基于粒子滤波的NAR模型状态过程估计

Estimation of NAR model state based on particle filter

针对状态转移方程是非线性自回归(NAR)模型的一类动态系统最优估计问题,提出利用粒子滤波(PF)方法估计NAR模型状态。该方法用正交最小二乘法建立NAR模型,得到系统状态方程和量测方程,利用PF方法估计NAR模型状态,减少因参数估计带来的状态估计误差。仿真实验表明,基于PF方法估计NAR模型状态是可行的,且比传统的NAR模型估计精度更高。

For a class of dynamic system optimal estimation problem that the state transition equation is nonlinear auto-regressive（NAR）model,aparticle filter method is proposed to estimate NAR model state.Firstly NAR model is established by using the orthogonal least squares,and the state equation and measurement equation of the system are established according to NAR model.And then NAR model state is estimated by the particle filter method to reduce the state estimation error.Simulation experiments show that the estimation of NAR model state based on particle filter is feasible,and the estimation precision is higher than the traditional NAR model.