Risk Modeling of Transmission Line Defects and Forecasting Method Based on SVM

• Published in: 2020 5th Asia Conference on Power and Electrical Engineering (ACPEE) pp. 47 - 53
• Main Authors: Ye, Wanyu, Zeng, Yongbin, Li, Shaopeng, Luo, Minhui, Ye, Zhijian, Wang, Xinghua, Yuan, Haoliang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2020
• Subjects: Analytical models; defect index; Maintenance engineering; meteorological factor; Meteorological factors; Pearson correlation coefficient; Power transmission lines; Predictive models; support vector machine; Support vector machines; Training; transmission lines
• DOI: 10.1109/ACPEE48638.2020.9136454

Establishing indicators for evaluating and forecasting the overall defect status of transmission lines are of great significance for the operation and maintenance, while studying the impact of different external factors on the development process of transmission line defects is also important. Therefore, this paper proposes a transmission line defect risk model and a SVM based prediction method. By dividing the transmission line into several component according to its own structure, the definition of the defect risk index is given and used to evaluate the status of the transmission line defects, which is based on the defect severity quantification and the membership degree analysis of each component. The historical defect risk value samples are calculated based on the historical defect data, and then use the Pearson correlation coefficient to select the meteorological factors with large impact on the defect development process to each component of the transmission line. Construct training samples with related factors, build a RBFSVM based transmission line defect risk value prediction model, and predict the defect risk value of the transmission line in the future. An example analysis is conducted on the transmission lines of Guangdong power grid to predict the defect risk value of a transmission line in each month of 2018. The results show the feasibility and accuracy of the proposed model and forecasting method in this paper.