量测不确定下多传感器交互式多模型自适应滤波算法  

Multi-sensors interacting multiple models adaptive filtering algorithm in measurement uncertain

作　　者：胡振涛[1] 金勇[1] 李小偎[2] 

机构地区：[1]河南大学图像处理与模式识别研究所,河南开封475001 [2]西北工业大学控制与信息研究所,陕西西安710072

出　　处：《传感器与微系统》2011年第8期116-119,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(60972119);河南省教育厅自然科学基金资助项目(2008A510001);河南省自然科学基金资助项目(092300410158);河南大学自然科学基金资助项目(2010YBZR045)

摘　　要：针对量测不确定下多传感器多模型混合系统中量测信息的有效利用和融合问题,给出了一种多传感器交互式多模型自适应滤波算法。采用交互式多模型机制实现目标运动模式的确认;通过计算每个传感器的量测似然度完成对于不含扰动传感器量测数据的选取;利用传感器量测数据间统计距离的构建实现对于量测系统中剩余传感器量测数据是否包含扰动影响的判定,并在此基础上实现传感器量测数据的合理选取和融合。新算法量测不确定下扰动对于滤波精度的不利影响。理论分析和仿真实验验证了算法的有效性。Aiming at the problem of the effective utilization and fusion of multi-sensors multi-models hybrid system in measurement uncertain,a novel multi-sensors interacting multiple models adaptive filtering algorithm is proposed.In the new algorithm,the interacting multiple models mechanism is used to confirm the motion pattern of target.The measurement likelihood of every sensor is calculated to realize the selection of measurement without interference.The statistic distance among the measurements is constructured to judge whether the influence of interference exists in remaining sensors,and the reasonable selection and fusion of measurements is accomplished.The new method effectively improves the adverse influence of interference for filtering precision in measurement uncertain.The theoretical analysis and experimental results show the validity of the algorithm.

关 键 词：多传感器信息融合 交互式多模型 量测不确定 统计距离 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]