Predicting Residual Useful Lifetime of Circuit Breakers Using K-Medoids and Holt-Winters Algorithms

• Main Authors: Liao,Chun-Yen, 廖俊彥
• Other Authors: Chung,Kuan-Jung
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/9242a7

碩士 === 國立彰化師範大學 === 機電工程學系所 === 105 === ABSTRACT The purpose of this essay is to focus on the prediction of the circuit breakers’ residual useful lifetime. Now the circuit breakers’useful lifetime is provided by the reliability test which is set up before leaving the factory. Its test conditions are mostly fixed. Such as temperature; electric pressure; are all set up in fixed numerical numbers. But products in reality are changeable in different environment conditions, so the estimates lifetime will be not coincide with the real lifetime. Therefore, the essay will use the insulation resistance value during the electric power cut over the years to classify and predict so as to tell customers to renew their equipments in advance. Hope to reduce customers’operating damage and raise both the electric quality and the reliability in the products. Keywords：Circuit Breakers, Machine Learning, Unsupervised Learning, K-Medoids, Holt-Winters