Determining Parameters in the Phase-Space Reconstruction of Multivariate Time Series on Genetic Algorithm

Abstract

Based the principle of minimizing average prediction error, in this paper genetic algorithm is adopted to determine parameters in the phase-space reconstruction of multivariate time series. First, the methods of phase-space reconstruction and multivariate time series prediction are introduced. Then the theory of genetic algorithm to select reconstruction parameters is given that chromosome coding is multi-parameters cascade binary string, fitness is average prediction error function and the optimal parameters combination is obtained through genetic operation. Finally, in Matlab2009b simulation environment, the algorithm is applied to confirm embedding dimensions and time-delays of Rossler coupling system, and the results show that the algorithm has high prediction precision and rapid calculating speed.