【作者】 张牧； 王建华； 万德成；

【Author】 ZHANG Mu;WANG Jian-hua;WAN De-cheng;Computational Marine Hydrodynamics Lab (CMHL),State Key Laboratory of Ocean Engineering,School of Naval Architecture,Ocean and Civil Engineering,Shanghai Jiao Tong University;

【机构】 上海交通大学船海计算水动力学研究中心(CMHL)船舶海洋与建筑工程学院海洋工程国家重点实验室；

【摘要】 本研究利用神经网络近似模型、Kriging模型和遗传算法来进行标准船模DTMB5415在设计航速下的总阻力优化。首先,采用对不同航速下船模的总阻力系数进行数值计算,并与实验值进行对比,保证总阻力系数求解的可靠性;然后使用自主开发的船型优化软件OPTShip-SJTU中的船型变换模块来获得一系列新船型,并计算这些船型在设计航速下的总阻力系数,形成样本船型数据库;接着基于该样本数据库,使用人工神经网络算法和Kriging算法分别学习构建出总阻力系数与船型参数之间的近似模型,并对此两个近似模型分别采用遗传算法求得两个最优解;最后将两个模型的对应的最优解,分别反馈到船型变换模块得到DTMB 5415船模最优船体型线。并且完成了两个近似模型得到的最优船型进行阻力的计算和对比分析。更多还原

【Abstract】 This paper uses neural network approximate model,Kriging model and genetic algorithm to optimize the total resistance of DTMB 5415 at design speed.First,the numerical calculation of the total drag coefficient of the ship model at different speeds is used and compared with the experimental value to ensure the reliability of the solution of the total drag coefficient.Then the ship type transformation module in the ship type optimization software OPTShip-SJTU is used to obtain a series of new Ship form.The solver is also used to calculate the total drag coefficients of these ship forms at the designed speed to form a sample ship form database.Based on the sample database,artificial neural network algorithm and Kriging algorithm are applied to learn to construct an approximate model between the total drag coefficient and the ship form parameters.Finally,the optimal solutions obtained by two approximate models are fed back to the ship type transformation module to obtain the optimal hull shape of DTMB 5415 ship model.Then the resistance calculation and comparative analysis of the optimal ship type are carried out.更多还原