基于混沌粒子群优化小波支持向量机的汇率预测  

Forecasting Exchange Rate Based on Chaos Particle Swarm Optimization——Wavelet Support Vector Machine

作　　者：廖淑娇[1] 冯晓霞[1] 刘家彬[2] 

机构地区：[1]漳州师范学院数学与信息科学系,漳州363000 [2]四川民族学院计算机科学系,康定626001

出　　处：《科学技术与工程》2012年第11期2660-2664,共5页Science Technology and Engineering

基　　金：福建省自然科学基金项目(2010J01018);漳州师范学院科学研究项目(SJ1103)资助

摘　　要：目前,支持向量机(SVM)常用的参数寻优方法存在易陷入局部极值的缺点,而其常用的核函数的逼近精度也有待提高。基于混沌映射的遍历性与随机性和小波变换的局部分析与特征提取能力,提出了一种混沌粒子群优化小波支持向量机(CPSO-WSVM)的算法,并应用它构建汇率预测模型。实验结果表明,相比传统的粒子群优化高斯核SVM(PSO-GSVM)的算法,CPSO-WSVM算法大大提高了预测的精度和效率,应用效果好。Nowadays,the common parameter optimization methods of support vector machine（SVM） are easy to lapse into local extremum,and the approximation accuracy of its frequently-used kernel functions also needs to be improved.Based on the ergodicity and stochastic property of chaos mapping as well as the local analysis and feature extraction abilities of wavelet transform,an algorithm is presented which is named as chaos particle swarm optimization wavelet SVM（CPSO-WSVM）.This algorithm is used to construct exchange rate forecasting model.The experimental results show that CPSO-WSVM method has good application effect,which obtains much higher forecasting precision and efficiency than the traditional particle swarm optimization-Gaussian kernel SVM（PSO-GSVM）.

关 键 词：混沌粒子群优化 小波核函数 支持向量机 汇率 预测 

分 类 号：F832.6[经济管理—金融学] TP391[自动化与计算机技术—计算机应用技术]