基于非对称函数的离散混沌序列长期预测算法

摘要

离散混沌信号的迭代长期预测是一个困难问题。根据混沌信号概率密度函数的分布特性和分形特征，提出一种使用形状可自适应变化的基函数的预测算法。通过自适应算法改变基函数的形状参数，使函数波形发生任意改变，以模拟混沌信号密度函数的非对称特性，从而使基函数能够反映出混沌信号的更多特征。通过该算法对离散超混沌信号的长期预测的仿真结果表明，本算法能够有效提高预测精度和预测长度。%The iteration-based long prediction for discrete chaotic series is a challenging task .According to the density function and the fractal character of a chaotic signal , the authors designed a prediction model by applying shape-variable functions as basis func-tions, which can adjust the shapes automatically , to show the asymmetric property of a chaotic signal statistic density function .The strategy made basis functions provide more information of the chaotic map .In simulation, the algorithm was applied to make a long pre-diction for a hyper discrete chaotic map , and showed an improvement performance in prediction accuracy and length .