Using the Integration of Manufacturing Process Simulations to Improve the Accuracy of Predicting Thermal Distortion in Outer Panels during the Painting and Drying Process

Kenjirou Baba･Kazuki Shouyama (TOYOTA AUTO BODY)･Shinichi Takezoe･Ryuichi Kaminishi･Katsushi Tsunoura (TOYOTA AUTO BODY R&D)･Takeshi Chino･Masahi Arai (JSOL)･Hirotaka Fukui･Takashi Fujiwara (TOYOTA AUTO BODY)

In order to achieve vehicle body lightweighting, the adoption of thin sheets for outer panels is desirable. However, it presents multiple manufacturing challenges. Among these challenges, predicting thermal distortion in outer panels during the painting and drying processes is particularly difficult. In this paper, we will introduce our efforts to improve prediction accuracy in this field by integrating several process simulations.

Development of press feasibility evaluation method using surrogate model

Takayuki Yoshimatsu･Shigeki Kojima･Takashi Kanai･Takaaki Harada･Koji Okamura･Daisaku Yanaga･Yuji Miyazawa･Mamoru Wakasa･Kiyoshi Nonomura (Toyota Motor)

Recent advancements in automotive styling design have made it more difficult to ensure the press formability of sheet metal and shorten development time using the conventional finite element method (FEM). Machine learning is considered an effective method for front-loading studies to ensure formability at an early stage. In this study, we propose methods to expand the training dataset for surrogate models and modify the target model for prediction. We also present the results of verifying the accuracy of surrogate model predictions.

Development of an Automated optimization Method for Resin Molding by Integrating CAE and Machine Learning

Yuto Takehara･Takayuki Nukui･Ryo Kamogawa･Yoshiko Hayashi･Kenji Takamura (AsahiKasei)

We developed an automated system that integrates CAE and machine learning to explore multi-objective optimization. By setting the exploration range, the system repeatedly executes CAE, builds surrogate models, and searches for optimal conditions. This presentation provides an overview of the algorithm and a practical example focusing on the molding analysis of resin products.

Development of a System for Efficient Confirmation Tasks to Prevent Return with Contents Using Object Detection (Report No.1)

Masahiro Kagi･Toru Hirai･Yuya Sakakibara･Riyo Kobe･Haruki Sei･Yuto Mori (Toyota Motor)

In the logistics process, all stacked boxes are visually inspected to prevent returns with contents; however, there is a problem with the long time required for this task. To address this, we developed a system that combines technology for automatically counting the number of boxes in images using machine learning with a weight scale, aiming to reduce the time required for the confirmation process.

Automated Technology for Determining the Quality of Cylinder Block Cavity through digital transformation

Hiroyuki Kimoto･Yuki Okahara･Tooru Iwaki･Hirotaka Sakamoto･Kunihiro Nobuhara (Toyota Motor)

To prevent oil leaks, skilled workers visually check the condition of Cavities, which is one of the causes of oil leaks. However, the lack of standards, which is based on intuition and tips, makes it difficult to transfer skills to other workers, which is an issue in the factory. To solve this problem, we have developed an automatic judgment technology that does not rely on humans, through digital transformation.