基于n阶平均发散度的混沌时序可预测尺度研究  

Study on the Predictable Size of Chaotic Time Series Based on n-rank Average Divergence Degree

作　　者：岳毅宏[1] 王金凤[2] 程国平[1] 

机构地区：[1]天津大学管理学院,天津300072 [2]郑州大学管理科学与工程系,郑州450002

出　　处：《系统仿真学报》2004年第11期2564-2566,共3页Journal of System Simulation

基　　金：国家自然科学基金资助项目(No.79970043)

摘　　要：深入分析了基于最大Lyapunov指数的混沌时序可预测尺度确定方法,指出了其存在的不足。在此基础上,提出了一种新的基于n阶平均发散度的可预测尺度确定方法。首先给出了n阶平均发散度的定义;然后阐述了新方法的基本原理,并总结了可预测尺度的计算步骤;最后将所提方法应用于电力负荷时序可预测尺度的确定中,通过对预测误差的分析和比较,验证了新方法的有效性。The method of determining predictable size of chaotic time series based on maximal Lyapunov exponent is deeply analyzed, and its defect is also pointed out. On this basis, a novel method used for the determination of predictable size based on n-rank average divergence degree is proposed. Firstly, the definition of n-rank average divergence degree is given. Then, the principle of the new method is demonstrated, and its computing procedure is also summarized. Finally, the suggested method is applied to the predictable size determination of power load time series, and by means of analyzing and comparing the forecast error, the validity of this method is verified.

关 键 词：n阶平均发散度 最大LYAPUNOV指数 混沌时序 可预测尺度 电力负荷 

分 类 号：TP273.4[自动化与计算机技术—检测技术与自动化装置]