Study on fault diagnosis of hydraulic pump based on sphere-structured support vector machines

This paper proposes the algorithm of defect classification of sphere-structured suppot vecor machines, constituting the multi-breakdown sorter to carry on the hydraulic pump´s fault recognition, in accordance with the insufficient data sample from fault diagnosis. The results show that training the classifier only needs a small quantity of fault data samples in time domain and does need signal preprocessing applied for multi-fault recognition and diagnosis. it has the advantage of strong ability of fault classification in the few sample situation compared to BP neural network.