Modern sezgisel yöntemler ile tek makine çizelgelemesi

IV ABSTRACT Scheduling is an important problem in manufacturing environment. Many techniques have been generated for these kinds of problems, and in recent years, modern heuristic techniques are the mostly used of them. In this study, scheduling problems are examined in terms of modern heuristic techniques, and a genetic algorithm (GA) was used for developing a model. This model is a group-based scheduling problem on single machine - rotational molding process of plastic parts. Since the products have different shapes and volumes, being processed on the same machine, the major loss of the system is the set up times. Furthermore, the machine utilization is the other critical measure, because of differences in dimensions that have an impact on the number of parts produced in the same group. The scheduling algorithm developed especially focuses on these two critical objectives; therefore the problem can be defined as a multi-objective scheduling. The most important part of a genetic algorithm is the objective function, which is called "evaluation function" in GA terminology; the other properties of GAs are similar to each other. In literature, most evaluation functions are based upon only one objective and include relatively less information. In this study, discriminant analysis on normalized rank values generates the evaluation function. First of all, terms in objectives (variation of setup time within the group, total volume, process time/ setup time) are ranked according to the appropriate sequence, and to eliminate the effect of size of ranks, ranks are normalized by their own mean rank value. At the end of the analysis, sequence weights of each term of objectives are obtained, then these weights are used as coefficients of terms in the objective function. Finally, by using this function as priority, and the operators "crossover" and "mutation", the best sequences of groups are obtained and a successful schedule is developed. MS-Access and its Visual Basic Module are used for creation of the initial population and execution of the algorithm. Keywords: Scheduling, heuristic techniques, genetic algorithm.