Fuzzy Expert System For Constraint Handling In Real World Optimization

Optimization techniques have increasingly been applied in industrial environments over the past years. Even so, requirements and restrictions of real world problems, like technical standards and including physical impossibilities, are not so easy to include. Often, these constraints are hard coded in the computer program and the user cannot update them without the help of a programmer. In most cases, they are used as must regions in the search space. However, in real world problems the restrictions usually are not so strict. Expert systems have a knowledge base that resides outside of the computer program and relies on a set of rules that can be easily updated by the user. Fuzzy logic can make flexible the restrictions provided by specialists’ knowledge about the problem. This paper aims to propose a Fuzzy Expert System (FES) to better handle the constraints in real life engineering and business optimization problems. The proposed system was tested for optimizing an engineering problem comparing two known swarm algorithms, PSO and wFSS. The latter one is multimodal and suggest multiple feasible solutions, which is even better for supporting decisions. The results of the study show that the proposed method can achieve good and feasible solutions on the given problems and, consequently, increase competitiveness.