MODELING AND OPTIMIZATION OF AN INTERVAL TYPE 2 FUZZY LOGIC SYSTEM FOR A CERAMIC COATING PROCESS

Process control is essential in Industry 4.0, and process modeling is an effective way to achieve it. For complex processes with high variability and uncertainty, Interval Type 2 Fuzzy Logic Systems are an efficient alternative, but they lack an appropriate methodology for selecting the Footprint of Uncertainty width. This work proposes a method that uses a genetic algorithm to optimize the Footprint of Uncertainty width and evaluates various Type-Reduction methods. ANOVA and R-2 and R-prediction(2) statistics are used to verify the model, which is applied to a manufacturing process that adjusts the density of a ceramic coating. The results indicate that the optimized model (R-2 = 0.886) outperforms the non-optimized model (R-2 = 0.796), linear regression (R-2= 0.498), and backpropagation neural networks (R-2 = 0.641). Additionally, a stability analysis of the proposed model was performed using cross-validation, obtaining an R-prediction(2) = 0.758, which indicates that the genetic algorithm-based method can be a suitable option for modeling complex processes.