On Wavelet Neural Network Based Fault Diagnosis Approach Of Hydraulic Pump And Experience

Hydraulic system is relied on by power transmission and servo control in more and more commercial equipment. Once the hydraulic system is broken- down, the economic losses will be enormous. The hydraulic pump is the"heart"of the hydraulic system, the pump's broken-down will lead to the whole hydraulic system unable to work normally. So the condition monitoring and fault diagnosis of the pump is crucial for the hydraulic system.Wavelet Neural Network who is based on wavelet analysis rationale is a sort of new style Neural Network. It incorporates the time-frequency character of the wavelet analysis and the self-study function of the Neural Network. In this paper, the Wavelet neural Network is applied in fault diagnosis of the pump. The results of test show that Wavelet neural Network applied in fault diagnosis will increase the accuracy and will have a good develop tendency.The main works in this dissertation are showed as follows.(1)The basic segment of the fault diagnosis system and some kind of hackneyed fault of rotary are expatiated. Common rotary fault diagnosis methods and their merits and drawbacks are introduced detailedly, and conclude the new development of the fault diagnosis.(2)The nature character of wavelet analysis is expatiated. Wavelet analysis method is used in the process of state signal of hydraulic pump. Through the multi-resolution decomposition, threshold value elimination and reconstruction, the high frequency noise in the system is excluded and the characteristic component of fault characteristic signal frequency band is abstracted.(3)The structure and the learning algorithm of BP Neural Network are analyzed exhaustively. The BP network is used in the approximation of function and fault diagnosis. There have a good emulation result by pretreatment the sampled signal and training the BP network.(4)The incorporation passes of the wavelet analysis and the Neural Network are introduced particularly. Elaborating the construct and learning algorithm of loose Wavelet Neural Network and compact Wavelet Neural Network.(5)The two types of Wavelet Neural Network are used in diagnosing the fault of the pump, from test analysis there have ideal diagnosis effects. Generalize the superiority of the Wavelet Neural Network.