Machine learning-based rapid forecasting method for dynamic response of hull grillage structure under underwater explosion load

Abstract

The invention relates to a machine learning-based rapid forecasting method for dynamic response of a hull grillage structure under an underwater explosion load. According to the method, any Euler-Lagrange method is adopted to carry out numerical calculation on ship body grillage structure response under the underwater explosion load to obtain a data sample; according to a hull grillage structure, dimension reduction is performed on the sample database so as to shorten the deep neural network training time; a deep neural network is adopted to learn the sample database after dimension reduction, and the learning effect is verified; the ant colony algorithm is adopted to optimize the structure and hyper-parameters of the deep neural network, the training efficiency and forecasting precision of the deep neural network are improved, and the deep neural network with the optimal generalization effect is output; and post-processing the dynamic response forecast result of the hull grillage structure under the underwater explosion load by using the deep neural network.

Description

technical field [0001] The invention relates to the technical field of ship damage under underwater explosive loads, and relates to a machine learning-based rapid prediction method for the dynamic response of hull plate structures under underwater explosive loads. Background technique [0002] Surface ships are the main force of naval equipment, and the damage caused by underwater explosions will pose a serious threat to the combat effectiveness and vitality of ships. Rapidly and accurately evaluating the damage characteristics of ships under underwater explosive loads in the battlefield environment has extremely important military value. For the research of underwater explosion, at present, it is mainly based on numerical simulation and model test. The published experimental data are basically limited to simple frame structures and experiments with small scale ratios. Small-scale scale model experiments have scale effects, and the boundary conditions are difficult to keep...

AI Technical Summary

Problems solved by technology

There are scale effects in the small-scale scale model experiment, and the boundary conditions are difficult to keep consistent with the actual working conditions. It is difficult to reverse the original model, and it can only be used as a preliminary mechanism study.

Large-scale experiments and actual ship tests are costly, and are limited by conditions such as feasibility and safety

Images