基于当前统计模型的消隐记忆SCKF算法研究

Memory Attenuation Square-Root Cubature Kalman Filtering Algorithm Based on Current Statistics Model

为解决容积卡尔曼滤波器对机动目标跟踪过程中因模型不准确出现的加速度跟踪超调问题,对消隐记忆滤波和平方根滤波理论进行了研究,提出了一种消隐记忆因子改进的平方根容积卡尔曼滤波算法(Memory Attenuation Square-root Cubature Kalman Filter,MASCKF)。首先推导了基于线性状态方程的简化平方根容积卡尔曼滤波算法,提高了算法的实时性;其次在简化算法的时间更新环节引入消隐记忆因子,提高新量测数据在最优估计中的比重,使得加速度跟踪超调得到了很好的抑制。通过实验比对,验证了新算法对加速度跟踪超调的抑制效果,提高了对目标跟踪的实时有效性和准确性。

In order to solve the overshoot problem of acceleration tracking due to inaccurate model when tracking maneuvering targets with cubature Kalman filter,we made study on the square root filter and memory attenuation filter,and proposed a square-root cubature Kalman filter improved by memory attenuation factor,i.e.,Memory Attenuation Square-root Cubature Kalman Filter（MASCKF）.Firstly,a simplified square-root cubature Kalman filter based on linear state equation was deduced for improving the real-time performance of the algorithm.Then,memory attenuation factor was applied to the link of time updating of the simplified algorithm.Therefore,the weight of new measurement data was increased during the estimation,and the overshoot of acceleration tracking was well suppressed.Experimental results demonstrate the effect of the new algorithm on overshoot suppression,which can improve the accuracy and real-time performance of target tracking.