Application of ANOVA in interval type-2 fuzzy logic systems for modeling the process of ceramic coating preparation in the foundry industry

• Published in: International journal of advanced manufacturing technology 2024-06, Vol.132 (7-8), p.3927-3938
• Main Authors: Olvera-Romero, Gerardo Daniel, Praga-Alejo, Rolando, Rodríguez-Reyes, Mario, Mancha-Molinar, Héctor, González-González, David, Vázquez-Obregón, Dagoberto, Luna-Álvarez, Jesús Salvador, de León-Delgado, Homero, Flores-Cárdenas, José
• Format: Article
• Language: English
• Subjects: Ambient temperature; CAE) and Design; Ceramic coatings; Chemical reactions; Complexity; Computer-Aided Engineering (CAD; Engineering; Fuzzy logic; Fuzzy systems; Industrial and Production Engineering; Liquid metals; Mechanical Engineering; Media Management; Original Article; Regression models; Sand casting; Statistical analysis; Statistical inference; Variability; Variance analysis
• ISSN: 0268-3768; 1433-3015
• DOI: 10.1007/s00170-024-13563-2

The preparation of ceramic coatings is a complex process, impacted by the variability and uncertainty inherent in its operational parameters. This coating, applied to sand cores in the iron casting production of monoblocs, serves primarily to shield them from physical and chemical reactions with molten metal that could lead to penetration. This study tackles such complexity by employing a type-2 interval fuzzy logic system (IT2 FLS), notably integrating analysis of variance (ANOVA) for statistical inference within the model. This methodology facilitates a detailed and quantitative analysis of the operational variables’ influence, enhancing both the understanding and the accuracy of the IT2 FLS model. The IT2 FLS implementation exhibited 96 % effectiveness in explaining the process variability, identifying the paint tank’s density as the most influential variable, unlike ambient temperature, which had a lesser impact. Furthermore, the IT2 FLS model not only displayed superior fit ( R 2 = 0.96 ) compared to type-1 fuzzy logic systems ( R 2 = 0.80 ) and linear regression models ( R 2 = 0.49 ), but also revealed, through cross-validation, a significant predictive capacity ( R prediction 2 = 0.86 ). These results validate the robustness of the IT2 FLS against conventional methods and highlight the novelty of integrating ANOVA to deepen the statistical analysis of the IT2 FLS model. Such an approach provides an effective tool for understanding and enhancing complex manufacturing processes, offering value to industries aiming to optimize efficiency and quality.