为避免经典固定模型集合中多模型算法造成的模型间竞争并减小计算量, 基于有向图切换的变结构方法, 建立有向图切换规则和较完备的模型集合.通过模糊逻辑推理得到模型集合中各模型的匹配度, 以此代替传统多模型算法中的模式概率来计算,降低了计算的复杂度.采用核粒子滤波克服了标准粒子滤波没有考虑到的量测信息问题, 即粒子能够通过核密度估计后朝着状态的后验概率密度的模型移动, 使真实和估计模型之间的均方误差最小.仿真结果表明, 基于模糊理论的变结构核粒子滤波算法能较大地提高跟踪准确度和减少计算量.

High precision tracking system for airborne laser communication is usual to adopt multiple models algorithm with a fixed structure,which brings about unnecessary inter-competition among the amount of models and calculation burden.In order to resolve this problem,variable structure multiple-Model based on digraph switching and fuzzy inference was presented,graph switching rule and complete model set were established,the fuzzy inference mechanism was introduced to get matched degree for each filtering,the calculation complexity was decreased obviously.The Kernel particle Filter can move particles toward the posterior,root mean squared error between estimated and true model is minimum,the method could be introduced into fuzzy VSMM framework.Simulation results show that the algorithm improves the accuracy of tracking by reducing the competition of the models as well as reducing the computation burden.