基于似然函数的自适应Singer模型滤波算法

Singer模型滤波算法可以对机动目标进行有效跟踪,但其模型参数的确定依赖于先验知识,且一旦确定,将在滤波过程中不再变化.因此,当事先确定的参数与目标机动不匹配时,跟踪精度会变得比较差.针对模型参数失配时,传统Singer模型不能有效跟踪机动目标的问题,提出一种自适应Singer模型滤波算法.在滤波过程中,构造多模型的模型似然函数,并随着滤波过程实时计算模型似然函数,根据似然函数的变化,自适应调整Singer模型加速度参数.仿真表明,该算法能够有效跟踪目标不同的机动情况,滤波效果较固定参数的Singer模型算法和离散自适应Singer模型算法更优.

An Adaptive Singer Model Filter Based on Likelihood Function

The Singer model filter is an effective algorithm for tracking maneuvering target. Since the parameters are determined depending on prior knowledge, and they can''t be changed in the filter process once determined. When the parameters determined in advance don''t match with the target maneuvering, it will bring poor tracking performance. In response to these problems which were caused by model mismatching in the traditional Singer model filter, an adaptive Singer model algorithm has been posed, which is based on the model likelihood function. And the model likelihood function is calculated in all the filter process. And the Singer model acceleration parameters is adjusted adaptively according to the likelihood function''s value. The simulation results show that the algorithm can track targets in different maneuvers effectively. It performances best in the comparison with traditional Singer model filter and the discrete adaptively Singer model filter.