基于多模型后验概率的机动目标跟踪方法

A Maneuvering Target Tracking Method Based on Multiple-Model Posterior Probability

传统单模型算法很难满足对机动目标的跟踪精度需求,自适应模型、多模型成为该领域的研究热点。转弯模型因其形式简单、计算复杂度低等特点在多模型跟踪中被广泛研究和使用。但是,网格调整算法估计转弯角速度的方法中由于机动目标角速度未知而无法确定最小的网格间隔,导致存在难以精确估计真实角速度且估计速度慢等问题,基于后验概率的机动目标跟踪方法能够有效地解决该问题。通过对多模型后验概率的计算获得各模型对应的权值,提高了真实角速度的估计精度和既定目标跟踪算法的估计性能。最后通过仿真试验说明了改进算法的有效性。

Traditional single-model methods are hard to meet the tracking accuracy of the maneuvering target,so the adaptive models and multiple models have become the focus in this field.The turn model has been studied widely because of its simple form and low computational complexity.However,due to the unknown turn angular velocity of the maneuvering target,the smallest grid distance cannot be determined in the grid adjusting algorithm,resulting in inaccurate estimation of the true angular velocity and slow estimation velocity.A modified maneuvering target tracking method based on the posterior probability can solve the problem well.Through calculating of the multiple models＇ posterior probability,the weights of the corresponding models in the multiplemodel algorithm are obtained.The simulation results demonstrate the effectiveness of the modified method.