Hybrid approach based on immune algorithm and support vector machine and its application for fault diagnosis of hydraulic pump

Abstract

This paper describes a hybrid fault diagnosis approach that combines the real-valued negative selection(RNS) algorithm and the support vector machine(SVM) and its application for fault diagnosis of hydraulic pump because it is very difficult to gain the fault samples in the fault diagnosis process of hydraulic pump. In this method, the RNS algorithm is used to generate the nonself set as the fault samples, which are used as the input to SVM algorithm for training purpose. The problem of lacking the fault samples is solved by using this new method. It is accomplished to eliminate the noise existing in the measured signals of hydraulic pump and pick up its features using the wavelet analysis method. Finally, the hydraulic pump fault samples are tested by using the hybrid approach. The classification right rate by this method is 90%, so it is valid for the fault diagnosis of Hydraulic Pump.