Fault diagnosis of an air-conditioning system using LS-SVM

Abstract

This paper describes fault diagnosis of an air-conditioning system for improving reliability and guaranteeing the thermal comfort and energy saving. To achieve this goal, we proposed a technique which is model based fault diagnosis technique. Here, a lumped parameter model of an air-conditioning system is considered and then characteristics of twelve faults are investigated in an air-conditioning system provided in passenger coach of an Indian Railway. Based on the variations of the system states under normal and faulty conditions of different degrees, the faults can be detected efficiently by using residual analysis method. The residual code is obtained through simple threshold testing of residuals, which are the output of a general scheme of residual generators. The pattern of residual is classified by using multi-layer LS-SVM classification. The diagnosis results show that LS-SVM classifier is effective with very high accuracy. © 2009 Springer-Verlag Berlin Heidelberg.