泊位-岸桥联合分配模型的模拟植物生长交替进化算法

Alternate evolution algorithm based on plant growth simulation for berth-quay crane joint allocation model

为了综合优化集装箱码头泊位和岸桥联合分配计划,分析了二者的相互独立性和系统关联性;利用相互独立性,分别针对泊位和岸桥分配建立了以平均在港时间和作业成本最小为目标的2个优化子模型;利用系统关联性,构建了泊位-岸桥联合分配的约束条件,将2个子模型紧密联系在一起,建立了完整的泊位-岸桥联合分配模型;分析了联合分配模型的特点,设计了模拟植物生长交替进化算法求解模型,利用基于模拟植物生长算法的交替进化算子对种群中每个个体的2个目标进行交替优化,进而实现种群进化,通过算法框架实现非支配解筛选,经多次种群进化和非支配解筛选,获得泊位-岸桥联合分配的Pareto满意解集;针对大连港集装箱码头3d中共计31艘真实到港船舶的泊位-岸桥联合分配计划进行优化计算,并与多目标遗传算法的计算结果进行对比。计算结果表明:共获得13个满意解,船舶平均在港时间为7.47~9.44h,使用岸桥次数为85~96台,作业总成本为20.868~21.114万元;与多目标遗传算法相比,进化算法的运算速度提高了6.07%,所得非支配解的数量增加了4个,增加幅度为30.76%,且计算结果更趋近于Pareto前沿,联合分配计划优化程度较高。可见,采用模拟植物生长交替进化算法能够最大限度地保持种群进化过程中个体的独立性,获得更多的非劣解,且交替进化的方式能够使结果更逼近Pareto前沿。

To optimize the berth-quay crane joint allocation plans in a container terminal synthetically,the independence and system relevance between berth and quay crane were analysed.Based on the independence,two optimized sub-models were established,one to target the berth with the minimum average time in port,and the other to target the quay cranes with the minimum operating cost.Based on the system relevance,the constraint conditions for berth-quay crane joint allocation were constructed,the two sub-models were linked closely,and a complete berth-quay crane joint allocation model was established.The characteristics of the joint allocation model were analysed,and an alternate evolution algorithm based on plant growth simulation wasdesigned to solve it.Alternate evolution operators based on the plant growth simulation algorithm were used to alternately optimize the two targets of each individual in the population to achieve population evolution.The non-dominated solutions were screened through the algorithm framework.After multiple population evolutions and non-dominated solution screening,the Pareto satisfactory solution set for the berth-quay crane joint allocation was obtained.A berthquay crane joint allocation plan for 31 vessels arriving within 3 din the container terminal of Dalian Port was optimized and compared with the multi-objective genetic algorithm.Calculation result shows that 13 satisfactory solutions are obtained.The average vessel time in the port is 7.47-9.44 h,the number of quay cranes used is 85-96 and the total operating cost is 208 680-211 140 yuan.Compared with the optimization results of the multi-objective genetic algorithm,the computation speed is 6.07%faster,and four more non-dominated solutions are achieved with an increase rate of 30.76%,the results are closer to the Pareto frontier and the optimization degree of joint allocation plan is higher.The designed alternate evolution algorithm based on plant growth simulation maintains the maximized independence of an individual in the population evolution process and obtains more non-inferior solutions,and the alternate evolutionary approach provides results closer to the Pareto frontier.3 tabs,7 figs,26 refs.