Name: DAYAN DE CASTRO BISSOLI
Publication date: 03/08/2018
Advisor:
Name |
Role |
|---|---|
| ANDRÉ RENATO SALES AMARAL | Advisor * |
Examining board:
| Name |
Role |
|---|---|
| ANDRÉ RENATO SALES AMARAL | Advisor * |
| LUCIANO LESSA LORENZONI | External Examiner * |
Summary: This study addresses the resolution of three different problems, widely encountered in
the real context of production planning and control. Initially, a GRASP metaheuristic is
proposed to solve an assembly-line balancing problem (SALBP-2). The proposed method
presented competitive results in relation to the literature, also focusing on a simplicity of
operation to be applied in real cases. Subsequently, the same method was used to solve the
Job Shop Scheduling Problem (JSP). The GRASP developed for the JSP also presented
good results, with low average relative deviation in relation to the best solutions known
in the literature. Next, we approached an extension of the JSP, the Flexible Job Shop
Scheduling Problem (FJSP). The JSP is limited to the sequencing of operations on fixed
machines, WHEREas in the FJSP the assignment of an operation is not preset and can
thus be processed on a set of alternative machines. Therefore, the FJSP is not restricted
to sequencing, extending in the assignment of operations to the appropriate machines
(routing). The FJSP is more complex than the JSP because it considers the determination
of the assignment of the machine for each operation. In order to solve the FJSP, we
proposed four meta-heuristics: GRASP, Simulated Annealing (SA), Iterated Local Search
(ILS) and Clustering Search (CS). SA presented lower results, however, incorporating it
into a hybrid version of ILS, which uses it as a local search, the results improved, especially
in more complex instances. Considering the hybrid characteristic of CS, the SA was also
used, in this case as a solution-generating metaheuristic. This approach also presented
superior results to SA. Both ILS and CS generated results with values equal to or close to
those of the best known solutions for an extensive set of instances for the FJSP, as well as
providing some new best known values.
