最小熵反馈式变结构多模型融合算法

统变结构多模型方法(VSMM)在处理高机动日标状态估计问题和大观测误差时存在因模型集合与真实模式匹配欠佳导致估计质量下降的问题.本文结合最小信息熵准则(ME)提出一种反馈式变结构多模型融合算法(MEVSMM),将在所有模型相关的在线估计信息进行反馈,进而选取状态估计分布信息熵最小的模型集作为当前有效模型集,计算多模型估计结果；结合粒子滤波算法(PF)和设计擂台赛算法(CM),构造了易于工程实现的次优算法(PF-MEVSMM).理论分析与仿真表明,与传统VSMM算法相比,本法具有模型集更精简、有效,融合估计结果鲁棒性更强、精度更高的优点.

Minimum entropy and feedback structure-based algorithm for variable structure multi-model fusion

When applying the traditional variable structure multi-model algorithms (VSMM) to the state estimation problems of high maneuver and large observation error, one may face the difficulty of estimation degradation caused by the mismatch between the prior model sets and the real modes. To deal with this difficulty, a minimum entropy VSMM algorithm (MEVSMM) is proposed based on the principle of minimum entropy. First, all model-based estimations are fed back online. Second, the optimal solution is found if the distributions of the related estimations satisfy the minimum entropy condition. A sub-optimal algorithm (PF-MEVSMM) is also designed by employing the particle filter (PF) and the challenge-match algorithm (CM). Comparing to some existing VSMM algorithms, the results demonstrate that the proposed algorithm can provide refined model sets with smaller sizes, as well as more robust and accurate estimation results.