Abstract: [Objectives] In order to solve the problem of breach damage assessment of ship structure under the action of underwater explosion, a breach prediction model based on PCA-BOA-KNN is established. [Methods] Firstly, the finite element model of the five-compartment section and seven-compartment section is established respectively, and the explosion simulation analysis is carried out for 21 groups of underwater explosion conditions. Then based on principal component analysis (PCA), the peak acceleration, peak velocity, peak displacement, peak stress, and peak overpressure are downscaled to obtain two eigenvalues. Finally, the results of principal component analysis are substituted into the KNN model of Bayesian network optimization (BOA), and the breach prediction model is established to predict the breach at different profiles of the ship under a set of working conditions. [Results] The results show that the cumulative contribution of the first two factors extracted by principal component analysis is 85.165%, so the first two factors can represent the main information of the five feature quantities. The results based on the PCA-BOA-KNN breach forecasting model are basically consistent with the simulation results.[Conclusions] The prediction modeling method proposed in this paper is effective for the ship structural breach prediction established in this paper, and has some reference value for the hull structural breach prediction of different main scales.