The simulation experiment indicated, fault diagnosis based on neural network is in good agreement with measured values. More the model the initial failure sample we chooses to train the BP nerve network, better the network fault-tolerant and the stability are. In view of the complexity of operating equipment, only use a single parameter in the diagnosis often made wrong judgments. But the method of fault pattern recognition based on nerve network can fully use the information characteristic, and realization of the mapping between the input and output, obtains the accurate diagnosis result. The neural network has provided a new theory and technical method for the condition monitor and the fault diagnosis.
展开▼