Reliability Analysis Of Smart Meters Under Compound Stresses

Typical operating conditions of smart meters measuring accuracy is vital for electricity settlement of fairness and justice.As a key index of operation reliability and measurement accuracy of smart electricity meter,basic error is affected by multiple stresses and various noises,and different features of smart meters company product is different,to the sensitivity of different stress es also has difference.Therefore,it is difficult to evaluate the degradation trend of smart meters under multiple stresses.This thesis built Optimized K-nearest-Neighbor(OKNN)and the Variation-Bias Wiener Process with Hierarchical Bayesian(VWPHB)model to analyze and study the operation characteristics of smart meter under complex stress in typical operating environment.Smart meters of several companies in typical operating environm ent laboratories in Xinjiang are selected to verify its applicability.Firstly,this thesis introduces the wide application of smart meters,which leads to the economic benefits and research necessity of the accuracy and reliability of smart meters in the typical operation environment under compound stress es.Then it introduces the research on the operation characteristics and reliability of smart meters by domestic and foreign scholars,and points out the shortcomings of the current research situation.Secondly,the relationship between reliability and basic error of smart meters is introduced,and realizes data acquisition of smart electrici ty meter by introducing typical operating environment test platform.The classification and source mechanism of basic errors are analyzed,and the data characteristics and correlation between stresses and basic errors are further analyzed.Aiming at the noise points and outliers in the measurement process,an OKNN algorithm is proposed.The nonlinear weighted factors corresponding to each stress are introduced to detect outliers.The parameters of OKNN are optimized by grid search,and the clustering quality evaluation index Silhouette Coefficient and clustering evaluation criteria Calinski Harabaz Index were used to evaluate the outlier detection results,select the optimal solution comprehensively,and modify the detected outlier.Then,according to the obvious correlation between various stress data and basic errors,the VWPHB model was built based on Bayesian theory,and partial pooling mode was selected to fuse multiple stresses and basic errors.The influence factors of each stress in the model were parameterized to evaluate and explain the influence of each stress on the basic error.Finally,the basic error prediction model based on OKNN-VWPHB was constructed,and the degradation analysis was carried out by combining the data of multiple smart meters from three companies of the typical operating environment test platform in Xinjiang,and the effectiveness of VWPHB’s fusion of various stress factors was verified by comparing multiple Bayesian models;several traditional artificial intelligence algorithms are combined with OKNN algorithm,and the experimental results verify the superiority and applicability of the model based on OKNN outlier detection and correction and OKNN-VWPHB.Several indexes were selected to verify the stability and convergence of OKNN-VWPHB model,and the results of solving the model were used to provide reference and theoretical value for the fairness of electric energy measurement and the manufacturing and operation reliability of smart meters.