Suggested formula to estimate relative smoothed grade for contaminated data in spectral analysis through using robust generalized M-estimates of ( Thomson & Al-Naqeeb )

Abstract

The aim of this study is to measure the efficiency coefficients in terms of prediction errors of time series in the presence of contaminators of the two types, Innovations and Additive outliers by application of the Robust generalized M-estimates for two methods: “Thomson” &“AL-Naqeeb” through autoregressive model AR(p), and introduces a suggested formula which enable to determine the relative smoothed grade. Where as two objectives were obtained, the 1st is to give an indicator to enable and determine the relative changes levels of progresses grade which were occurrences on the data according to the robust filtering process and the 2nd can be used as an initial indicator to determine the significant level of the robust methods under compromised and through determining the grade of covering the lowest level of process which would be needed and then take off the decision for testing the suitability of the robust methods for obtaining the needs estimates and then to estimate the power spectra. The application view of the two efficient indicators in terms of predication errors and the suggested relative smoothed grade was in the analysis of (EMG) signal according to constraint simulation technique for the pre whitening signal {ε (z)}.

Keywords

M-estimates