Failure decision-making based on contracted support vector machine for indiscernible system

Due to inherent delivery fluctuation of piston pump, its measurable signals are full of structure coupling and noise besides failure feature that make the system illegible and fault diagnosis difficult. This paper presents a contract support vector machine to extract the effective information from data and eliminate the redundant attribute among different data. Then utilize the support vector machine to classify the failures effectively on condition of limit samples. Application of piston head looseness indicates that the contract support vector machine not only can decrease the calculation of feature extraction but also can classify the failures effectively under limit samples.