混沌行为预测：一种变型的码本预测方法

利用混沌周期点之全体在其不变集中稠及其不变集的几何结构呈一定程度的自相似性性质,提出了一种变形的码本预测方法,对带和不带观测噪声的混沌时间序列进行预测.在保证精度的前提下,克服了全局非线性化模型和局部线性化模型对噪声敏感及预测时间短的缺点.

Chaotic Behavior Prediction:An Improved Codebook Prediction Algorithm

The existence of a countable periodic orbits and the structure of the strange attrac-tors presenting some self-similar properties are the main features of chaotic behavior. From thesignal processing point of view, the exploration of the property is made and provided with animproved codebook prediction algorithm to predict the chaotic time series with or without additive noises. Tbe advantage of the method is that it prolongs the prediction time and overcomesthe shortcomings of globally nonlinear autoregresive model and locally linear model that is unfitfor additive noises.