Discrimination of Measured Outliers of Ground Total Electric Field of DC Transmission Line Based on ARIMA and Exponential Smoothing Method

The ground total electric field of DC transmission line is one of the environmental impact factors of public concern. The total electric field is vulnerable to the environmental conditions, which leads to great fluctuation of measured results, and suspicious abnormal data(outliers) often appear. At present, the environmental impact is not fully taken into consideration in the theoretical model of total electric field, thus it is difficult to accurately discriminate those outliers. From the perspective of time series analysis, this paper proposes an outlier discrimination method based on auto regressive integrated moving average model (ARIMA) and exponential smoothing method(ETS). The numerical case analysis shows that the proposed method, by which the root mean square error (RMSE) of prediction is about 1 kV/m, can be adopted to discriminate all 12 outliers, while the classical mathematical statistics method can only be adopted to discriminate 3 outliers. This method can be adopted to effectively deal with the typical cases such as relatively stable data, fixed trend data and slowly but significantly fluctuant data series, especially the case of suddenly increased non-extreme value in data series. This method is more reasonable to discriminate outlier of measured results of total electric field than classical mathematical statistics. © 2023 Science Press. All rights reserved.