| Abstract [eng] |
This work presents a tower-building algorithm for packing boxes of different sizes into one container. The essence of the procedure is to generate a set of towers, which consist of different boxes, using a genetic algorithm, and then place these towers on the floor of a container, also using a genetic algorithm. Several heuristic algorithms have been developed for the genetic algorithm to fill the bottom of a container with towers of boxes. Heuristic mutation and crossover operators have been developed. The crowding technique was applied. The heuristic algorithm was also parallelized for efficient execution on many CPU cores. Also, stability and orientation constraints are included in the work. In addition to this, a script was developed to visualize loading, which can operate in several modes. The heuristic algorithm was tested on 2 datasets and compared with algorithms from other authors. This showed very good results that were above other metaheuristic approaches. |
