建筑设备故障的模糊神经网络诊断方法

A Failure Diagnosis for the Architectural Equipment Using Fuzzy Neuro Approaches

提出故障诊断系统的总体构造和各个故障原因诊断的神经网络结构，依据模糊推理和Ｉｆ—Ｔｈｅｎ型产生式规则，采用隶属度传递法则，进行故障知识处理，把表示原因和症状间因果关系的频数分布作为网络层间单元的权重。这个诊断方法的主要特征是能够定量地表示所归纳的故障原因。

The overall architecture of diagnostic systems and the fuzzy neural networks of failure diagnosis are presented. The transfer of membership grades based on fuzzy infe rence and IF THEN production rules is applied for diagnostic knowledge process. The frequence distribution about the relation between the cause and symptom in failure modes is taken as the connection weights in units. The main feature of this approach is that the failure modes can be quantitatively indicated. The diagnosis of diesel engine as an example is shown.