高速动车组多质点模型的极大似然辨识

Maximum Likelihood Identification of Multiple Point Model for High-speed Electrical Multiple Units

针对高速动车组多质点模型的参数估计问题,为列车稳定运行和提高效能,提出了适合于高速动车组多质点模型的极大似然辨识方法。首先,建立高速动车组的随机离散非线性状态空间模型,并将高速动车组参数的极大似然估计问题转化为期望极大的优化问题。给出适合于高速动车组的改进粒子滤波算法,构造了高速动车组的条件数学期望。然后,给出高速动车组参数优化的梯度搜索方法,进而得到高速动车组参数的辨识算法。以CRH3型高速动车组为对象进行仿真,结果表明提出的方法有效,为列车高速稳定运行提供了科学依据。

A maximum likelihood solution to the problem of identifying parameters for a multiple-point mass model of high-speed electrical multiple units（EMU） was presented in the paper. A stochastic discrete nonlinear statespace model for the multiple-point mass model was proposed to describe the dynamic behavior of multiple-point mass model of high-speed EMU. And the expectation maximization algorithm was employed solve the problem of ML pa- rameter estimates. In addition, an improved particle filter approach was given to estimate the state of high-speed EMU, which was used to compute approximation of conditional expectation. And the conditional expectation was opti- mized by gradient-based search method. Furthermore, the identification algorithm was given for parameter estimation of multiple-point mass model of high-speed EMU. Finally, numerical simulation study on parameter estimation for multiple-point mass model of high-speed EMU was implemented and the results show the effectiveness of the proposed ML identification method.