基于IMM-PF的再入目标数据融合算法与仿真

摘要

A data fusion algorithm for reentry target based on Interactive Multiple Models-Particle Filter(IMM-PF) is developed in order to improve the tracking accuracy of ballistic reentry target. The algorithm combines IMM with PF, and utilizes limited motion model to approximate motion state of reentry target. On the basis of discretely processing the motion equation and the observation equation of reentry target, the state estimation value and covariance of each model are calculated by PF algorithm, and the residual re-sampling method is adopted to overcome the degradation of the particle’s weight. In the process of PF, the system continues to improve the probability density function of particles, and constantly updates probability of each model, so the accurate estimation of unknown parameters in the reentry target tracking is achieved. Simulation results show that compared with other algorithms, this algorithm is of higher tracking accuracy, shorter running time, better algorithm convergence, and suitable for fast and accurate tracking of the reentry target.%为了提高弹道再入目标的跟踪精确度，提出了一种基于交互式多模型粒子滤波(IMM-PF)的再入目标数据融合算法。该算法将交互式多模型和粒子滤波相结合，用有限个运动模型来逼近再入目标的运动状态，在对再入目标的运动方程和观测方程离散处理的基础上，采用粒子滤波算法计算各模型的状态估计值和协方差，并采用残差重采样方法克服了粒子权重的退化问题；在粒子滤波过程中，系统不断改善粒子的概率密度函数，不断更新各个模型的概率，从而实现对再入目标跟踪中未知参数的精确估计。通过实例仿真表明：与其他算法相比，该算法的跟踪精确度较高，运行时间较短，算法收敛性较好，适合对再入目标的快速、精确跟踪。