水下爆炸载荷下柔性支撑板架结构防护效能快速预报与优化方法  

Rapid prediction and optimization method for protective effectiveness of flexibly supported plate structure under underwater explosive

作　　者：郭桐桐 郭煜 余俊[1,2] 陈娟 王海坤[1,2] 张伦平[1,2] GUO Tongtong;GUO Yu;YU Jun;CHEN Juan;WANG Haikun;ZHANG Lunping(China Ship Scientific Research Center,Wuxi 214000,Jiangsu,China;Taihu Laboratory of Deep Sea Technology Science,Wuxi 214000,Jiangsu,China;China Ship Development and Design Center,Wuhan 430000,Hubei,China)

机构地区：[1]中国船舶科学研究中心,江苏无锡214000 [2]深海技术科学太湖实验室,江苏无锡214000 [3]中国舰船研究设计中心,湖北武汉430000

出　　处：《爆炸与冲击》2024年第10期167-176,共10页Explosion and Shock Waves

摘　　要：为实现柔性支撑板架结构在水下爆炸下防护效能的快速评估和设计优化,基于高置信度仿真,建立了水下爆炸作用下柔性支撑板架防护效能的评估方法并开展试验验证。采用验证后的高置信度仿真方法生成样本工况数据,并通过径向基神经网络模型构建能快速评估柔性支撑板架结构防护效能的代理模型。结合多岛遗传算法对建立的代理模型进行防护结构高防护效能和轻量化的多目标优化并获取最优结构参数。建立的快速预报与优化方法可以为相关的结构设计优化提供重要的技术支撑。In order to make a rapid assessment and design optimization of the protective performance of flexibly supported plate structure subjected to underwater explosion,a high-confidence simulation method is first established for the protective performance of flexibly supported plate structure subjected to underwater explosion.Then,underwater explosion tests were conducted on the flexibly supported plate structure to validate the computational accuracy of the developed high-confidence simulation method by comparing the deformation between the simulation results and the experimental results.The thickness of the blast-facing panel,the thickness of the flexible supports,and the thickness of the stiffened web are identified as the three key characteristic parameters that affect the protective performance of the flexibly supported plate.Utilizing optimized Latinhypercube sampling method,15 sample conditions are extracted from the sample space.The validated high-confidence simulation method is then used to generate protective performance data for these 15 sample conditions,which is subsequently employed to construct a proxy model for rapid assessment of the protective performance of the flexibly supported plates by using a radial basis function(RBF)neural network.The accuracy of the proxy model is assessed by using 5 randomly selected conditions,and the results show that the prediction error is within 7%,indicating a high level of prediction accuracy.The multi-island genetic algorithm(MIGA)is applied to the proxy model to perform multi-objective optimization and obtain a pareto set of solutions.The condition with the maximum specific ultimate energy absorption per unit mass is selected as the optimal structural parameters for the flexibly supported plate,achieving the goals on enhancing the ultimate protective performance and reducing the total structural mass.The rapid prediction and optimization method developed in this study provides significant technical support for the design and optimization of flexibly supported plate,

关 键 词：水下爆炸 柔性支撑板架 防护效能 抽样方法 神经网络 多目标优化 

分 类 号：O383.1[理学—流体力学]