In-house-developed AI system, which uses high-quality component X-ray images, has improved inspection accuracy from 93% to 99.9%, and identifies small bubbles and damage in parts during manufacturing and assembly, thereby detecting defects that human inspectors might miss.
05 Aug 2024 | 2079 Views | By Autocar Pro News Desk
Hyundai Transys is boosting its quality competitiveness by implementing its in-house-developed AI system, ‘Transys Advanced Data Analytics’ (TADA), at its production sites. TADA essentially is a digital platform driving innovation across Hyundai Transys’ business operations
TADA encompasses developing AI systems to implement smart manufacturing technology, a big data platform ...
This is an Autocar Pro Plus article. Subscribe to continue reading.
Hyundai Transys is boosting its quality competitiveness by implementing its in-house-developed AI system, ‘Transys Advanced Data Analytics’ (TADA), at its production sites. TADA essentially is a digital platform driving innovation across Hyundai Transys’ business operations
TADA encompasses developing AI systems to implement smart manufacturing technology, a big data platform to support data-driven decision-making, and training programs to enhance employee’s digital transformation capabilities.
In particular, the TADA Smart Solution and TADA Edge Solution are self-developed programs that utilise AI deep learning technology to enhance production efficiency and defect inspection accuracy on the manufacturing floor, significantly contributing to Hyundai Transys’ quality management.
TADA Smart Solution is a vision inspection system where deep learning AI detects defects during the scanning process, identifying small bubbles and damage in parts during manufacturing and assembly. The TADA Smart Solution, trained with high-quality component X-ray images, has improved inspection accuracy from 93% to 99.9%, even detecting defects that human inspectors might miss.
As of July 2024, Hyundai Transys has implemented this in seven factories in Seosan, including Jigok and Seongyeon, as well as in 54 processes at the Georgia Powertrain Plant in the United States.
TADA Edge Solution deployed at four plants
TADA Edge Solution is a small AI inspection system developed to easily create and apply customized deep learning models on-site, even by nonexperts.
Typically, at the production site, production process images such as the mounting status and quantity check of the parts to inspect are collected through a small camera and then work proceeds with the labelling work. The TADA Edge Solution will conduct defect inspection after learning. It is used to check the number of bolts assembled, mounting status, and joint lifting inspection at four Hyundai Transys Seosan Plant locations.
The TADA Edge Solution can be installed at one-tenth the cost of traditional AI deep learning inspection equipment (vision systems). In addition, it does not require network construction work or specialized maintenance technicians, allowing small and medium-sized partners to enhance production efficiency and reduce defect rates effortlessly. It is currently installed and operational at the partner company responsible for manufacturing sheet parts.
TADA AI tech pushes the worker safety envelope
Hyundai Transys has installed AI-based responsive CCTV in four areas of the Jigok Plant, where logistics vehicles and forklifts frequently pass through. These cameras sound an alarm to prevent accidents by alerting workers when vehicles approach. From an environmental perspective, around 30 responsive CCTVs detect equipment damage and immediately send alerts and messages to facility managers in case of oil or wastewater leaks, enabling swift action.
What’s more, in the R&D field, TADA AI has been utilised to improve work efficiency. The ‘gears’ used in transmissions and reducers are crucial components that affect a vehicle‘s safety, efficiency, and quietness. The Electrification R&D Division at Hyundai Transys has incorporated AI technology into the durability testing process for powertrain gear development.
A deep learning AI measurement machine can now analyze the damage range and complete the report within an hour. This task previously took researchers up to 24 hours to determine the damaged areas in gear durability tests. Furthermore, the collected test data can assist in root cause analysis and the design process.