Hydraulic system fault diagnosis method based on a multi-feature fusion support vector machine

Abstract

Real-time fault prevention and diagnosis in hydraulic systems is one of the challenges in engineering applications. In order to solve the problems of feature extraction and pattern recognition of different types of fault signals such as leakage, blockage, and cavitation in the hydraulic system, this paper presents a fault diagnosis method based on a multi-feature fusion support vector machine. This method first extracts the typical features of the fault signal from the time domain, frequency domain, and time−frequency domain. Then use the information entropy to calculate the weight of each feature, select the feature with the larger weight to participate in the stock index diagnosis. Finally, a multi-classifier based on the support vector machine is used to realise the fault diagnosis of the hydraulic system. The experimental results show that this method can achieve high accuracy in the hydraulic system fault diagnosis when the Gaussian radial basis function is selected. Also, compared with the commonly used improved back-projection neural network fault diagnosis method, this method has more excellent diagnostic performance.

Keywords

• support vector machines
• fault diagnosis
• radial basis function networks
• hydraulic systems
• entropy
• feature extraction
• pattern recognition
• signal processing
• mechanical engineering computing
• hydraulic system fault diagnosis method
• multifeature fusion support vector machine
• feature extraction
• pattern recognition
• fault signal
• stock index diagnosis
• back-projection neural network fault diagnosis method
• time-frequency domain