基于改进NSGA-Ⅱ的船舶管路路径设计

Ship pipe route design based on improved NSGA-Ⅱ

针对船舶机舱空间设备繁多、约束复杂,导致人工布管效率低、易出错,而现有布管算法只能生成少量布局结果,不能为管路工程师提供丰富参考等问题,提出一种用于求解船舶管路路径设计的改进多目标优化算法。用网格分解法描述障碍物和主要布局约束,将管路长度、折弯数目、贴壁敷设、折弯限制、“凹兜”数目、成束布局效果作为优化目标参与个体支配关系确定和择优过程,以连接点策略和改进A*算法构建种群个体,为带精英策略的快速非支配排序遗传算法(NSGA-Ⅱ)框架引入局部爬山搜索、精英个体非支配合并、种群多样性保持等改进策略以提高算法寻优能力,提出基于改进多目标算法和人工决策进行多管路成束布局和分支管路布局的算法流程。最后,通过仿真算例和实际算例验证了所提算法的可行性和先进性。

To solve the piping problems in ship engine room such as various equipment and complex constraints lead to low efficiency and error prone of manual pipe design,and some existing algorithms only generate a small number of routing results which cannot provide engineers with rich references,an improved multi-objective optimization algorithm for solving Ship Pipe Route Design(SPRD)was proposed.The cell decomposition method was used to describe the obstacles and the main routing constraints,and the length of path,the number of bends,the location to walls and floors,the limited distance between bends,the number of“pocket”structures and the effect of bundle layout were taken as the optimization objectives to determinate the dominance and comparison relations of individuals.The population of algorithm was constructed by the connection point strategy and the improved A*algorithm.For the fast elitist Non-dominated Sorting Genetic Algorithm(NSGA-Ⅱ)framework,some improved strategies such as local hill-climbing search,non-dominated merge of elite individuals and population diversity maintenance were introduced to improve the optimization ability.Then the workflows for routing parallel pipes and branched pipe based on the improved algorithm and human decisions were put forward.The feasibility and advancement of the proposed algorithms were demonstrated by the simulation and actual piping cases.