Abstract:

A Dimensional Boxing Problem (1BKP) is one of the most studied NP-Zor combination problems on industrial engineering. Finding the best solution for problems that contain a large number of (more than a hand) pieces can take hundreds of years with classical brutal force algorithms. Therefore, intuitive algorithms are often used that can be found in short-term with (approximately)-optimal solutions or with low performance losses. With this study, a hyper-scientific parallel algorithm (HPGG-1BKP) was developed, instead of classic approaches using only one of the intuitive cutting techniques used in the GGA (Grouping Genetic Algorithms), using multiple intuitive cutting techniques at the same time. The most suitable emptiness fills (EUBD), the first emptiness you find fills (IBBD) and the smallest emptiness fills by leaving (EKBBD) intuitive box filling algorithms have been used simultaneously in this algorithm. The results of 1228 trials on the benchmark problem were achieved with 88.1% success and 1070 optimal results. For the remaining problems, only one box of more solutions were produced and the results were reduced. The recommended algorithm compared to Falkenauer GGA, up to 9% improvements were achieved.

Keywords: