基于改进自适应IMM算法的机动目标跟踪

Maneuvering Target Tracking Method Based on Improved Adaptive Interacting Multiple Models

文中提出了一种改进的跟踪机动目标的自适应IMM(AIMM)算法,采用协同转弯运动模型作为目标的机动模型,通过估计目标的角速度,并在估计参数值的周围选择数量一定的模型组成模型集来覆盖目标所有可能的运动模式,采用交互模型算法进行状态估计;应用变结构思想,将模型之间的切换理解为随机有向图,并综合利用前一时刻模型的后验概率和当前时刻模型的预测概率,计算模型之间的转移概率,从而能够根据目标(转弯)机动的情况,自适应地建立IMM算法的可变模型集。仿真结果表明,改进的AIMM算法的跟踪精度有了较大的改善,算法也更加平稳。

A adaptive interacting muhiple models（AIMM） algorithm is proposed to track maneuvering target, turn model is applied as maneuvering model of target, angle speed is estimated and the model set which is composed of several models around the estimation value are chosen to cover all possible maneuverability, then IMM algorithm is utilized to estimation target state. Varied structure idea is adopted that switch between models is regarded as digraph switching, and posterior probability of former moment and forecasting probability of current moment are utilized to calculate transferring probability between models, as a result, varied model set is adaptively designed according to maneuverability（turn） of target. Simulation results indicate that AIMM has higher precision, better stability than IMM algorithms in application of maneuver target tracking.