Prediction of 10kV distribution feeder monthly outage rate based on decision tree

This paper proposes a new approach to predict the 10kV distribution feeder monthly outage rate. Based on investigation of existing IT systems in distribution companies, influential factors of 10kV distribution feeder outage are first selected according to availability. Correlation analysis is performed to determine the outage critical influential factors. A Decision tree with C4.5algorithm is constructed to predict feeder's monthly outage rate. Case studies on 131 10kV distribution feeders indicate that the proposed approach can reach 90% accuracy level, which can be used to assist in developing inspection and maintenance strategies.