This paper presents a new method of determining the causes of faults by using decision tree analysis. These days, the high economic growth in Japan has led to a rapid increase in the number of customers receiving high voltages (i.e., > 50kW electric facilities). Therefore, the maintenance of these facilities has become important from a power quality point of view, and electric facilities for private use install an insulation level monitoring device to detect the leakage current from an electrical fault. The insulation level monitoring device detects the leakage current based on threshold current of 50mA. Usually, faults are detected by combining analytical elements of the threshold or majority decision method. However, these methods cannot achieve high accuracy without the optimal threshold being defined.
In this paper, we propose a new method for determining the causes of faults by using decision tree analysis with an insulation level monitoring device. We compared the differences in the accuracy among the threshold method, majority decision method, and decision tree analysis. We found that decision tree analysis can determine the optimal threshold to classify the fault cause. By repeating this method in multiple stages, the optimal threshold of each fault cause can be determined more accurately.
J-STAGEがリニューアルされました！https://www.jstage.jst.go.jp/browse/-char/ja/
