基于遗传算法的船舶避碰决策辅助

摘要

针对海上船舶避碰问题,提出一种基于多种群遗传算法(Genetic Algorithm,GA)自动生成最优避碰路径的船舶避碰辅助决策方法.该算法采用多种群协同进化的方式,通过建立移民算子和人工选择算子保持种群之间的联系.这种改进的GA不仅能解决标准GA中遗传算子参数设定的问题,而且能提高算法的有效性和效率.利用船舶避碰方面的知识和启发式方法生成初始路径,使其决策方向符合避碰规则的要求,并对种群中的个体进行适应度评价与优化.以精英种群中最优个体的最少保持代数作为算法终止条件,这种判据充分利用GA在进化过程中的知识积累,比最大遗传代数判据更为合理.仿真结果证明了多种群GA在辅助船舶避碰决策方面的可行性和优越性.%For the ship collision avoidance problem at sea, a method to ship collision avoidance decision aids is presented, where the multi-population Genetic Algorithm(GA) is adopted to automatically generate the optimal path of collision avoidance.In the algorithm, multiple populations evolve simultaneously, and the immigration operator and the artificial selection operator are established to keep relationships among populations.The improved GA can not only solve the problem of parameter settings for genetic operators in the traditional GA, but also improve the algorithm's effectiveness and efficiency.The knowledge of ship collision avoidance and the heuristic method are used to generate the initial path whose decision direction meets the requirements of collision avoidance rules, and to carry out the fitness evaluation and optimization for individuals of the populations.The termination condition of the algorithm is the minimum iteration times predefined that the optimal individual is kept in the elite population, which makes full use of the knowledge accumulation in the iteration process of GA, and is more reasonable than the criterion of the maximum genetic iteration times.Simulation results demonstrate the feasibility and superiority of the multi-population GA in aiding ship collision avoidance decision.