机动目标跟踪的HIMM算法研究

HIMM Algorithm for Maneuvering Target Tracking

交互多模型算法(IMM)的子滤波器都是基于Kalm an滤波的,它要求知道精确的噪声统计特性,然而在许多情况下噪声信号的统计特性是未知的,只能得到噪声信号的近似模型,这在一定程度上降低了IMM算法的跟踪精度。基于以上问题,将H∞滤波算法应用于IMM算法的滤波过程。H∞滤波对干扰信号的统计特性不作任何假设,与Kalm an滤波相比,H∞滤波器对噪声形式的不确定性不太敏感,鲁棒性好。在跟踪过程中还引入了一种数值稳健的模型概率计算方法,能有效防止计算过程中出现数值溢出现象,提高了算法的可靠性。最后通过仿真实验,证明了算法的有效性。

The sub-filters of Interacting Multiple-Model algorithm are based on Kalman filtering, which requires the accurate statistical characteristics of noise. Unfortunately, it is usually unknown and only the approximative models can be obtained, which may result in decrease of IMM performance. Therefore, the H∞ filter is used in IMM algorithm because the H∞ filter does not make any supposition to the noise statistical property. Compared with Kalman filter, it is not so sensitive to the noise, and has robustness to the uncertainty of noise. In the process of target tracking, a numerical robust method is introduced to calculate the model probability, which can effectively prevent the phenomenon of numerical overflow in computation process, and can improve the reliability of algorithm. Simulation results are provided to verify the proposed algorithm.