基于组合式模糊神经网络的旋转机械故障诊断模型

A Faults Diagnosis Model for Rotating Machinery Based on Modular Fuzzy Neural Network

针对旋转机械故障所具有的层次性、相关性和模糊性的特点 ,提出1种基于组合式模糊神经网络的旋转机械故障诊断模型。它由第 1层的决策模糊神经网络和第 2层的多个诊断模糊神经网络组合构成 ,依据大隶属度优先为真原则进行推理 ,按推理的过程和隶属度的大小给出可能的故障结果及相应的隶属度值 ,供现场工程技术人员结合辅助故障特征参数等进行进一步的联想推理 ,得到最终的故障诊断结果。实验研究结果表明 ,该系统可以有效地对具有模糊性的单一故障和复合故障进行诊断。详细讨论了模型建立、隶属度函数定义和模型推理过程 ,并给出实验结果。

With the view of the fault characteristics of rotating machinery ,such as hierarchy , correlation and fuzziness , this paper puts forward a novel fault diagnosis model for rotating machinery based on modular fuzzy neural networks.A two-level integration strategy is applied in this model . The first level is composed of a fuzzy BP neural networks that determines which fuzzy neural networks block at the second level is activated according to the output member function values .The second level is composed of six fuzzy BP neural networks blocks,each of the blocks is further applied to diagnose some similar faults . The inference principle of the model is that the output node with larger member function value is first activated . With the input fault feature data , the model yields diagnosis conclusion that gives the possible faults with their corresponding member function value and inference route.According to this diagnosis conclusion and other information , the engineer can get further diagnosis conclusion . The experimental results show that this model improves the ability to diagnose multi-fault (several faults occur concurrently) and gives more information in the diagnosis conclusion . The structure of the model , the definition of the member function , the inference process of the model and experimental results are discussed in detail .