Trend Extraction by Difference Method for Least Square Combining with EMD
Yuhua Dong, Yanhui Sun - · International Journal of Advancements in Computing Technology · 2012
A difference method to order decision for least square in trend fitting is proposed in this paper, combining with EMD, which can extract the trend items. The signal to be analyzed is decomposed by EMD, and the correlation method is used to determine the IMF which probably contains the trend item. Then for such IMF, the least square is used for fitting trend, and the order number of the least square is determined by difference method. The method solves the order decision problem of least square in trend fitting effectively, and combining with EMD, the trend extraction has high accuracy. Simulation processing results and the test data show that this method is effective in trend extraction. Keywords; Least Square, Trend Fitting, Order Decision, EMD