Study On Wavelet Transform Applications To Fault Diagnosis And Measurement Of Electric Machine

With the development of modern science and technology, electric machines play a more and more important role in modern industrial plants. The risk of machine failing can be remarkably reduced if normal service conditions can be arranged in advance. In other words, one may avoid very costly expensive downtime of plant by proper time scheduling of machine replacement or repair if warning of impeding failure can be obtained in advance.As the fault signals are non-stationary transient ones, the traditional signal analysis methods, such as Fourier Transform, are not so efficient and useful for the fault signal detection. However, Wavelet Analysis has the excellent time-frequency local performance, it can detect the different frequency components of the fault signals by its adjustable time-frequency window. In view of the superiority of Wavelet Transform to non-stationary signals, this paper focuses on how to detect and analyze the fault signals and the electric machine measurement signals by Wavelet Transform.This dissertation contains the following parts:The extraction algorithm for instantaneous frequency of asymptotic signals based on wavelet ridge is proposed in this dissertation, and the asymptotic estimation method for wavelet transform of asymptotic signals is deeply studied. The algorithm is also applied in the broken rotor bars detection in squirrel cage induction machines, which is dependent on detecting the twice slip frequency modulation due to the speed or torque in the stator current, and the rotor bar faults can be detected effectively without some preprocessing.The feasibility of using Mallat algorithm to extract the feature component of rotor bar faults is discussed, and we find that Mallat algorithm is not suitable for rotor bar faults detection because of the similar representations of the fault and the normal signals. The paper point outs that how to distinguish load oscillation and rotor bar faults is still the essential problem up to now, and a method based on Wavelet Transform to solve the problem is proposed. A fast algorithm for Wavelet Transform is also proposed and applied in the extraction of the phase change rate for complex wavelet transform of stable-state stator current, which can detect broken rotor bar faults effectively.At last the floating threshold algorithm based on wavelet packet transform is proposed to remove the noise from the signal of DC motor measurement in the dissertation, which is proved effectively by the experimental results.