Granular computing theory in the application of fault diagnosis

A novel intelligent fault diagnosis method based on binary granular computing-neural network (BGrCNN) was presented on this paper. To a fault diagnosis system of an internal combustion engine, the binary granular of granular computing (BGrC) method was used to reduce the information brought by the measured original data, and then feed-forward neural networks was added into the fault diagnosis system using the reduction samples. A simulation example was given in the end of this paper, and the simulation result was compared with the diagnosis results only used artificial neural network (ANN), which lies on the less time required in training and effectiveness of fault diagnosis. The conclusion indicates that the BGrCNN method can reduce the amount of useless data and bring an effective structure to neural network.