摘要
针对EKF对初值选取的敏感性和线性化误差较大等问题,根据目标初始距离可能分布,进行距离区间的分割,建立每区间所对应的滤波模型;利用协方差旋转变化原理,确定旋转变换的近似矩阵;分析这种近似处理的误差,提出基于协方差旋转变换的多模型滤波算法,采用蒙特卡罗方法进行了仿真计算。
Extended Kalman Filter（EKF） is sensitive to initial value selection and has high linear error. According to the initial range to where the underwater target may locate, we divided the initial range into sections （sub-ranges）. Filtering models were built up for each section correspondingly. The covariance rotated transform approximate matrix was constructed by use of the principle of covariance rotated transform. After analyzing the errors of the approximating process, a single platform bearings-only target tracking algorithm based on covariance rotated transform was put forward. Monte Carlo method was used for simulation. The result showed that the algorithm is effective, stable and has high accuracy.
出处
《电光与控制》
北大核心
2008年第5期20-23,共4页
Electronics Optics & Control
基金
海军工程大学自然基金资助(HGDJJ06015)
关键词
目标跟踪
纯方位
多模型
滤波
target tracking
bearings-only
multiple model
filter
