| dc.description.abstract | Just-in-Time Job Shop Scheduling (JIT JSS) is one of the job shop scheduling
problems, in which every cost of operations is taken into consideration. Distinguishing
this kind of problem of scheduling from others (makespan optimization, tardiness
optimization) is that the cost occurs whenever there is a different between due date and
completion time of the operation (earliness and tardiness penalty), which makes it
much more complicated to solve. In this study, a hybrid metaheuristic – Genetic Bees
Algorithm is developed for solving the problem. The basic Bees Algorithm (BA) has
proved its effectiveness in performing local search in recent years through other
research, but the global search is not its strength. From that idea, this study references
some models of combinations metaheuristics, then comes to the proposal model, which
hybrid some global search operators of a well-known metaheuristic – Genetic
Algorithm, to enhance the BA search. The local search operators have been studied
and chosen carefully to implement into the model depending on its suitability to the
JIT JSSP and unlike other classical ones, they try to avoid randomness as much as
possible. By solving a set of 36 benchmark instances ranging from 20 operations to
200 operations, the outcomes obtained from the are then compared to an exact method,
two recent studies, and best-known solutions. In general, the proposed model has its
strength and weaknesses since it performs well in some instances while do not in
others. However, computational result shows that the model can find 23, 18, 11 and 9
best solutions among 36 instances compared to exact method and two recent study
models and best-known solutions respectively, which means it outperforms the exact
method. Lastly, the last comparison shows that it achieves five new best solutions.
This study hopes that the model can help save operational cost of companies. | en_US |
