使用遗传算法的自适应Kalman滤波器  

Adaptive Kalman filter with genetic algorithm

作　　者：曲铁军[1] 杨旭东[1] 葛升民[1] 

机构地区：[1]哈尔滨工业大学控制科学与工程系,黑龙江哈尔滨150001

出　　处：《哈尔滨工业大学学报》2003年第6期655-659,共5页Journal of Harbin Institute of Technology

基　　金：国家自然科学基金资助项目 ( 60 10 40 0 3 )

摘　　要：基于Magill的Kalman滤波器池结构 ,设计了使用遗传算法的自适应Kalman滤波器 ,给出了离线和在线两种实现方案 .离线方案以辨识滤波参数为主要目的 ,进而可以对状态进行较准确的事后估计 ;在线方案以实时地对状态进行估计为目的 .对滤波参数寻优使用具有良好性能的浮点数编码遗传算法 ,该算法与二进制编码遗传算法相比收敛速度更快、搜索全局最优的能力更强 .Adaptive Kalman filter with genetic algorithm was designed on the configuration of Magill's Kalman filter bank with. Two realization schemes including off-line and on-line given. The off-line scheme aims primarily at identifying filtering parameters, and the postmortem estimation of states can be obtained more accurately while. The on-line scheme aims at real-time estimation of the states. A well-performed float-coded genetic algorithm was used to optimize the filtering parameters. In comparison with the binary-coded genetic algorithm, this algorithom has higher convergence speed and greater ability of finding global optimum. The efficiency of the proposed adaptive Kalman filter has been verified through simulation.

关 键 词：遗传算法 自适应Kalman滤波器 自动控制 滤波参数 编码 目标函数 滤波器池 估计误差 

分 类 号：TN713[电子电信—电路与系统] TP273.2[自动化与计算机技术—检测技术与自动化装置]