Fault Diagnosis Of Rotating Machineries Based On The Extraction Of Weak Signal

Rotating machinery has been widely applied in many actual industrial engineering, for example aircraft, gas turbine, wind turbine etc. Some faults may occur when the machines rotate, which bring huge economic losses, so it is necessary to study the fault diagnosis method of signal. The signal should be detected before the fault occurs or the fault is weak, so that to reduce the losses to lowest. Thus, the research on the extraction method of weak signal has become central issue of concerns in engineering areas, which attracts many researchers all over the world.The common methods of weak signal processing is reviewed first, for example the method based on chaotic oscillator, the wave analysis method, compressed sensing(CS) method and so on. The CS method of dictionary learning based on the theories of sparse dictionary and CS is proposed.Then simulation of bearing fault is studied on the basis of stagewise matching morphological component analysis(SMMCA) of sparse decomposition theory. The impulse signal caused by bearing fault is extracted between harmonic signal and noise successfully, the efficiency of the signal processing method is verified at the same time. We do experiment research on actual bearing faults of cars and aircraft engine respectively. The actually weak signal fault is extracted from noise successfully. The efficiency of the SMMCA method is verified from the aspect of experiment.At last, the extraction method of weak signal is applied in the rotor-bearing system. The 23 degrees of freedom(DOF) rotor system with pedestal looseness is established by the Newton’s second law, the fault signal component is extracted by the SMMCA method. The proper orthogonal decomposition(POD) is used to reduce the 23-DOFs model to a 2-DOFs one. The reduced model reserves the bifurcation and amplitude-frequency characters of the original one. These results not only manifest the applied value of the extraction method of weak signal, but also the efficiency of the reduced method.