacturers,the particle swarm algorithm with adaptive weight change and the support vector machine are used to establish a family recognition model,and the high-order statistics of the verification data of the energy meter under different loads are used as the characteristic input.,To carry out family recognition.Finally,based on the Monte Carlo verification method,the Gaussian mixture model is used to fit the distribution of the electric energy meter verification data produced by different manufacturers,and the M-H sampling method is used to fit the distribution and then sample to generate a large amount of simulation data.The analysis and comparison of clustering and high-order statistics are carried out on the simulation data of the electric energy meter verification and the real verification data to verify whether the method in this paper analyzes the characteristics of the electric energy meter family is reasonable.The results show that the use of clustering methods and high-order statistics can effectively reflect the family characteristics of electric energy meters produced by different manufacturers,and the identification model established based on the statistical characteristics of the metering and verification data of different manufacturers can effectively identify the family of electric energy meters.