高速列车非线性模型的极大似然辨识

衷路生; 李兵; 龚锦红; 张永贤; 祝振敏

doi:10.3724/SP.J.1004.2014.02950

高速列车非线性模型的极大似然辨识

doi: 10.3724/SP.J.1004.2014.02950

1.
华东交通大学电气与电子工程学院南昌 330013

基金项目:

国家自然科学基金(61263010,60904049),江西省青年科学基金(20114BAB211014),江西省教育厅研究项目(GJJ14399),国家留学基金(2011836118)资助

详细信息

作者简介:
李兵华东交通大学电气与电子工程学院硕士研究生. 主要研究方向为系统辨识, 统计学习理论. E-mail: libing87@gmail.com

通讯作者:
衷路生华东交通大学电气与电子工程学院副教授. 主要研究方向为系统辨识, 统计学习理论. 本文通信作者.E-mail: lszhongzju@gmail.com

计量
- 文章访问数: 1621
- HTML全文浏览量: 107
- PDF下载量: 1070
- 被引次数: 0
出版历程
- 收稿日期: 2013-10-10
- 修回日期: 2014-03-19
- 刊出日期: 2014-12-20

Maximum Likelihood Identification of Nonlinear Model for High-speed Train

1.
School of Electrical and Electronic Engineering, East China Jiaotong University, Nanchang 330013

Funds:

Supported by National Natural Science Foundation of China (61263010, 60904049), Natural Science Foundation for the Youth of Jiangxi Province (20114BAB211014), Research Project of Jiangxi Educational Department (GJJ14399), and Grant of China Scholarship Council (2011836118)

摘要

摘要: 提出高速列车非线性模型的极大似然(Maximum likelihood, ML)辨识方法,适合于高速列车在非高斯噪声干扰下的非线性模型的参数估计.首先,构建了描述高速列车单质点力学行为的随机离散非线性状态空间模型,并将高速列车参数的极大似然(ML)估计问题转化为期望极大(Expectation maximization, EM)的优化问题; 然后,给出高速列车状态估计的粒子滤波器和粒子平滑器的设计方法,据此构造列车的条件数学期望,并给出最大化该数学期望的梯度搜索方法,进而得到列车参数的辨识算法,分析了算法的收敛速度; 最后,进行了高速列车阻力系数估计的数值对比实验. 结果表明, 所提出的辨识方法的有效性.
- 高速列车 /
- 系统辨识 /
- 极大似然 /
- 平滑滤波器 /
- 梯度搜索
Abstract: A maximum likelihood (ML) system identification method is proposed for parameter estimation of nonlinear dynamic high-speed train model subject to non-gaussian noise. Firstly, a stochastic nonlinear discrete state-space model is established to describe the dynamic behavior of high-speed train as a single-point-mass object. The expectation-maximization (EM) approach is employed to compute the ML parameter estimates. In addition, the techniques of particle filtering and particle smoothing are given to estimate the nonlinear state of high-speed train, which is used to compute approximation of the conditional expectation. Furthermore, gradient-based search method is presented to maximize the conditional expectation. And the identification algorithm is given for parameter estimation of high-speed train. The convergence rate of the identification algorithm is also discussed in detail. Finally, numerical simulation study of parameter estimation for high-speed train is implemented and the results show the effectiveness of the proposed ML identification method.
- High-speed train /
- system identification /
- maximum likelihood (ML) /
- smoothing filters /
- gradient-based search

HTML全文

参考文献(14)

[1]	Zhai Wan-Ming, Jin Xue-Song, Zhao Yong-Xiang. Some typical mechanics problems in high-speed railway engineering. Advances in Mechanics, 2010, 40(4): 358-374(翟婉明, 金学松, 赵永翔. 高速铁路工程中若干典型力学问题. 力学进展, 2010, 40(4): 358-374)
[2]	[2] Song Q, Song Y D, Tang T, Ning B. Computationally inexpensive tracking control of high-speed trains with traction/braking saturation. IEEE Transactions on Intelligent Transportation Systems, 2011, 12(4): 1116-1125
[3]	[3] Song Q, Song Y D. Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures. IEEE Transactions on Neural Networks, 2011, 22(12): 2250-2261
[4]	Zhong Lu-Sheng, Yan Zheng, Yang Hui, Qi Ye-Peng, Zhang Kun-Peng, Fan Xiao-Ping. Predictive control of high-speed train based on data driveb subspace approach. Journal of the China Railway Society, 2013, 35(4): 77-83(衷路生, 颜争, 杨辉, 齐叶鹏, 张坤鹏, 樊晓平. 数据驱动的高速列车子空间预测控制方法. 铁道学报, 2013, 35(4): 77-83)
[5]	Zhong Lu-Sheng, Yan Zheng, Gong Jin-Hong, Zhang Yong-Xian, Zhu Zhen-Min, Fan Xiao-Ping. Adaptive subspace predictive control of high-speed train based on time-varying forgetting factor. Journal of the China Railway Society, 2013, 35(5): 54-61(衷路生, 颜争, 龚锦红, 张永贤, 祝振敏, 樊晓平. 时变遗忘因子的高速列车自适应子空间预测控制. 铁道学报, 2013, 35(5): 54-61)
[6]	Zhong Lu-Sheng, Yan Zheng, Gong Jin-Hong, Zhang Yong-Xian, Zhu Zhen-Min. Adaptive subspace predictive control method of high-speed train. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2013, 41(8): 28-35(衷路生, 颜争, 龚锦红, 张永贤, 祝振敏. 高速列车的自适应子空间预测控制方法. 华中科技大学学报, 2013, 41(8): 28-35)
[7]	[7] Gibson S, Ninness B. Robust maximum-likelihood estimation of multivariable dynamic systems. Automatica, 2005, 41(10): 1667-1682
[8]	Zhong Lu-Sheng, Song Zhi-Huan. Hierarchical optimization identification of LTI state-space by projected gradient search Acta Automatica Sinica, 2008, 34(6): 711-715 (衷路生, 宋执环. 基于正交梯度搜索的动态系统递阶优化辨识. 自动化学报, 2008, 34(6): 711-715)
[9]	Zhong Lu-Sheng, Fan Xiao-Ping, Yang Hui, Qu Zhi-Hua. Output error identification of LTI state-space models by orthogonal gradient search. Control and Decision, 2011, 26(5): 685-689(衷路生, 樊晓平, 杨辉, 瞿志华. 状态空间模型基于正交梯度搜索的预报误差辨识, 控制与决策, 2011, 26(5): 685-689)
[10]	De Brabanter K. Least Squares Support Vector Regression with Applications to Large-Scale Data: a Statistical Approach [Ph.D. dissertation], Katholieke Universiteit Leuven, Faculty of Engineering, Belgium, 2011
[11]	Schon T B, Wills A, Ninness B. System identification of nonlinear state-space models. Automatica, 2011, 47: 39-49
[12]	Cheng Yu-Hu, Feng Huan-Ting, Wang Xue-Song. Expecta- tion-maximization policy search with parameter-based exploration. Acta Automatica Sinica, 2012, 38(1): 38-45(程玉虎, 冯涣婷, 王雪松. 基于参数探索的期望最大化策略搜索. 自动化学报, 2012, 38(1): 38-45)
[13]	Dempster A P, Laird N M, Rubin D B. Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, 1977, 39(1): 1-38
[14]	Zhang Gong-Yuan, Cheng Yong-Mei, Yang Feng, Pan Quan, Liang Yan. Design of an adaptive particle filter based on variance reduction technique. Acta Automatica Sinica, 2010, 36(7): 1020-1024