Enhancing optimization and reducing machining time of freeform shapes through modeling, simulation, and Taguchi design of experiments with artificial neural networks

Freeform machining is one of the trickier machining operations characterized by prolonged processing durations, which lead to high energy consumption. Addressing this issue is imperative for enhancing industrial efficacy and minimizing energy consumption. This article presents modeling, simulation, optimizing, and an algorithm for reducing the long machining time of freeform geometric shapes using the Taguchi optimization technique and an artificial neural network (ANN). Firstly, the models (i.e., stock and impeller) were designed in Solidworks and imported into simulation software (SolidCAM) for machining simulation. The Taguchi L9 (3^3) was used to arrange the data using spindle speed, feed rate, and cutter diameter as the cutting conditions, and the simulation was conducted using the corresponding settings, and the machining times were recoded. Then the optimized settings were obtained, and other statistical analyses were conducted and discussed. Also, the data generated through the application of the Taguchi method was utilized to develop an advanced ANN algorithm. In addition, a second-order regression model was constructed using the Bayesian approach. This algorithm achieved a perfect 100% prediction accuracy for both the training and testing phases. These methods resulted in reducing the machining time for roughing operations from 69.68 to 33.07 min, and the ANN and other statistical results shown and discussed in this article show the effectiveness of the methods.