遗传算法的改进及其在超大型油船结构优化中的应用

遗传算法是一种基于适者生存理念的随机搜索算法,它具有极强的全局搜索能力,且不需要知道问题的导数信息.然而,简单遗传算法局部搜索能力差以及易于早熟.文章编制了一种基于实数编码的适用于连续型变量的遗传算法,比较适合于多峰函数的全局寻优,且对之略作改进,也可用于离散型变量优化.采用大量经典数学测试函数对该遗传算法的优化能力进行测试,取得了很好的优化结果.在此基础上,选用经典10杆桁架结构对该算法的寻优能力进行了验证.最后,以一艘超大型油船的典型中横剖面作为研究对象,选取396个设计变量,所有变量在优化过程中都进行了离散化处理,应用JTP规范1]作为校核依据,采用该遗传算法进行优化设计.经过优化后,船中剖面面积下降了2.6%.

Modification of genetic algorithm and its application in ship structural optimal design of a VLCC

Genetic algorithm is a kind of random optimization algorithm based on "best survive" concept.It has powerful global optimization ability,regardless of gradient information of the objective function.However,simple genetic algorithm has low efficiency in local extreme searching and might be easily premature.In this paper,a real-coded genetic algorithm used for continuous design variables optimization is presented,which is quite suitable for multimodal functions.By making some extensions,it can also be used in discrete variable optimization problems.For verification and validation of this modified genetic algorithm,first a set of classical test functions and a 10-bars truss structure are optimized.By comparison with the theoretical solution and published results,the efficiency and practicability of the algorithm are confirmed.For its application,the structures of the midship section of a very large crude oil carrier are optimized,396 structural dimensions are seen as design variables,the requirements of the JTP common structural rules are satisfied.All design varialbles are considered as discrete.After optimization,dimensions of some structural members are improved,and the total weight of the midship section is reduced.