模型转移概率自适应的交互式多模型UKF算法

本文给出了一种基于UKF算法的参数自适应交互式多模型方法,较好的解决了非线性条件下机动目标跟踪的问题,可获得比基于EKF算法的交互多模型方法更好的稳定性和计算精度,还避免了复杂的Jacobi矩阵运算;由于该方法结合了模型概率转移自适应技术,实现了对模型转移矩阵的在线估计,降低了人为因素的影响。最后,通过Monte Carlo仿真进一步验证了该方法的正确性和有效性。

An Interacting Multiple Model UKF Algorithm with Adaptive Markov Transition Probabilities

This paper presents an interacting multiple model UKF algorithm with adaptive markov transition probabilities, which can effectively tackle the maneuvering target tracking problem in practice. This algorithm demonstrates better computational stability and precision than those based on EKF, and avoids computing the Jacobi matrix. Moreover, by using the adaptive Markov transition probability technique, the transition matrix can be modified in real time according to measurements. Monte Carlo simulation shows the efficiency and effectiveness of the algorithm.