Research On The Method Of State Evaluation And Prediction Of Smart Meter

As an important part of the smart grid,the operating status of smart meters affects the reliability and stability of the power supply of the entire grid directly.According to the relevant regulations,smart meters must be verified by the professional meter verification center before being put into use,which is the first mandatory verification to confirm its measurement accuracy and operational reliability.But the measurement accuracy and operational reliability of each smart meter will decrease in different degrees after a period of operation.The thesis mainly studies the state evaluation of smart meters and the prediction of the unhealthy degree of smart meters.Based on the results of the state evaluation,assist the power company to remotely manage the health status of smart meters in operation.Mastering the operation health state of smart meters can provide electric power company with the objective reference of timely repairation,replacement and purchase.The main contents are as follows:(1)The reliability index selection of smart meters.Aiming at the four dimensions including clock abnormality evaluation of smart meters,operation failure evaluation of smart meters,reliability evaluation of smart meters and comprehensive non-health evaluation of smart meters,we study the corresponding specific items in detail.(2)The process of assessment and analysis of the state of smart meters.We summarize the reasons for status assessment,then clean and filter dirty data by using relevant data cleaning techniques;Combining with historical data and using the entropy weight method,we establish a non-health model of smart meters,calculate the information entropy of each index item of the status assessment,determine the weight of the assessment and calculate the unhealthy value of smart meters.(3)The prediction of non-health value of smart meters.We study the characteristics of linear regression prediction model and ARIMA prediction model,and their respective modeling process.We predict the unhealthy value of smart meters by using two prediction models respectively,and then compare the two prediction results with the real results.