The work aims to study the algorithm of unit load device to solve the problems that in the aviation industry of China, the assembly and arrangement of the cargo on the air container is completed manually, and there is no software system to realize automatic calculation. The elite genetic algorithm which made full use of the elite selection strategy was combined with the simple genetic algorithm of roulette to study the optimal packing of multi unit load device. Historical data of 20 d of an international airline were selected for experiment. The result showed that availability of space in the international route was increased by 5%. The packing model was built and the algorithm of multi unit load device was optimized. It improves the use ratio of air freight hold capacity, effectively improves the packing and achieves optimal packing.