航空学报 > 2018, Vol. 39 Issue (8): 322071-322071   doi: 10.7527/S1000-6893.2018.22071

带异步相关噪声的战斗机蛇形机动跟踪算法

卢春光, 周中良, 刘宏强, 寇添, 杨远志   

  1. 空军工程大学 航空工程学院, 西安 710038
  • 收稿日期:2018-02-01 修回日期:2018-05-22 出版日期:2018-08-15 发布日期:2018-05-21
  • 通讯作者: 卢春光 E-mail:15891782912@163.com
  • 基金资助:
    国家自然科学基金(61472441)

Algorithm for fighter zigzag maneuver target tracking with correlated noises at one epoch apart

LU Chunguang, ZHOU Zhongliang, LIU Hongqiang, KOU Tian, YANG Yuanzhi   

  1. Aeronautics Engineering College, Air Force Engineering University, Xi'an 710038, China
  • Received:2018-02-01 Revised:2018-05-22 Online:2018-08-15 Published:2018-05-21
  • Supported by:
    National Natural Science Foundation of China (61472441)

摘要: 针对异步相关噪声背景下战斗机蛇形机动模式转弯角速度辨识问题,考虑到目标状态与转弯角速度之间相互耦合的特性,从联合优化的解决思路出发,基于期望最大化(EM)算法框架,提出了一种带异步相关噪声的联合估计与辨识算法。首先采用"去相关框架"解除过程噪声与量测噪声之间的相关性,从而将异步相关噪声背景下的转弯角速度辨识问题转换成具有一步状态延迟的转弯角速度辨识问题,其次通过解除目标状态与转弯角速度之间的非线性耦合关系,基于期望最大化算法实现了战斗机蛇形机动目标状态与转弯角速度的联合估计与辨识,从而获得转弯角速度闭环形式的解析解:在E-step,通过利用异步相关噪声背景下的高阶容积卡尔曼平滑器(HCKS),获得目标状态的后验估计;在M-step,通过极大化条件似然函数,获得转弯角速度的解析解。最后通过仿真验证了所提算法的目标状态估计与角速度辨识的精度均优越于传统的扩维法。

关键词: 蛇形机动, 异步相关噪声, 去相关框架, 期望最大化, 联合估计与辨识

Abstract: Aiming at the turn rate identification of fighter zigzag maneuver in the context of noise at the epoch apart, taking into account the characteristics of the coupling between the target state and turn rate, starting from the joint optimization solution an algorithm for joint estimation and identification with correlated noises at one epoch apart is proposed based on the Expectation Maximization (EM) algorithm. First, the de-correlating framework is utilized to eliminate the correlation between process noise and measurement noise, and thus the problem of turn rate identification with correlated noises at one epoch apart is transformed into the problem of turn rate identification with one-step delayed state. Second, by eliminating the non-linear coupling between the target state and turn rate, joint estimation and identification of both states and turn rate are achieved using the EM algorithm. A closed-loop analytic solution for the turn rate is then achieved:in the E-step, the state of the target and the expectation are achieved accurately by use of the High-degree Cubature Kalman Smoothers (HCKS) with correlated noises at one epoch apart; in the M-step, the analytical identification result of turn rate is obtained by maximizing the expectation. Simulation results show that the algorithm proposed is superior to the traditional augmentation method in terms of state estimation and turn rate identification.

Key words: zigzag maneuver, correlated noises at one epoch apart, de-correlating framework, Expectation Maximization (EM), joint estimation and identification

中图分类号: