基于多新息理论的EKF改进算法.pdf

优先出版 计 算 机 应 用 研 究 第32 卷 基于多新息理论的EKF 改进算法* 刘毛毛，秦品乐，吕国宏，常 江 ( 中北大学 计算机与控制工程学院，太原 030051) 摘 要：针对标准的扩展卡尔曼滤波算法 （EKF）在强非线性系统中估计精度较低的问题，提出了一种改进的扩展卡尔 曼滤波算法 （MI-EKF），使得滤波精度得到很大的提高。MI-EKF 是在标准EKF 基础上，结合多新息理论，不仅考虑了 系统当前的测量值，而且也充分考虑了之前时刻的有用信息，从而使得MI-EKF 的滤波精度和稳定性得到改善。最后， 讨论了新息数量对改进算法精度的影响，仿真结果表明包含两个新息的MI-EKF 算法滤波效果最佳。 关键词：非线性；扩展卡尔曼滤波；多新息；多新息扩展卡尔曼滤波；仿真分析 中图分类号：TP391.9 文献标志码：A Improvement of extended Kalman filter based on multi-innovation theory LIU Mao-mao, QIN Pin-le, LV Guo-hong, CHANG Jiang (College of Computer Control Engineering, North University of China, Taiyuan, 030051, China) Abstract: Because of the low estimation accuracy of normal extended Kalman Filter (EKF) in strong nonlinear system, this paper developed an improved extended Kalman Filter (MI-EKF) to solve the problem, and it improves the filtering accuracy greatly. MI-EKF is proposed by combining multi-innovation theory and the standard EKF. MI-EKF has better precision and stability, because MI-KF considers not only the current measured value, but also give full consideration to the time before state of motion. Finally, discuss the impact of algorithm precision which include different numbers of innovations, and simulation results show that the improved algorithm MI-EKF includes two innovations is optimal. Key Words: nonlinear; extend Kalman filter; multi-innovation; multi-innovation extend Kalman filter; simulation analyses 估计能力。文献[7]提出一种基于插值公式的EKF(扩展卡尔曼滤 0 引言 波)改进算法，利用基Stifling 插值公式的差分方法代替EKF 中 在运动目标跟踪领域中,卡尔曼滤波器在线性