The work aims to address the practical issues of three-dimensional packaging in logistics operations, such as material packaging, pallet stacking, assembly packaging, and container loading, so as to improve the efficiency, accuracy, and utilization of three-dimensional packaging. The traditional single chain genetic algorithm was improved to double chain, the selection, crossover, and mutation operators were modified based on the characteristics of three-dimensional packaging. Update operators were introduced to establish a packaging model between strong heterogeneous hard cubes (many) and standardized packaging boxes (many). The result showed that simultaneously encoding the packing order and placement direction and establishing associated keys to form a double stranded genetic model could avoid local too fast convergence while ensuring accuracy and reliability. It achieved global fast optimization, and guided the implementation of 3D packaging in the offline mode. In conclusion, with a strong heterogeneous hard cube as the packing object, and three-dimensional standardized packaging as the packing objective, a "many to many" three-dimensional packaging model is constructed through double chain improvement of genetic algorithm, and a set of modeling techniques and algorithm programs are proposed to solve the "many to many" three-dimensional packaging combination optimization problem, which can effectively improve the efficiency, accuracy, and utilization of three-dimensional packaging.