论文部分内容阅读
针对高速列车纵向动力学特性,分析了牵引力、制动力、阻力与速度和加速度的关系;考虑了天气和线路对高速列车运行状态造成的随机干扰,以及机械磨损和运行环境对列车模型结构参数造成的随机影响,建立了噪声干扰下的高速列车纵向动力学参数化状态空间模型,利用期望极大化准则,计算了列车模型参数的条件数学期望,并结合粒子滤波理论估计了参数粒子下的列车状态;基于贝叶斯后验概率理论,建立了高速列车非线性动力学模型的时变参数辨识方法,估计了列车的实时状态,并在噪声与参数分布均属于高斯分布、噪声属于高斯分布与参数属于指数分布、噪声属于伽玛分布与参数属于高斯分布的3种工况下,进行了蒙特卡洛仿真试验。仿真结果表明:在3种工况下,高速列车位移和速度的估计值与真实值的相对误差小于5%,列车模型参数估计值与真实值的相对误差小于10%,满足实际系统需求,因此,在高斯或伽玛噪声的干扰下,针对给定概率分布的时变参数,本方法均能实现系统状态的估计和模型参数的辨识。
According to the longitudinal dynamic characteristics of high-speed trains, the relationship between tractive force, braking force, drag and speed and acceleration is analyzed. The random disturbance caused by the weather and the line on the running status of high-speed trains and the mechanical wear and the operating environment are considered , The parametric state space model for longitudinal dynamics of high-speed trains under noise disturbance was established. The conditional mathematical expectation of train model parameters was calculated by using the expectation maximization criterion. The particle filter theory was used to estimate the train parameters State. Based on the Bayesian posterior probability theory, a time-varying parameter identification method is established for the nonlinear dynamics model of high-speed trains to estimate the real-time state of the train. The noise and parameter distribution belong to the Gaussian distribution. The noise belongs to Gaussian distribution and The parameters belong to the exponential distribution, the noise belongs to the gamma distribution and the parameters belong to the Gaussian distribution under the three conditions, the Monte Carlo simulation test. The simulation results show that the relative error between the estimated value and the true value of the displacement and speed of the high-speed train is less than 5% and the relative error between the estimated value and the true value of the train model is less than 10% under the three conditions, thus satisfying the actual system requirements Under the interference of Gaussian or gamma noise, this method can estimate the state of the system and identify the model parameters for a given time-varying parameter of the probability distribution.