Parameter Estimation for Asymptotic Regression Model by Dynamical Evolutionary Algorithm
Hai-peng Ye, Hao Hu · 2010
The dynamical evolutionary algorithm (DEA) is a novel evolutionary computation technology, which is based on the theory of statistical mechanics. In this paper, an improved dynamical evolutionary algorithm (IDEA) with multi-parent crossover and differential evolution mutation is proposed and IDEA is applied to estimate parameters for asymptotic regression model for the first time. In order to confirm performance of our algorithm, IDEA is verified on six groups of actual data and several sets of random sampling data, and then how sampling range and data with Gaussian noise influence on the performance of IDEA is considered. Experimental results show that IDEA is a stable, reliable and effective method in parameter estimation for asymptotic regression model and it's robust to noise.