An Improved Interval-halving Algorithm for Qualitative Trend Extraction
Rui Zhou · Keisou gijutsu · 2011
Step signals comprise the important fault information.A polynomial-fit based interval-halving algorithm in qualitative trend analysis can automatically identify the qualitative shapes of process data trends.When the parameter data comprise step signals,the interval-halving algorithm has some problems such as being hard to partition the window size felicitously,inaccurate of trend extraction.This paper improves the interval-halving algorithm for trend extraction.Data are divided into two parts by data preprocessing,one comprises step signals only and the other comprises non-step signals.The method that discriminates the effectiveness of interval-halving algorithm is also given.The simulation results show that the improved algorithm is correct.