论文部分内容阅读
针对高速列车参数估计中参数增广为状态变量时所出现的非线性问题,提出一种基于边缘粒子滤波的参数估计方法。在Rao-Blackwellised(RB)框架下,将高速列车性能参数估计的概率模型进行分块化处理。应用卡尔曼滤波对线性的状态块进行一步预测和测量更新,应用粒子滤波对非线性的参数块进行一步预测与测量更新,实现参数的动态估计,并通过理论分析和高速列车参数估计实例验证了方法的有效性。分析结果表明:与经典的扩展卡尔曼滤波相比,提出的方法具有对初值免疫和算法稳定的特点;参数估计误差快速收敛到5%以内,且提出的参数估计方法是无偏估计,具有较好的工程适用性。
In order to solve the nonlinear problem in the parameter estimation of high-speed train extended to state variables, a parameter estimation method based on edge particle filter is proposed. Under Rao-Blackwellised (RB) framework, the probability model of high speed train performance parameter estimation is divided into blocks. The Kalman filter is used to predict and measure the linear state block in one step. Particle filter is used to predict and update the nonlinear parameter block in one step. The dynamic estimation of parameters is achieved. Theoretical analysis and high speed train parameter estimation The effectiveness of the method. The results show that compared with the classical extended Kalman filter, the proposed method has the characteristics of initial immune and stable algorithm. The error of parameter estimation quickly converges to less than 5%, and the proposed parameter estimation method is unbiased estimation. Better engineering applicability.