船舶主尺度设计的高维多目标多方向进化算法

摘要

针对现有船舶主尺度优化模型往往只考虑经济性能，而忽略安全性能的问题，引入初稳性作为安全性指标，建立了4目标优化模型，并提出一种高维多目标多方向进化算法对其进行优化求解。通过一组方向向量将搜索空间分解成多个寻优方向，并利用改进的方向角差分算法结合SBX算子加强各方向上的寻优能力和方向间的信息交互；最后，以改进的模糊支配和密度估计因子构造精英保留策略，提高种群的先进性和分布性。实验结果表明，高维多目标多方向进化算法能够迅速、客观地选择合理的船舶主尺度，可以给设计人员提供更多的选择，为船舶初步设计提供了一种简单、高效的新方法。%In view of the fact that the existing principal ship dimensions optimization model tends to merely consider the economic performance, while ignores the problem of the safety performance, a four goal optimization model is introduced by using stability as a safety measure.In this paper, a high-dimensional many-objective multi-direction evolutionary algorithm is proposed.The search space is decomposed into several optimized directions using a set of direction vectors.Then,a modified hybrid direction angle differential evolutionary algorithm is combined with SBX operators to improve the searching ability in every direction and the information exchange between different direc-tions.And further, the improved fuzzy dominance and density estimation factor are used to maintain the size of the archive set to enhance the progressiveness and the distribution of the population.Experimental results showed that the high-dimensional many-objective multi-directional evolutionary algorithm can search the reasonable principal ship dimensions more quickly and objectively, which provided more choices for the designers.The new method is simple and efficient for preliminary ship designs.