基于期望极大化算法的转弯机动目标跟踪

针对目标做转弯机动时产生的运动模式的不确定性问题，提出了基于期望极大化（EM）算法的转弯机动目标跟踪算法，采用转弯速率来描述目标的转弯机动，并将转弯速率作为待估量，用EM算法对转弯速率序列进行估计，从而获得对目标状态的精确估计．最后，给出了该算法的批处理形式和递推形式．新算法有效地提高了目标的跟踪精度．Monte—Carlo仿真表明，与标准的交互式多模型算法相比，批处理EM算法使目标的位置和速度的跟踪精度提高了70％以上，当转弯速率处于稳态时，递推EM算法使目标的位置和速度的跟踪精度提高了20％以上．

Maneuvering Target Tracking with Coordinated Turn Motion Based on Expectation Maximization Algorithm

Aiming at the target motion uncertainty when target has a coordinated turn maneuver,a state estimation algorithm based on expectation maximization(EM) algorithm was derived.The coordinated turn maneuver was described by the turn rate.Firstly,the coordinated turn series were estimated using EM algorithm;then,with the estimated coordinated turn series,the target states could be estimated accurately and the batch and recursive EM algorithms were given.The tracking accuracy can be greatly improved using the new algorithm.Monte-Carlo simulation(results) show that the position and velocity accuracy based on the batch EM algorithm are(improved) by 70% compared to the interacting multiple model algorithm.When the turn rate is stable,the position and velocity accuracy are improved more than 20% by the recursive EM algorithm.