【作者】 史健芳； 赵丹丹； 何承伟； 冯志永；

【Author】 Shi Jianfang Zhao Dandan He Chengwei Feng Zhiyong (College of Information Engineering of Taiyuan University of Technology,Taiyuan 030024,China)

【机构】 太原理工大学信息工程学院；

【摘要】 D-S 证据理论中,进行目标识别需要获取各项证据的基本概率分配函数。鉴于人工神经网络具有表达模糊信息能力的特点,本文从多级数据融合的角度出发,将 D-S 证据理论、人工神经网络技术和多传感器数据融合技术有机结合,提出了一种利用神经网络构造基本概率分配函数,将神经网络的输出作为证据,利用证据组合规则、决策规则进行目标识别的三级数据融合目标识别系统模型。对飞机目标的仿真表明,本文设计的三级数据融合目标识别系统具有较高的识别率。此融合系统模型可推广应用于其它目标的识别。更多还原

【Abstract】 In the D-S evidence theory,it’s necessary to obtain the basic probability assignment function of all kinds of evidence in order to carry on target recognition.In view of artificial neural network having the ability to express fuzzy information,the paper embarks from the perspective of multi-level data fusion,or- ganically combines neural network with D-S evidence theory.A kind of target recognition system model with three-level data fusion was proposed.In the model,the basic probability assignment function is con- structed by neural network and the neural network’s outputs as evidence.Evidence combination rules and decision-making rules are used to realize target recognition.The simulation to aircraft target proves that the new designed system model has higher recognition rate.The model can be extended to other kind of target recognitions.更多还原