Abstract: Aiming at the prediction precision problem using prediction model for chaotic time series，the 4 prediction models based on chaos theory，such as Radial Basis Function neural network（RBF） model，Lyapunov exponent model，local-region prediction model and Volterra filter model are introduced.Besides，the 4 prediction models are comparative studied.The time series of several typical nonlinear systems are predicted by the 4 prediction models.The simulation results show that the proposed 4 prediction models have effective prediction results for typical nonlinear systems.The RBF model and the Volterra filter model have the better performances on the precision accuracy.

Key words: time series, prediction model, chaos theory, phase space reconstruction

摘要： 针对目前混沌时间序列预测模型预测结果差异较大的问题，归纳了4种混沌时间序列预测模型：BRF神经网络模型、最大Lyapunov指数模型、局域线性模型和Volterra滤波器自适应预测模型，并对这4种预测模型进行了比较研究。应用4种预测模型对几个典型的非线性系统进行预测仿真。结果表明，这4种预测模型对典型混沌时间序列预测都具有很好的预测效果；在预测精度上BRF模型和Volterra模型明显优于最大Lyapunov指数模型和局域线性模型。

关键词: 时间序列, 预测模型, 混沌理论, 相空间重构