Robust intelligent modeling for developing an optimal predictive model of metal powder injection molded dental braces using fuzzy-logic-based multi-objective design.

ObjectiveMetal Injection Molding (MIM) is an advanced manufacturing technology suitable for producing complex, small, and miniature components such as dental braces and biomedical devices. However, product warpage, deformation, and non-uniform temperature distribution during molding often compromise dimensional accuracy, assembly precision, and aesthetic quality. This study aims to optimize the MIM process for dental braces and develop accurate predictive models for quality characteristics.MethodsA robust process design integrated with fuzzy theory was employed to determine optimal process parameters for both single-objective and multi-objective quality optimization. Predictive models based on a Back Propagation Neural Network (BPNN) and an Adaptive Network-based Fuzzy Inference System (ANFIS) were developed. Hyperparameter structures were optimized during model development to improve prediction performance and model robustness.ResultsCompared with the manufacturer's original process parameters, single-objective optimization improved warpage and average temperature by 85.7% and 9.1%, respectively. For multi-objective optimization, the corresponding improvements were 85.7% and 7.0%. After hyperparameter optimization, the BPNN and ANFIS models achieved prediction accuracies of 96.09% and 97.11%, respectively. The ANFIS model demonstrated superior predictive capability for nonlinear process relationships while requiring less complex parameter tuning.ConclusionsThe proposed intelligent modeling framework effectively improves process quality and provides accurate prediction of key quality characteristics, thereby offering a promising approach for intelligent process optimization and potentially reducing experimental effort during MIM process development. The results demonstrate that ANFIS provides a robust and accurate prediction approach for dental-brace MIM applications and offers significant potential for intelligent process optimization in advanced manufacturing.