Research On The De-noising Algorithm Of Strong Noises In The EMAT System

Electro-magnetic acoustic transducer (EMAT) is a new trend of the development of the nondestructive testing, and recently it has been wisely adopted in the railway department of our country. When it is applied, because of the bad condition of the surface of the wheel and the existing electromagnetic interference, the strong noises will appear so as not to realize the data analysis.This paper aims at studying the data of the system, and achieving an algorithm to process the data so that it can be saved after de-nosing, which includes the status of wheel tread.This thesis firstly provides introduction of the aim of detecting and characteristics of the signal, and the basic situation at home and abroad, and the work of this paper. Then it presents the principle of the adaptive filtering algorithm and its classification, and to solve the noise problem exiting in the detecting procedure, a changing step size of normalized least mean-square algorithm (NLMS) is adopted. Subsequently, the two methods of phase calculations are described; a method of the simple trigonometric function is designed to realize the phase calculation, which is combined with the Blackman window and spectrum characteristics of the data. After that, based on the features of noises and useful EMAT signal, a wavelet soft-threshold de-noising algorithm is presented. Finally, based on the final de-noising results, the three integrated algorithms combined several algorithms is designed, by means of phase calculation, wavelet-de-noising, signal amplification, and smoothing processing, and it realizes the aim that the useful data can be saved after de-nosing, which includes the status of wheel tread.Its feasibility has been proved through the experimental data and accounting data. This algorithm will play an important role in the application of EMAT system.