Research on Time Series Modeling by Genetic Programming and Model De-noising

Abstract:- In order to cast off the subjective assumptions of traditional methods for modeling, this paper brings forward the Genetic Programming (GP for short) algorithm to establish a reasonable system model dynamically for time series signal. Meanwhile, the approach of wavelet threshold is adopted to de-noising for the GP models. On the basis of these theories, the simulation experimentations about two instances are carried on. The results indicate that the threshold approach of wavelet de-noising for time series signal models take on better impacts, which can improve the GP models to some extent, and enhance the forecast precision of the model