【摘要】 为了诊断直升机自动驾驶仪故障并诊断到外场可更换单元,在分析自动驾驶仪中信息与知识的基础上,提出了基于知识推理的融合诊断策略,分别采用系统中的案例、规则和模型知识进行信息处理和决策生成,并提出外场故障诊断的自学习型融合框架和实现方法,即首先在部件级进行局部融合决策,然后再结合系统级的信息和知识进行系统级融合决策。给出了利用D-S证据理论进行决策融合的可信度分配方法。试验结果表明所提出的方法能有效降低诊断结果的不确定性。更多还原

【Abstract】 In order to improve the ability of fault isolation to Line Replaceable Unit(LRU) level,this paper analyzed the information and knowledge of autopilot of helicopter,and brought forward a new knowledge based fusion strategy.It processed the information by using the cases,rules and model knowledge to generate the decision respectively,then the self-learning two level fusion frame and implementation method for outfield fault diagnosis was put forward.Firstly the component level fusion was carried out,then the information and knowledge of system level were combined for final fault decision-making.The belief assignment method of D-S evidence theory was also given to realize the decision level fusion.The experimental results show that the uncertainty can be reduced remarkably based on the proposed idea.更多还原