Development of Smart AI Battery Factory... Annual Savings of 2.2 Billion Won

[Financial News] A domestic research team has successfully applied artificial intelligence (AI) technology to the production process of lithium-ion battery cathode materials to reduce defect rates and increase yield. It is estimated that more than 2 billion won in annual production costs can be saved.
The team led by Professor Jeong Imdu of the Department of Mechanical Engineering at Ulsan National Institute of Science and Technology (UNIST), in collaboration with the senior team of Yugi Seong at Pohang Institute of Science and Technology (RIST), announced on the 27th that they have designed process conditions to reduce the defect rate of NCM precursors and developed AI technology to control them in real-time.
NCM precursor is a powdery material mixed with nickel (Ni), cobalt (Co), and manganese (Mn), which is agglomerated at high temperatures to make cathode materials for electric vehicle batteries. The higher the nickel content in the precursor particles, the greater the battery capacity, but it is prone to 'leaching', where nickel does not precipitate properly and remains in the solution or re-dissolves. Leaching leads to defects with irregular particle shapes and compositions, reducing battery life and performance.
The research team optimized process conditions to suppress such nickel leaching and developed AI-based real-time equipment anomaly detection technology. By controlling the stirring speed of the raw material solution containing metal ions, acidity (pH), and ammonia concentration, nickel is designed to be placed inside the particles, while cobalt and manganese are placed outside. When nickel is positioned inside the particles, the likelihood of leaching decreases, and structural stability increases.
Furthermore, domain adaptation AI technology significantly improved defect detection performance. Existing AI was optimized only for conditions learned in the laboratory, so performance dropped significantly even with slight changes due to equipment aging or long-term mass production. In contrast, domain adaptation AI can recognize changes in the production environment in real-time and self-correct, allowing stable quality prediction in various situations.
As a result of demonstrating this AI technology in an industrial 11.5-ton reactor, the number of defective batches was reduced to 1/15 of the previous level, and the AI-based anomaly detection accuracy reached 97.8%. It is also analyzed that this could reduce raw material and production losses by about 2.2 billion won annually.
Professor Jeong Imdu said, "Unlike small-scale experimental environments in laboratories, large-scale production sites require significant costs and efforts to manage quality and yield, but this AI technology has been applied to actual sites to induce stable high-quality production," adding, "This can be applied not only to secondary batteries but also to the entire large-scale manufacturing industry, including chemicals, machinery, and semiconductors."
This research was participated by Seojun Young and Kim Taekyung, researchers from the Department of Mechanical Engineering at UNIST, as the first authors, and the research results were published on May 8 in 'InfoMat', a world-renowned journal in the field of materials (Impact Factor: 22.7, within the top 3% of JCR).

jiany@fnnews.com Reporter Yeon Jian