基于遗传规划算法的Lorenz序列多步预测方法研究  

Multi-step prediction methods for Lorenz series based on GP algorithm

作　　者：吴晓军[1,2] 杨磊[1] 张玉梅[1,2] 马悦[2] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]陕西师范大学计算机科学学院,陕西西安710062

出　　处：《西安建筑科技大学学报（自然科学版）》2013年第3期447-451,456,共6页Journal of Xi'an University of Architecture & Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(11172342);教育部"新世纪优秀人才支持计划"项目(NCET-11-0674);陕西省自然科学基础研究计划项目(2012JQ8051)

摘　　要：Lorenz序列是由Lorenz系统产生的一组离散的时间序列,在Lorenz序列混沌特性的基础上提出了一种基于改进的GP算法的Lorenz序列预测方法.针对Lorenz序列预测模型求解,在GP算法中引入多种群及爬山算法,建立了Lorenz序列预测模型,在此基础上通过粒子群优化算法对预测模型参数进行优化,通过优化后的预测模型对Lorenz序列进行预测.最后通过实验对预测模型进行评价,结果表明,利用本文提出的方法获得的预测模型能有效的对Lorenz序列进行预测.Lorenz sequence is a set of discrete time series generated by the Lorenz system.Proposed is in this paper,a Lorenz system prediction method based on the improved GP algorithm on the basis of Lorenz’s chaotic characteristics.For solving Lorenz series prediction problems,multi-population and hill-climbing algorithms were introduced into GP algorithm and then the Lorenz series model was built by using this method.The particle swarm optimization is used to optimize the parameters of the model and then optimized model is used to predict the Lorenz series.Results of the model evaluation experiments showed that the prediction model obtained by the proposed method in this paper can effectively be used to predict the Lorenz series.

关 键 词：遗传规划算法 LORENZ系统 多步预测 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]