Health status identification of connecting rod bearing based on support vector machine

Abstract

Connecting rod bearing (CRB) is an important component which joins the reciprocating and rotating movements together in the internal combustion engine (ICE). It is very difficult to identify health status of CRB because of variable working process, complicated excitation and distribution sources, and lack of fault samples. Support vector machine (SVM), which has excellent capability in small data case, was introduced to identify the health status of CRB. In this paper, faults of the CRB were simulated in an ICE with the type of EQ6100. Vibration features were extracted from vibration signals acquired from the shell of ICE. And a SVM multi-classifier was designed to identify health status of CRB by using the radial basis kernel function. Experimental results indicated that the presented fault diagnosis method could effectively recognize different conditions of CRB. © 2011 IFIP International Federation for Information Processing.