Neural Networks for Pattern Recognition
Neural Networks for Pattern Recognition
Computers and Industrial Engineering
Hi-index | 0.00 |
In the case of fault diagnosis of the plant machinery, diagnostic knowledge for distinguishing faults is ambiguous because definite relationships between symptoms and fault types cannot be easily identified. This paper propose a sequential fuzzy diagnosis method for condition monitoring of a rolling bearing used in a centrifugal blower by the possibility theory and a neural network. The possibility theory is used for solving the ambiguous problem of the fault diagnosis. The neural network is realized with a developed back propagation neural network. As input data for a neural network, the non-dimensional symptom parameters are also defined in time domain. Fault types of a rolling bearing can be effectively, sequentially distinguished on the basis of the possibilities of the normal state and abnormal states at early stage by the fuzzy diagnosis approach. Practical examples of diagnosis are shown in order to verify the efficiency of the method.