The work aims to maximize the use of container loading space with the packing problem of goods on the assembly line under the background of air freight as the research object. Considering a series of realistic constraints such as cargo loading order, quality, volume, non-overlap, stability, etc., a container stacking model was established, and a combined heuristic algorithm combining anthropomorphic heuristics and genetics was designed. First, by setting rules and assigning weights, an ordered set of peaceable points was obtained, and the anthropomorphic heuristic algorithm was used to construct the cargo loading strategy to obtain the initial packing solution; then the solution was optimized through crossover and mutation operations in the genetic algorithm. Among the feasible schemes, the packing scheme with the largest container space utilization rate was selected; finally, the actual cargo data of an airport logistics company was used for the experiment, and the loading scheme was visualized. Compared with the hybrid simulated annealing algorithm, the combined heuristic algorithm had good convergence, fast search speed, and can obtain a loading solution with a high space utilization rate in a short period of time, and the cargo stack planning was more compact and more stable. The proposed algorithm has better adaptability to different kinds of cargoes, can realize the reasonable layout of strong heterogeneous cargoes, guarantee the stability of the stacking shape during cargo loading, and provide a reference for the engineering application of air cargo packing.