Abstract:

The presence of wild points in the measurement and control system, brings serious difficulties to the further processing of data. Since traditional methods cannot be completely exclude outliers and retain high frequency information signal, this paper presents a new method for excluding the outliers based on wavelet transform and moment theory. In this method, the original signal wavelet decomposition of lowfrequency components as the original signal, estimates and removes outliers with adaptive method, reconstructs signal after removing outlier, then obtains ideal signal. Comparing with traditional methods, the method not only retains the highfrequency signal components but also effective exclude outliers. The fixed threshold and adaptive threshold method are compared by Matlab simulation, the results show that the adaptive method is better.

Key words: wavelet transform, outliers removed, adaptive threshold