中文摘要：

提出了一种双层网格变结构多模型（DLG-VSMM）算法用于复杂系统的状态估计．DLG-VSMM算法中的全体模型集由1个拓扑结构自适应调整的基础模型网格M和1个能在系统模式空间自由滑动的修正模型网格C组成．首先利用可能模型集技术实时对基础模型网格结构进行调整；然后利用自适应网格技术产生修正模型网格的区域中心并生成自适应修正模型网格；最后利用最优融合原理得到系统的整体估计．仿真结果表明，DLG-VSMM算法在计算量增加不大的情况下提高了算法的跟踪精度．

英文摘要：

A variable structural multi-model method is propose based on a dual-layered grid （DLGVSMM） for state estimation of hybrid system. The total model set of DLG-VSMM is composed of a basic grid. The structure of the new method can be adaptively tuned and an amendatory model grid C which can move freely in the space of system mode. Firstly, the likely model set technique is used to adjust the to pology structure of the M. Then, the central position of C produced by the adaptive grid approach is further used to produce the adaptive amendatory model grid C. Finally, the total estimation can be obtained by optimal information fusion. Simulations demonstrate that the proposed DLG-VSMM wiil improve the tracking performance at a limited extra computational cost.