Damage detection on hull girder of ship subjected to explosion loading

Abstract

Considering the limitation of current direct measurement method for ship damage identification, an indirect damage identification method based on dynamic parameters is taken into account. A 1500-ton ship simplified as the hull girder is taken as the research object. Artificial neural network method, the coordinate modal confidence (COMAC) and modal curvature index are applied to identify the ship's damage under different conditions through numerical experiments. Migration matrix method is introduced to obtain the damage database. The damage identification accuracy of two indexes based on modal shapes is discussed. Results show that the neural network intelligent method based on frequency changes can well locate and quantify the damage of the ship structure. Damages were correctly located, and the average accuracy of damage severity reached 95.78%. For damage indexes based on modal shapes, COMAC requires less data to identify the damage and has high calculation efficiency, but there is a certain range of error. When used in conjunction with various modes of the index modal curvature, damage localization is relatively accurate. Both indicators based on modal shapes can reflect the damage degree to some extent, but can't accurately identify the specific extent of damage.