《计算机应用研究》|Application Research of Computers

基于自适应变异PSO的ARMA模型参数寻优及预测应用

Parameters optimization and prediction application of ARMA model based on PSO with adaptive mutation

免费全文下载 (已被下载 次)  
获取PDF全文
作者 李怀俊,谢小鹏,李军
机构 1.广东交通职业技术学院 车辆安全工程技术中心,广州 510650;2.华南理工大学 汽车摩擦学与故障诊断研究所,广州 510640
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2015)04-1004-03
DOI 10.3969/j.issn.1001-3695.2015.04.010
摘要 为提高ARMA模型在时间序列预测中的精度,提出一种基于改进粒子群算法(AMPSO)的模型参数智能寻优估计方法。AMPSO算法以粒子熵的判别为依据,在寻优过程中对算法的关键参数进行多次自适应变异,以提高其跳出局部、面向全局寻优的能力。模型参数寻优过程基于ARMA(2n,2n-1)模型架构依次跳阶,每阶的参数初估后运用AMPSO算法展开寻优,适应度判定标准为当前模型残差方差最小。针对齿轮箱轴承的输出扭矩进行预测,结果表明,AMPSO算法在收敛性和寻优速度方面效果良好;参数寻优方法可显著提高参数预测精度,具有良好的工程应用价值。
关键词 粒子熵;PSO;ARMA模型;参数寻优;残差方差
基金项目 广东省自然科学基金资助项目(S2011010002118)
广东省高校优秀青年教师培养项目(Yq2013178)
本文URL http://www.arocmag.com/article/01-2015-04-010.html
英文标题 Parameters optimization and prediction application of ARMA model based on PSO with adaptive mutation
作者英文名 LI Huai-jun, XIE Xiao-peng, LI Jun
机构英文名 1. Vehicle Safety Engineering Technology Center, Guangdong Communication Polytechnic, Guangzhou 510650, China; 2. Automobile Tribology & Fault Diagnosis Institute, South China University of Technology, Guangzhou 510640, China
英文摘要 To improve the prediction precision with auto-regressive moving average(ARMA) model in time series, this paper presented an ARMA parameters estimation method based on improved particle swarm optimization(AMPSO). The algorithm determined the adaptive mutation scheme with the key parameters of AMPSO based on the discrimination of swarm entropy to improve the ability of global optimization in searching process. It optimized the model parameters using jumping steps scheme from smallest step based on ARMA(2n, 2n-1) model, and used the AMPSO algorithm that the fitness criterion was to get the minimum residual variance after the initial parameters estimation in every step. Using the method to estimate the torque of the gear box bearing and the result shows there is good engineering application value with notable performance in astringency and optimization speed to AMPSO algorithm and the significant precision to the prediction method.
英文关键词 swarm entropy; PSO; ARMA model; parameter optimization; residual variance
参考文献 查看稿件参考文献
  [1] 时维国, 邵诚, 孙正阳. 基于AR模型时延预测的改进GPC网络控制算法[J] . 控制与决策, 2012, 27(3):477-480.
[2] 王黎明. 应用时间序列分析[M] . 上海:复旦大学出版社, 2009.
[3] 桂中华, 韩凤琴. 小波包—最大嫡谱估计及其在水轮机故障诊断中的应用[J] . 电力系统自动化, 2004, 28(2):62-66.
[4] SAPANKEVY N, SANKAR R. Time series prediction using support vector machines:a survey[J] . IEEE Computational Intelligence Magazine, 2009, 4(2):24-38.
[5] HUANG J C. A genetic algorithm approach to the spectral estimation of time series with noise and missed observations[J] . Information Sciences, 2008, 178(24):4632-4643.
[6] ALUAREZ R, LUIS J. Support vector method for robust ARMA system identification[J] . IEEE Trans on Signal Processing, 2004, 52(1):155-164.
[7] 单黎黎, 张宏军, 王杰, 等. 一种改进粒子群算法的混合核ε-SVM参数优化及应用[J] . 计算机应用研究, 2013, 30(6):1636-1639.
[8] LEI Kai-you, QIU Yu-hui, WANG Xue-fei, et al. High dimension complex functions optimization using adaptive particle swarm optimizer[C] // Proc of the 1st International Conference on Rough Sets and Knowledge Technology. 2006:321-326.
[9] SHI Yu-hui, EBERHART R. A modified particle swarm optimizer[C] //Proc of IEEE International Conference on Evolutionary Computation. Piscataway:IEEE Press, 1998:69-73.
[10] 郭敬, 董彦良, 赵克定, 等. 基于混合优化策略的自回归—滑动平均模型建模[J] . 机械工程学报, 2007, 43(4):229-233.
[11] 张树京, 齐立心. 时间序列分析简明教程[M] . 北京:清华大学出版社, 2003.
[12] 李学军, 蒋玲莉, 杨大炼. 基于双谱分布区域的齿轮聚类分析与故障诊断[J] . 振动工程学报, 2011, 24(3):304-309.
[13] XU Yun-jie, XIU Shu-dong. A new and effective method of bearing fault diagnosis using wavelet packet transform combined with support vector machine[J] . Journal of Computers, 2011, 6(11):2502-2509.
收稿日期 2013/12/26
修回日期 2014/4/24
页码 1004-1006,1015
中图分类号 TP181;O329
文献标志码 A