基于规则的故障诊断计算复杂性分析

由于大量故障的相互作用使得多故障识别及计算量成为故障状态判断的难题.在复杂系统故障诊断中,基于规则的故障诊断特别是专家系统已得到广泛应用.规则数量是诊断精度的关键因素,潜在规则数量直接影响搜索匹配的计算量,其计算量将成为大系统诊断的重要指标.在分析规则构成的基础上,以征兆数量为输入,确定了诊断系统规模判定指标.以汽轮机轴系振动专家系统为例,展示了规则诊断问题的复杂性和诊断规模的确定方法.

On Computational Complexity of Rule-based Fault Diagnosis

What is expected from a diagnosis system is its ability of coping with a large amount of computational work and of identifying a multi-number of faults.Problems arise due to the interaction among different faults.Rule-based fault diagnosis,and especially expert systems,are already widely used in diagnosing faults of complex systems.The number of rules concerned is the dominant factor,affecting diagnosing accuracy,and the number of latent rules has a direct influence on the amount of computational searching and matching work required;which may serve as an important diagnosis index for large systems.Based on the analysis of the rules' build-up and taking the number of symptoms as the input,the scope of rule-based diagnosis is being identified.Exemplified by the scope of an expert system for diagnosing rotor-string vibration of steam turbines,the complexity of rule-based diagnosis and the way of determining the diagnosing scope is demonstrated.