混沌统计量的预测研究

Research on the Chaotic Statistic Data Prediction

提出了一种基于混沌吸引子的统计量序列预测技术 .对于一个经诊断为混沌的统计量序列 ,应用神经网络建立模型 ,短期预测混沌序列 .实验证明 :Lyapunov指数越大 ,预测效果越差 ;神经网络的预测结果优于 AR模型的预测结果 .混沌序列是完全可以预测的 .

A forecasting technique for chaotic statistic data is presented. Short term predictions of chaotic series are realised with neural network model, after diagnosing the time series as chaotic statistic series. Experiment results are shown that the accuracy is decreaced with Lyapunov exponent increasing, predict results of neural network are superior to those of AR model.Chaotic time series can be predicted totally.