In view of the great pressure of operation and maintenance of power energy data collection terminal after putting into operation, an intelligent fault diagnosis system for power equipment terminal based on computer was developed. This paper proposes an Agent based on autonomy, promotility, cooperation and interactivity. It also provides a flexible interactive communication interface and load balancing optimization method. It can meet a variety of data transmission specifications and a large number of data service transmission requirements. The SA-KNN algorithm is used to analyze the power quality. Methods The operation data of maintenance electricity were collected. Learning and cluster analysis are carried out by using relevant data, and the calculation model of automatic check device failure is established. At the same time, the outlier points in the collected data are used to accurately locate the fault points of the power grid. This method can realize the comprehensive fault diagnosis of the power grid by using the interaction between the power grid and each component.