摘要
多诊断Agent(智能体)合作求解可以有效地解决单一诊断方法或专家知识水平的局限性,但是也带来了诊断结果的综合评价问题。根据多Agent诊断结果的特点,对证据理论中的概率分配函数进行了约束和具体化,引入了不确定性函数,从而有利于在多Agent诊断结论综合中的应用。在此基础上,提出了多诊断Agent合作诊断解的综合步骤,并通过各诊断Agent的结论一致、部分冲突和完全冲突3个实例进行了综合分析,说明了该方法的有效性。用该方法进行诊断结果的综合能够提高多Agent诊断系统对故障模式的分类识别能力。
Cooperative solution of multi-diagnosis agent for common task can effectively overcome some disadvantages of unitary diagnosis method or expert knowledge level, but also brings problems about synthesis and appraisement of diagnosis conclusions. In order to synthesize diagnosis conclusions of multi-agent system, the probability distribution function in evidence theory was restrained and reified, according to the characteristics of multi-agent diagnosis results, and the uncertainty function was introduced into evidence theory. On above foundation, the result synthesis steps of multi-diagnosis agent were presented, and the validity of this method was explained through analyzing three examples including consistent results, partial conflict results and complete conflict results. The syntheis of diagnosis results with this method can raise the ability to classify the fault patterns in multi-agent diagnosis system.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2006年第4期737-740,共4页
Acta Armamentarii
关键词
人工智能
证据理论
智能体
故障诊断
概率分配函数
artificial intelligence
evidence theory
agent
fault diagnosis
probability distribution function