A rapid prediction method for underwater explosion impact response based on compartment structural characteristics

A novel method for rapidly predieting the peak impact response of ship structures under underwater near-field non-eontact explosion loads was proposed. A multi-compartment ship model was constructed, and the experimental conditions were designed using the Latin hypercube sampling method. Sample measurement points were arranged at typical positions of each layer of decks and double bottoms at different rib positions along the transverse sections for each set of conditions. The peak impact response of each sample measurement point was extracted to establish a database of peak impact responses. A rapid prediction model for the peak impact response of ship structures was established using the non-dominated sorting genetic algorithm- TJ I (NSGA- TJ) optimized BP (back propagation) neural network method. The Parameters of the explosive (TNT (trinitrotoluene) equivalent, stand-off distance, angle of attack), parameters of the ship structure form (number and positions of decks, longitudinal and transverse bulkheads, etc.), and the relative positions of measurement points to the detonation point were separately taken as inputs to the neural network. Peak acceleration, velocity, and displacement were used as Outputs for neural network training. After training, the accuracy of peak impact responses for 6 sets of explosion conditions was verified. The results indicate that compared to existing finite element Simulation prediction methods, the method proposed in this paper is simpler and more practical. Compared to existing rapid prediction methods, this method takes into account the influence of ship structural features on underwater explosion impact response, thus better meeting the requirements for rapid prediction of underwater explosion impact response of ship structures. © 2025 Chinese Vibration Engineering Society. All rights reserved.