Neural network prediction model-based blast damage assessment of medical cabin in coast guard ship
10.19745/j.1003-8868.2024229
- VernacularTitle:基于神经网络预测模型的海警舰艇医疗卫生舱室爆炸损伤评估
- Author:
Meng-lei JIA
1
;
Yun-xia CHENG
;
Yan LI
;
Zun-feng DU
;
Chen-guang HAN
Author Information
1. 天津大学建筑工程学院,天津 300350
- Publication Type:Journal Article
- Keywords:
neural network;
coast guard ship;
medical cabin;
blast damage;
damage assessment
- From:
Chinese Medical Equipment Journal
2024;45(12):14-18
- CountryChina
- Language:Chinese
-
Abstract:
Objective To evaluate the blast damage of the medical cabin in the coast guard ship based on the neural network prediction model.Methods Firstly,a cabin finite element model was established with the typical medical cabin in some coast guard ship as the subject;secondly,ABAQUS simulation software was used to construct a blast damage simulation model and perform blast damage numerical simulation;thirdly,three neural network prediction models respectively based on Levenberg-Marquardt,Bayesian regularization and conjugate gradient algorithms were formed with the sample data from the simulation results and MATLAB software;finally,visualization over the prediction results of the neural network prediction models was carried out in terms of three blast conditions at the interior,bulkhead and near field of the cabin,and the blast damage to the cabin was analyzed with reference to the simulation results.Results Near the observation points the damage prediction results by the models were similar to those by the simulation.The blasts at the interior and bulkhead both resulted in obvious damages to the cabin framework,with serious effects and large damage areas;near-field blasts did not cause direct damage to the inter-nal framework of the cabin.Conclusion The neural network prediction model forecasts the blast damage to the medical carin in the coast guard ship,and references are provided for the structural design of the cabins in the coast guard ship.[Chinese Medical Equipment Journal,2024,45(12):14-18]
