In order to overcome the existing difficulty in processing huge amount of data obtained by on-line detection, a new method is proposed in this article, where outlier detection algorithm is used to identify abnormal points in the data and obtain statistics pattern of the entire dataset. To satisfy the demand of partial outlier points, local outlier factor algorithm is used and data of transformer winding DC resistance experiment is processed. Outlier points in the dataset are identified successfully. Besides, the huge amount of data obtained by on-line detection is divided into several state parameters, anomaly coefficient is introduced and state parametr outlier algorithm is used to obtain the radar map of the electrical equipment’s fault and state parameter. Statistic pattern of relativity between fault and parameter is studied to provide reference for further equipment repair. This method is also applicable to the abnormal data detection and analysis of other types of transmission and transformation equipment. By applying this method to the test data of substation equipment of Guangxi Power Grid, 10 main transformers with abnormal direct-current(DC) resistor test data and 75 capacitor voltage transformers (CVTs) with abnormal dielectric loss test data were successfully detected. This method may bring positive significance and broad application prospect in promoting the operation and maintenance level of electrical devices.