证据网络是证据理论的层次化推广,联合信任测度是证据网络中的一种推理方法,推理参数一般由专家经验给出.为了更准确地获得参数,提出一种将冲突大小作为评价函数,利用格雷编码遗传算法,通过评价输入来自动获取优化参数的方法.以机载多传感器平台为例的仿真实验表明,所提出方法的融合识别性能优于专家知识设定参数的方法,而且初始种群数量越大,越能获得好的识别结果. The evidence network is a hierarchical promotion of evidence theory, and the joint trust measure is a kind of reasoning method in the evidence network, and the reasoning parameters are generally given by experts’ experience.In order to get the parameters more accurately, a method is proposed to use the conflict size as the evaluation function, Gray coding genetic algorithm, which can automatically obtain the optimal parameters through the evaluation of input.Experimental results show that the proposed fusion fusion method is superior to the method of expert knowledge in setting parameters, and the number of initial population The bigger the better the recognition result is.