一种强跟踪有限差分滤波二元估计算法  

Dual Estimation Algorithm Based on Strong Tracking Finite-difference Filtering

作　　者：李江 杨慧中 丁锋 

机构地区：[1]江南大学系统工程研究所,江苏无锡214122

出　　处：《控制工程》2006年第4期381-383,387,共4页Control Engineering of China

基　　金：国家自然科学基金资助项目(No.60574051)

摘　　要：针对非线性工业过程测量的滞后性和模型不确定性给系统状态估计和模型参数估计造成的困难,在扩展Kalman滤波器(EKF)的基础上,引入有限差分滤波器(FDEKF)和次优渐消因子,提出了一种强跟踪有限差分滤波状态和参数二元估计算法。该二元估计算法将滤波器分解为参数滤波和状态滤波两个过程,分别估计模型参数和系统状态。最后,将该算法应用于一化学反应过程的仿真,结果表明,这种强跟踪有限差分滤波的二元估计算法在原模型或参数存在偏差的情况下,仍能较准确地估计系统状态和模型参数,并具有较强的数值鲁棒性。To the difficulty in estimating the states and parameters of nonlinear industrial processes with measurement delays and model uncertainty, based on extended Kalman filter （EKF）, a dual estimation algorithm is presented by introducing finite-difference filtering （FDEKF） with suboptimal fading factors. The dual strong tracking finite-difference filter is divided into the parameter filter and the state filter which are used to estimate model parameters and system states respectively. This algorithm is applied to a chemical reaction process as simulation cases. The results show that this algorithm can estimate the states and parameters of aforementioned processes accurately and run well even in the case of model or parameter deviation occurring, and good numerical robustness of the algorithm is shown.

关 键 词：二元估计 KALMAN滤波 有限差分滤波 次优渐消因子 强跟踪滤波 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]