基于先验知识的抗野值kalman滤波算法  

Anti-outlier kalman filter algorithm based on prior knowledge

作　　者：王徐民 赵贵 WANG Xumin;ZHAO Gui(Nanjing Corad Electronic Equipment Co.,Ltd.,Nanjing 211100,China)

机构地区：[1]南京科瑞达电子装备有限责任公司(原924厂),南京211100

出　　处：《计算机应用文摘》2023年第21期104-106,110,共4页Chinese Journal of Computer Application

摘　　要：在使用标准Kalman滤波算法对观测数据进行滤波时,如果观测数据包含野值(即大误差数据点),这些野值及其后续点的滤波结果可能会明显偏离真实值,导致较大的滤波误差。文章充分利用了先验知识,包括在无源测向系统正常工作时观测方位基本稳定、当观测方位出现大幅度变化时观测数据不可靠,以及信号幅度与观测方位精度之间存在相关关系等,在标准Kalman滤波算法中,引入了不同的惩罚因子函数,并分别对两组不同的观测数据进行了仿真实验。仿真结果表明,引入惩罚因子函数后,观测方位的均方根误差值显著减小,从而显著提高了无源测向系统的测向输出精度。When the standard Kalman filtering algorithm filters the observed data,if there are outliers(i.e.,large error points)in the data,the filtering results of outliers and subsequent points will seriously deviate from the true value,which will lead to large filtering errors.This paper,by using passive direction finding system to work normally when the bearings basically stable,the bearings appeared significantly beat is unreliable,and the precision of the signal amplitude and the bearings there is relationship between prior knowledge,in the standard kalman filtering algorithm introduces different punishment factor function,respectively on two groups of different observation data of simulation,simulation and comparison results show that the after introducing the penalty factor function,the root mean square error of the output direction is significantly reduced,and the output accuracy of the passive direction finding system is significantly improved.

关 键 词：KALMAN滤波 先验知识 野值 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]