Chaotic Time Series Forecast Modeling Based on Biased Wavelet Neural Network

Chaos and artificial neural networks have been providing a new rout for investigating the complicated nonlinear time series. As the traditional neural networks are easy to get slow convergence and produce large redundancy which might consequently bring low efficiency, the biased wavelet neural networks is employed to build chaotic time series forecasting. Efforts are also made to assess the forecast modeling process, characteristics and key coefficients selection. Together with the phase space reconstruction theory, this paper discusses the application to month inflow runoff time series, which shows that the presented method is feasible and obtains satisfying forecast precisions.