A Study Of Signal De-noising Based On Wavelet Transform

The theory of wavelet analysis achieves the attention in all over the world of late years. It is also a pop-researching field internationally and arouses extensive attention and upstairs recognition in science and technology field. The development of wavelet analysis forces the development of many others study and field. Typical signal processing mechanism, such as pure time domain, pure frequency domain, Short term Fourier transform, wigner-ville distribution and so on have their own limitations. As the result, their application extents are restricted.As a new signal processing method, Wavelet analysis decomposed different kinds of frequency element to non-overlapped frequency bands and this method put forwards an effective way to signal filter, signal-noise separating and character picking-up, especially it has well de-noising capability. This paper introduced the classic de-noising method and as to it's applied of scope and results to carry on the analysis and comparison. Then studied the wavelet transform theories, analyzed the characteristics of the wavelet transform, aim at the non-stable signal, the system introduced the common use wavelet de-noising methods: maximum de-noising method, coefficient of high frequency place to zero method and threshold de-noising method. The threshold de-noising includes the method of soft threshold de-noising, hard threshold de-noising and soft-hard threshold de-noising.This paper analyzes the arithmetic of these three kinds of methods and come true through the experiment simulations, it analyzed the results of the experiment.Based on the threshold de-noising, this paper puts forward a de-noising method that combines mean approximation and threshold together and come true through the experiment simulations. The result indicates that it increases the SNR, de-noising effect is better than the threshold de-noising only.In spite of the limited samples, the simulation result shows that using the wavelet analysis to signal de-noising is an efficient way of fault location.