Container Ship Three-Dimensional Loading Problem Based On Hybrid Genetic Algorithm

With the coming of industrial 4.0, "Smart Logistics" has become a hot research topic. As a key element in the logistics process, the loading process of container ship at port has a significant impact on the efficiency of the whole shipping logistics process and directly impacts the transport costs. Researches on "Smart" algorithms about container ship loading problem have also been given new significances.Container ship loading problem belongs to the general three-dimensional container loading problem. Based on the analysis of container ship loading problem, as the goal of maximizing the space utilization, a mathematical model of general container ship loading problem was established. According to the characteristics of the mathematical model, a hybrid genetic algorithm integrating heuristic algorithm and genetic algorithm was presented. A heuristic algorithm combined dynamic space division method and a new real-number genetic encoding method consisted of cargo's sequence and rotation was designed. The corresponding adaptive crossover operator and adaptive mutation operator were constructed which were adapted to the container ship loading problem. Based on the research of general container ship loading problem, four kinds of constraints may be encountered during the loading process including weight limit constraint, orientation constraint, upper stacking constraint, and under stacking constraint were taken into account. As the goal of maximizing the space and weight utilization, a mathematical model of multi-objective and multi-constrain container ship loading problem was established. According to the constrainsm, new heuristic method for dividing space was designed. The genetic operators must be improved to satisfy the rotation constraint. The loading system software was designed with MATLAB GUI software. The loading system software can be both apply in general container ship loading problem and multi-objective and multi-constrain container ship loading problem. The loading system software can visually display the loading results, and its interface is friendly and easy to operate.Some simulations and experimental confirmations were used to verify the algorithm. Taking fifteen classic sets of test data in Loh's paper as weak heterogeneous problem data, the space utilization reached 70.63% on average. Taking three sets of test data in George and others' paper as strong heterogeneous problem data, the space utilization respectively reached 84.42%, 86.93% and 83.76%. Taking a set of test data in Lei Pu's paper as multi-objective and multi-constrain problem data, the hybrid genetic algorithm only iterated 10 times and used 2 seconds on average to achive the global optimal solution. Compared with the similar algorithm applying in the similar container loading problem, the hybrid genetic algorithm had a distinct improvement. The experimental results showed the hybrid genetic algorithm had an excellent effect both on solving weak heterogeneous container ship loading problem, strong heterogeneous container ship loading problem, multi-objective and multi-constrain container ship loading problem.