| 基于混合滤波最大期望算法的高速列车建模 |
| |
| 引用本文： | 王呈,陈晶,荀径,李开成.基于混合滤波最大期望算法的高速列车建模[J].自动化学报,2019,45(12):2260-2267. |
| |
| 作者姓名： | 王呈 陈晶 荀径 李开成 |
| |
| 作者单位： | 1.江南大学物联网工程学院 无锡 214012 |
| |
| 基金项目： | 国家自然科学基金(61603156, 61973137), 高速铁路基础研究联合基金(U1734210), 北京交通大学教育基金会基金(9907006519)资助 |
| |
| 摘 要： | 针对高速列车非线性单质点模型的特殊结构及含有隐含变量问题, 提出一种基于混合滤波的最大期望辨识方法. 借助递阶辨识理论, 将高铁列车状态空间模型分解为线性子系统模型和非线性子系统模型. 进而, 分别利用卡尔曼滤波和粒子滤波对速度和位移状态进行联合估计. 最后, 使用最大期望方法辨识高铁列车子系统模型参数, 解决了隐含变量辨识问题. 和传统方法相比, 本文所提出方法计算量小, 且具有较高的辨识精度. 仿真对比实验结果验证了该方法的有效性.
|
| 关 键 词： | 参数估计 卡尔曼滤波 粒子滤波 递阶辨识 最大期望算法 |
| 收稿时间： | 2019-03-19 |
Hybrid Filter Based Expectation Maximization Algorithm for High-speed Train Modeling |
| |
| Affiliation: | 1.School of Internet of Things, Jiangnan University, Wuxi 2140122.School of Science, Jiangnan University, Wuxi 2140123.State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing 1000444.National Engineering Research Center of Rail Transportation Operation and Control System, Beijing Jiaotong University, Beijing 100044 |
| |
| Abstract: | For the special high-speed train model structure with hidden variables in the form of the single mass-point, a hybrid filter based expectation maximization (EM) algorithm is proposed. By employing the hierarchical identification theory, the high-speed train state-space model is decomposed into a linear subsystem and a nonlinear subsystem. Furthermore, the Kalman filter and the particle filter are provided to estimate the velocity and displacement, respectively. Finally, the parameters of subsystems are identified by using the EM algorithm. Compared to the classical methods, the proposed algorithm can produce high accuracy estimation with less computational effort. The simulation results verify the effectiveness of the algorithm. |
| |
| Keywords: | |
|
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
|
