摘要
针对机动目标跟踪问题中,固定结构多模型(FSMM)算法效费比不高、容易滤波器发散以及交互式多模型(IMM)算法模型的后验概率计算过程繁琐的问题,研究了一种防发散自适应网格模糊神经交互多模型(AD-AG-AIMM)算法。上述算法对标准卡尔曼滤波器进行了防发散处理,通过自适应网格调整实现了模型集自适应,通过ANFIS系统得到模型集中各个模型的匹配度。仿真结果表明,AD-AG-AIMM算法与标准的IMM算法相比,可以有效提高多模型算法的精度和效费比,特别是对目标机动的适应能力显著提高,且适合工程实用。
ABSTRACT ： For the maneuvering target tracking problem, the efficiency-cost ratio of fixed structure multi model of （FSMM） algorithm is low, which easily leads to the filter divergence, and the calculation process of posterior proba- bility of interacting multiple model （IMM） algorithm is too complex. In this article, an algorithm of fuzzy neural in- teracting multi-model is proposed based on anti-divergent adaptive grid （AD-AG-AIMM）. The algorithm performed an anti-divergence treatment in standard Kalman filter, and realized the model set adaptation through adjusting adap- tive grid. Thus, each matching degree of model in model set was obtained through ANFIS system. Simulation results prove that compared with standard IMM algorithm, the AD-AG-AIMM algorithm can effectively improve the accuracy and efficiency-cost ratio of multi-model algorithm. Especially, it remarkably improves the self-adaptive ability of target maneuvering, which is suitable for engineering.
出处
《计算机仿真》
北大核心
2018年第1期239-242,250,共5页
Computer Simulation
关键词
机动目标跟踪
防发散
自适应网格
模糊神经
Maneuvering target tracking
Anti-divergent
Self-adaptive mesh
Fuzzy neural
