The formation of a hat-profile is significantly influenced by springback and the final cross-sectional geometry, both of which are sensitive to die profile design. This study introduces a scalar-based artificial neural network (ANN) surrogate model combined with genetic-algorithm (GA) optimization to enhance die and process design efficiency. An automated ABAQUS finite-element workflow was established to generate 900 design cases. For each case, seven scalar geometric and angle responses characterizing the post-forming cross section were extracted and used to train a multilayer perceptron. This network maps four die design variables to the final geometry. The surrogate model demonstrated high predictive accuracy, with geometric and angular errors remaining small and coefficients of determination (R²) nearing 1.0. This enabled quick evaluation of new designs without the need for additional finiteelement analyses. By integrating the ANN surrogate within a GA, optimal die geometries were identified that reduce springback while meeting target dimensions, showcasing the proposed framework as an effective AI-driven design tool for sheet-metal forming.

This study developed a defect-detecting system for automotive wheel nuts. We proposed an image processing method using OpenCV for efficient defect-detection of automotive wheel nuts. Image processing method focused on noise removal, ratio adjustment, binarization, polar coordinate system formation, and orthogonal coordinate system conversion. Through data collection, preprocessing, object detection model training, and testing, we established a system capable of accurately classifying defects and tracking their positions. There are four defect types. Types 1 and 2 defects are defects of products where the product is completely broken circumferentially. Types 3 and 4 defects are defects are small circumferential dents and scratches in the product. We utilized Faster R-CNN and YOLOv8 models to detect defect types. By employing effective preprocessing and post-processing steps, we enhanced the accuracy. In the case of Fast RCNN, AP values were 0.92, 0.93, 0.76, and 0.49 for types 1, 2, 3, and 4 defects, respectively. The mAP was 0.77. In the case of YOLOv8, AP values were 0.78, 0.96, 0.8, and 0.51 for types for types 1, 2, 3, and 4 defects, respectively. The mAP was 0.76. These results could contribute to defect detection and quality improvement in the automotive manufacturing sector.