Prediction of Zone Area Line Loss Anomalies based on PCA and Improved CHAID Decision Tree

• Published in: 2019 34rd Youth Academic Annual Conference of Chinese Association of Automation (YAC) pp. 221 - 226
• Main Authors: Tao, Xiumei, Yu, Mingxing, Zhu, Jiazheng, Yu, Changle, Zhang, Shuo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2019
• Subjects: cision tee; Decision trees; gain evaluation; HAID; Indexes; line loss in station area; PCA; Power grids; Power supplies; Prediction algorithms; Principal component analysis; Propagation losses
• DOI: 10.1109/YAC.2019.8787604

Reduction of line loss plays an important role in economic operation of the power grid. In view of the huge losses brought by line loss of zone area power distribution network to the economy and life, this paper proposes a daily simultaneous line loss prediction system based on PCA and improved CHAID decision tree. Firstly, the influencing factors affecting line loss of zone area distribution network are analyzed by the PCA method, and the influencing factors are decomposed according to intelligence factor, lean factor and density factor, thus providing regional distribution of the important factors affecting line loss. Secondly, decision branch computing is made on the influencing factors using CHAID method and the improved CHAID method, and gain evaluation of the two methods is compared. Finally, the daily line loss data of the actual zone area is predicted by the established zone area line loss prediction system, which verifies reliability of the proposed model and the prediction accuracy is ideal. It thus provides ideas on power grid theory research and loss reduction strategies for the relevant engineering personnel.