Wavelet Robust Filtering of Out-trajectory Data

摘要

Wavelet-based robust filtering of process data is proposed in order to reduce the influence of the outliers and noise in Out-trajectory data. We utilize the moving median filtering method to reject outliers in the original data and then combine wavelet de-noising method with empirical Wiener threshold to suppress noise. Simulation calculation and real engineering application has shown that the novel algorithm reliably preserves the information encapsulated in a process signal corrupted with noise and outliers. The methodology has been proved to be reliable and robust.