基于K-L信息因子的变结构多模型算法模型集自适应研究

针对传统多模型机动目标跟踪算法对模型数量的增长会产生组合爆炸和模型竞争现象，降低跟踪系统的性能，提出了一种模型集自适应的变结构多模型算法。采用基于Kullback-Liber（K-L）信息因子分析多模型算法中模型集各模型之间的匹配程度，并在此基础上实现对模型集的自适应调节，同时推导了各时刻模型概率与模型之间的模型转换概率矩阵，从而为变结构多模型算法的实现提供了基础。仿真实验验证了所提算法的可行性和有效性。

Model-set adaptive algorithm of variable structure multiple-model based on K-L criterion

The traditional multiple-model algorithm of maneuvering target tracking always causes combinatorial explosion and model competition with the growth in the number of model， which greatly deteriorates the performance of the tracking system. To solve this problem, a kind of model-set adaptive algorithm of variable structure multiple-model algorithm is proposed. Based on the Kullback-Liber (K-L) criterion, the matching situation of each model-set is analyzed, and then the model-set adaptive regulation is carried out. The model probability and transition probability matrix between models at each moment are derived, which provides the support for the realization of the variable structure multiple-model algorithm. Simulation results demonstrate the feasibility and effectiveness of this proposed algorithm.