Echo state network based on improved fruit fly optimization algorithm for chaotic time series prediction

Abstract

Chaos is a common phenomenon in nature and society. Chaotic system affects many fields. It is of great significance to find out the regularity of chaotic time series from chaotic system. Chaotic system has extremely complex dynamic characteristics and unpredictability. The traditional prediction methods for chaotic time series have some problems, such as low accuracy, slow convergence speed and complex model structure. In this paper, an echo state network prediction method based on improved fruit fly optimization algorithm for chaotic time series is proposed. The phase space reconstruction is introduced for the prediction of chaotic time series. The C–C method is used to determine the delay time. The embedding dimension is obtained by the G–P method. After reconstructing the phase space of the chaotic time series, an improved echo state network is proposed as the prediction model. In order to improve the prediction accuracy, an improved fruit fly optimization algorithm is proposed to optimize the parameters of the prediction model. Three typical chaotic time series, including Lorenz, Mackey–Glass, and short-term wind speed, are selected as simulation objects. The simulation results show that the prediction method proposed in this paper has good prediction indicators. At the same time, the results of the reliability and Pearson's test also show the better predictive effect.