Kernel Regression Residual Transformation And Its Application In Fault Diagnosis

Rolling bearing,gear and valve are the vulnerable parts of mechanical equipment for petrochemical,metallurgy,railroad and pipelines industries.The random noise generated under complex working condition,it has always been a problem that affects the accuracy of faults detection in these basic components using vibration signal processing methods.Therefore,a new method of signal decomposition and reconstruction and its diagnostic principle is presented in this paper,and this method is applied to the fault diagnosis of rolling bearing,gear and valve.The main contributions and research contents are given as follows:1.A new method called kernel regression residual transform method is presented.Aiming at the de-noising result needs to be improved in signal processing,combined kernel regression function with wavelet threshold and empirical mode decomposition iterative stopping criterion of modern signal processing analysis method,a new method called kernel regression residual transform method is presented.Compared with the traditional signal processing methods through the simulation signal,the new method has the advantage that different from the de-noising method of wavelet transform and the complete ensemble empirical mode decomposition in the signal processing.2.A new fault diagnosis method based on kernel regression residual transform and Hilbert envelope demodulation is presented.Collecting inner race,outer race and combine fault of rolling bearing and tooth and crack fault of gear vibration signal through the MFS-MG machinery fault simulator and QPZZ-? rotating machinery vibration analysis simulator.The purified signal is obtained by using the kernel regression residual transform,according to the characteristics of bearing and gear fault signal is modulated,and then fault features is obtained by using Hilbert envelope demodulation technique.Experimental result shows the effectiveness of this method.3.A fault diagnosis research of valve internal leakage based on kernel regression residual transform is carried out.The on-line leakage diagnosis system based on kernel regression residual transform method is established and effectively applied to internal leakage fault diagnosis of pneumatic ball valve in piping system in the internship enterprise.