Intelligent Modeling Method to Optimal Predictive Model for Metal Powder Injection Molding Gears Using Fuzzy-Logic-Based Multi Objective Design

This study will use intelligent modeling to optimize the metal powder injection molding gear process. The robust process combined with fuzzy theory would be used to find process parameters for optimizing process single and multiple quality characteristics. And establish a reverse transfer neural network and adaptive neuro-fuzzy system prediction model. The model hyperparameter structure will be optimized during the modeling process to improve accuracy. The industry can use it to reduce process uncertainty, cost, and extremely a process innovation. The experimental results are single-objective optimization. After optimizing the parameters, the process target displacement, volume shrinkage rate, and powder concentration, compared with the manufacturer's original process, the improvements are increased by 5.79%, 5.66%, and 1.57%. Optimizing multiple quality characteristics has risen by 5.37%, 5.66%, and 0.31%. In the prediction model building part, after the optimization experiment of hyperparameter setting, the accuracy of the backward pass neural network model can reach 96.99%; the accuracy of the adaptive neuro-fuzzy system model can reach 97.45%.