Enhancing propeller design with freeform contours through NURBS interpolation for 2D fabrication, CAD/CAM for 3D production, optimized with Taguchi method and artificial neural network

Freeform machining of complex geometries like propellers presents significant challenges, including prolonged machining times, high energy consumption, and the risk of tool-workpiece collisions. Traditional optimization approaches often rely on trial-and-error methods or simplified models, leading to suboptimal performance. This study addresses these limitations by developing an integrated framework that combines NURBS interpolation for precise propeller design with a hybrid Taguchi-ANN optimization strategy to minimize machining time and energy usage. First, a 2D propeller curve was generated using NURBS interpolation to ensure smooth and accurate geometry for machining. The design was then extended to 3D using SolidWorks, and machining simulations were conducted in SolidCAM. To optimize the process, a Taguchi L27 (33) design evaluated three key parameters cutting feed, Z-direction feed rate, and cutter diameter with machining time as the response variable. The Taguchi analysis revealed that cutter diameter contributed 95.52 % to machining time variation, followed by cutting feed (4.42 %) and feed Z (0.06 %). The Taguchi-derived dataset was then used to train an ANN model with a Levenberg-Marquardt algorithm, achieving 100 % prediction accuracy (R = 1.0, MSE ≈ 10⁻24) across training, validation, and testing phases. Simulation validation confirmed that the optimized parameters reduced roughing time by 69.2 % (from 42.85 to 13.20 min). Also, a redesigned Y-shaped stock minimized collision risks, further enhancing process efficiency. The study shows that combining NURBS-based design with hybrid Taguchi-ANN optimization enhances freeform machining, delivering faster production and energy efficiency for industrial use. The method also scales well for optimizing complex geometries like propellers.