Application of hybrid wavelet neural network for missile fault diagnosis system

In missile weapon system, exact fault prediction and diagnosis is very important for missile security, according to the specialty and complexity of the missile fault diagnosis, a novel expert system design method based on the hybrid neural network ensembles is proposed in this paper. To improve the limitation of applying traditional fault diagnosis method to the diagnosis method of the diagnosis of missile fault, with large amounts of typical missile fault samples and raw measurable parametric data available, the missile fault diagnosis system based on wavelet neural network ensembles can be created applying general construction techniques of the wavelet neural network fault diagnosis system, including signal binary wavelet transform, fault feature extraction/ selection and network training. The back-propagation (BP) algorithm is used to fulfill the parameter initialization and the neural network structure (WNN). By means of choosing enough practical samples to verity the wavelet neural network (WNN) and the information representing the faults is inputted into the trained WNN, and according to the output result the type of fault can be determined. It's proved that through diagnosis of the missile from several different sides by use of different parameters the diagnosis result is more reliable. The method can be generalized to other devices' fault diagnosis.