The impulsive environment calculation of ship subjected to underwater explosion is the key of shock resistance design of ship equipment. The approach to improve the accuracy of calculation about impulsive environment was concerned with the researchers.  In this research, ten FEM ships with varied dimensions and types were established for prediction. On varied decks of ships, 400 groups were uniformly distributed when the ship subjected to underwater explosions. More than 120 million impulsive environment data was produced to build the impulsive environment data base. In this paper the impulsive environment prediction model was constructed base on radial basis function network. The input parameters were main dimension parameters, underwater explosion cases and measuring point locations, and the only output parameter was velocity shock spectrum. The model network parameters were optimized by clustering algorithm. Then the prediction results about the impulsive environment of an unknown ship subjected to specified underwater explosion case expresses the high precision, the better generalization ability and optimal robust performance. The method present a new choice for rapidly prediction of ship impulsive environment in design stage.
Key words: shock response spectrum of ship; prediction; RBF neural network; optimization algorithms