Application Research Of Wavelet Transform On Denoising And Structural Dynamic Characteristic Analysis

Wavelet transform is a new signal-processing tool developed in resent years. Its ability of multi-resolution time-frequency analysis not only suits for stationary signals, especially suits for dynamic signals characteristic of non-stationary and transient. Based on Natural Science Foundation of Anhui Province and National 863 Foundation of China, this thesis performs a lot of research work on wavelet denoising and wavelet-based structural dynamic analysis, for the purpose of providing an advanced and effective tool for actual machine and structure, especially for robot's dynamic characteristic analysis.In the studying of wavelet denoising, we compare several popular wavelet denoising methods, and propose a selecting rule for wavelet denoising. We also apply wavelet denoising to solve two engineering problems and get satisfactory results. In the area of structural dynamic analysis, traditional FFT and time domain method have a variety of drawbacks. So we use wavelet method to extract impulse response function of system and compare it with FFT method. Through a lot of simulation examples, we can see that wavelet method is superior to FFT method and can obtain more accurate results. We also analysis and explain it in nature why wavelet has more advantages over FFT method. Step excitation is a very important excitation mode for industrial robot, because it reflects the actual working conditions of robot. In this thesis we present impulse response extraction by wavelet when using step input, and the results are quite satisfactory. Finally we validate the correctness and practicability of wavelet method in engineering application by a real structural experiment.