Increase judgment reliability up to 96% and reduce inspection time to 30 seconds
[Asia Economy Reporter Kim Jong-hwa] The Ministry of SMEs and Startups has begun to see results from the recently launched 'Artificial Intelligence Manufacturing Innovation Platform (KAMP) project.'
Recently, a company participating in the Ministry's KAMP project made headlines by introducing artificial intelligence for the first time in the world to inspect defects in refractory products. Chosun Refractories Co., Ltd., established in 1947, is the only comprehensive refractory manufacturer in Korea that pioneered the refractory industry domestically, localizing 100% of domestic refractory products and serving as a pillar of Korea's heavy industry development for over 70 years.
The company developed the necessary algorithms and analysis models for inspection with the help of AI and manufacturing process experts. At the KAMP service portal opening ceremony held on the 14th of last month, they demonstrated the process of analyzing actual data to inspect and judge defective products. Through the demonstration, they improved judgment reliability to 90-96% and significantly reduced inspection time from 1.5 minutes to 0.5 minutes.
Until recently, the company’s workers visually inspected defects such as cracks and foreign substances inside finished refractory products using X-ray equipment. However, due to product shape diversification and the aging of inspectors, inspection reliability declined, and additional personnel were required due to inspectors’ eye fatigue, leading to increased manufacturing costs and other issues.
After much consideration, participation in the KAMP project became the solution. The KAMP system extracts image data from X-ray videos of the final refractory products and attaches labels indicating 'normal/defective' status to each data set. Then, this data is used to reinforce the learning of the developed AI analysis model, enabling accurate judgment of whether the actual refractory image data input into the AI computation method is normal or defective.
The system improved judgment accuracy, which had been inconsistent depending on the inspector’s skill level, up to 96%, shortened inspection time to 30 seconds, and strengthened the ability to respond to new defect patterns deviating from existing types.
Lee Gwi-seon, Executive Director of Chosun Refractories Co., Ltd., said, "I hope this project will be continuously managed and become an opportunity for our company to leap forward again through KAMP." Kim Il-ho, Head of the Smart Manufacturing Innovation Planning Group at the Ministry of SMEs and Startups, emphasized, "If small and medium enterprises receive AI services at the level of large companies through the Artificial Intelligence Manufacturing Innovation Platform (KAMP), more excellent demonstration cases like Chosun Refractories Co., Ltd. will be discovered in the future."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
