Prediction of chaotic time series of neural network and an improved algorithm

This paper mainly proposes a neural networks model with hybrid algorithm, named HAENN (Hybrid Algorithm Elman Neural Network). This model is based on Elman neural network, using an improved algorithm instead the standard BP training algorithm. This improved algorithm is combined Particle Swarm Optimization algorithm with Simulated Annealing´s idea, which has faster convergence speed and better solution quality. In this paper, The Mackey-Glass chaotic time series and the Henon series are used for testing and imitating. The results indicate that by using this model can get faster convergence speed, better stability, higher the precision of prediction, and stronger adjustability.