Particle Classification in Automotive Transmission Manufacturing by Artificial Neural Networks

Particle Classification in Automotive Transmission Manufacturing by Artificial Neural Networks

Minh-Thuan Tran, Van-Tien Truong, Trong-Dat Huynh, Van-Tron Tran, Huy-Tuan Pham, Quoc-Bao Phan

Minh-Thuan Tran, Trong-Dat Huynh, Van-Tron Tran, Huy-Tuan Pham,

Faculty of Mechanical Engineering, Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam

Van-Tien Truong,

Faculty of Engineering and Technology, Nguyen Tat Thanh University, Ho Chi Minh City, Vietnam

Corresponding author:

Quoc-Bao Phan,

Advanced Manufacturing Lab, Artificial Intelligence And Digital Transformation Institute, Binh Duong University, Ho Chi Minh City, Vietnam

Email: pqbao@bdu.edu.vn

Tel: +84903060459

Abstract

Residual particles generated during automotive transmission manufacturing can compromise system reliability, functional stability, and driving safety if they are not effectively removed before assembly. Those particles – burr, cast, chip, debris - can be generated from casting, CNC milling, CNC turning, high pressure water jet deburring, brushing deburring or unknown sources. Identifying the production origin of these particles is therefore essential for cleanability control and corrective action in manufacturing. This study proposes a morphology-based particle-classification framework using an artificial neural network (ANN) and ten extracted descriptors, including geometric features, fractal dimension, and curvature-related parameters. A two-stage classification strategy was developed to evaluate performance under different levels of classification difficulty. In Stage A, particles in the 300–1000 µm range were classified into seven representative classes, while five difficult subclasses were grouped into an “Unknown” category. Under this practical screening setting, the ANN achieved an overall accuracy of 91.8%. In Stage B, the original 12 classes were restored for particles in the 50–300 µm range, and the overall accuracy decreased to 65.3% because of increased morphological overlap among fine-grained subclasses. The results show that ANN is effective for representative particle screening in automotive transmission manufacturing, but ANN alone is not sufficient for highly reliable recognition of small and morphologically similar particles. These findings provide a practical foundation for future integration of richer descriptors and complementary algorithms to support more intelligent cleanability analysis.

Key Words: Particles × Burr × Classification × ANN × Cleanability.

Funding This work was financially funded by Advanced Manufacturing R&D Project at Vietnam Metal Hardware Co., Ltd, grant number: 20250101.

Trong công trình đầu tiên mang tên Particle classification in automotive transmission manufacturing by artificial neural networks, nhóm tác giả mang đến một góc nhìn thực tiễn vô cùng sâu sắc. Bằng cách sử dụng mạng nơ-ron nhân tạo ANN phân loại hai giai đoạn, mô hình đạt độ chính xác 91,8% ở các hạt kích thước lớn nhưng lại phản ánh đúng thực tế giới hạn 65,3% ở các hạt nhỏ do sự chồng chéo hình thái. Sự minh bạch này cung cấp nền tảng vững chắc để tiếp tục tối ưu hóa các thuật toán trong tương lai. Bạn đọc quan tâm có thể đọc chi tiết tại liên kết: https://link.springer.com/article/10.1007/s00170-026-18355-4 🎯

Dưới sự dẫn dắt của TS. Phan Quốc Bảo cùng các cộng sự, nhóm nghiên cứu đã xuất sắc công bố liên tiếp hai bài báo trên tạp chí danh giá The International Journal of Advanced Manufacturing Technology thuộc danh mục Springer Q1 với chỉ số ảnh hưởng IF 3.5 🏆

Tham khảo nguồn: https://fee.bdu.edu.vn/index.php/nghien-cuu-khoa-hoc/double-q1-hanh-trinh-ai-vuot-gioi-han-tai-vien-aidti-737.html