| Abstract [eng] |
A wide area of scheduling problem is industrial so called shop scheduling. There are three classes of shop scheduling: Job Shop, Open Shop and Flow Shop. General problem specification could be specified as follows: there is set of jobs and set of machines, which should interact with each other in some specific way. Typically these problems are hard to solve in traditional (exact) methods. Metaheuristics algorithms mostly produce only nearby-optima, but in proper time. We implemented several metaheuristics: genetic algorithms (separated by values of their parameters) and several Tabu search algorithms (separated by neighborhood of solution). Some strategies of genetic algorithms are suggested as conclusion of genetic algorithm parameter research. Eight algorithms are examined for random shop scheduling problems in terms of initial solutions and minimum, gained solutions and minimum, processing time and difference between initial and gained solutions. In the end, author concludes, that one kind of genetic parameters (crossover and mutation rates) are especially demanding in sense of algorithm convergence, diversification and intensification aspects, other (number of iterations and population size) should depend on resources, third (elitism) is good “buffers”. Finally, while with its simplest form, Tabu search seems to be less competitive in algorithm effectiveness research, its dynamic modification outperforms all proposed genetic algorithms, but both – tabu search with enlarged diversification and dynamic parameter strategy of genetic algorithm – performs quite well and requires more solid future studies to compete with other advanced metaheuristics. |
