基于改进遗传算法的航空集装箱装载问题研究

针对标准遗传算法求解装载方案时存在收敛速度慢、易早熟、寻优结果欠佳的问题，基于拟人装载策略，提出了一种以集装箱空间利用率最大为目标，考虑货物装载顺序、体积、质量、重心、不重叠等多种实际约束的改进遗传算法。首先，采用与货物放置状态相结合的实数编码，随机产生初始种群；然后，在常规选择操作中加入最优解保存策略，并将稳定性、支撑限制、重心约束考虑到进行线性尺度变换后的适应度函数中，以此来计算每种装载方案的评估值；最后，输出评估值最高的方案作为最优装载方案。实验采用异构性不同的测试算例进行性能测试，结合3组具体货物装载数据证明算法的普适性与实用性。结果表明:所提算法在求解强异构货物装载过程中具有较好的优化效果，适用于求解集装箱装载问题。与标准遗传算法相比，收敛性与搜索速度有所提高，2种不同箱型的集装箱空间利用率分别提高了3.82%和3.66%，运行时间分别缩短了7.9 s和5.58 s，能快速找到最优装载方案，可有效解决规则、不规则集装箱的货物装箱问题。基于MATLAB软件实现装载方案的可视化，为集装箱的实时装载决策提供了理论基础。 

Air container loading based on improved genetic algorithm

Aimed at the problems of slow convergence speed, premature maturity, and poor optimization results when the standard genetic algorithm solves the loading plan, based on the anthropomorphic loading strategy, an improved genetic algorithm is proposed to maximize the utilization of container space, considering the loading sequence, volume, and quality of the goods, center of gravity, non-overlapping and other practical constraints. First, the real number code combined with the placement state of the goods is used to randomly generate the initial population. Second, the optimal solution preservation strategy is added to the routine selection operation, and the stability, support constraints, and center of gravity constraints are taken into account after linear scale transformation. In the fitness function, the evaluation value of each loading scheme is calculated by this. Finally, the scheme with the highest evaluation value is output as the optimal loading scheme. In the experimental part, the performance test was performed using test cases with different heterogeneity, and then three sets of specific cargo loading data were combined to prove the universality and practicability of the algorithm. The results show that the proposed algorithm has better optimization effect in solving the process of strong heterogeneous cargo loading, and is suitable for solving the container loading problem. Compared with the standard genetic algorithm, the convergence and search speed have been improved. The space utilization of the two different container types has increased by 3.82% and 3.66%, and the running time has been shortened by 7.9 s and 5.58 s. The optimal loading can be found quickly. The solution can effectively solve the problem of cargo packing in regular and irregular containers. At the same time, the visualization of the loading plan is realized based on MATLAB software, which provides a theoretical basis for the real-time loading decision of the container.