Abstract:

This paper describes the application of neural networks for structure failure detection for a fighter. As compared with traditional model based failure detection for nonlinear systems, neural network methods have the advantages of strong nonlinear approximation ability and fast detection. A prediction neural network scheme for fault detection has been developed, along with multiple step direct prediction algorithm and threshold selection principle in this paper. Finally, the proposed scheme is demonstrated using the model of a fighter and the results show that the neural network method is an effective tool for structure fault detection of a fighter.

Key words: prediction neural network, failure detection, threshold