In this paper, a new method of smart meter performance rating based on data mining method is proposed considering the lack of comprehensive performance evaluation of smart meters now. On the basis of the performance test data of a large number of smart meter samples, the representative test data is selected as the performance rating index. Then the principal component method is used to analyze the influence degree of different test points under a certain rating index, and the weight value of different test points is given, so the data matrix of each rating index is obtained. The k-means clustering analysis method is used to construct the comprehensive performance rating model of the smart meter, and the rating results are given for samples. Finally, selecting smart meter samples from three different manufacturers randomly and analyzing their distribution of rating results considered failure rate in the field.
The rating results of the samples verify this rating method of comprehensive performance to smart meters is reasonable and reliable.