Fault forecast of electronic equipment based on ε-SVR

Abstract

In order to ensure security and reliability of the equipment, so as to decrease the maintenance cost, combining with the characteristics of fault data, this paper adoptsε-support vector regression to establish a fault forecast model and evaluation system to prediction model effect which are proper to the electronic equipment. Selecting multi-electronic equipment and training on the ε-SVR with different kernel functions. It is demonstrated that the prediction effect is better and it is still of vital realistic significance for realizing conditionbased maintenance of modern electronic equipment. © Springer-Verlag Berlin Heidelberg 2012.