The work described in this paper investigates the fault diagnosis of water hydraulic motor by the optimization and automatic classification of the feature values. The second generation wavelet for the vibration signals analysis of the water hydraulic motor was proposed to extract the feature values. The new optimization method by bi-classification support vector machine (SVM) was proposed to select the optimal feature values based on a rank criterion and the algorithm was developed here. In order to classify the conditions of the pistons used in the hydraulic motor, a two-level structure based on the multiclassification was developed in this work. The multi-classification method of SVM for the fault diagnosis of a piston crack was investigated. The winner-takes-all scheme was studied. The results of the classification were found to be successful.

Keywords:Fault diagnosis, Support vector machine, Water hydraulic system, Feature extraction, Wavelet transform, Neural network