South KoreaManufacturing2025 – ongoingHwashin
Predictive maintenance on the battery-case line
An auto-parts maker added predictive maintenance to its EV battery-case line and reported a 20% reduction in the product defect rate
Problem
Hwashin, an automotive parts maker, produces battery cases for electric vehicles. Equipment condition on the line affects the quality of the cases it turns out.
Approach
Through the Ministry of Trade, Industry and Energy's AI Factory leading project, the company introduced a predictive maintenance system for the equipment on its battery-case line. The ministry included the result in its August 2026 review of 42 early sites out of about 170 it supported.
Results
Takeaway
Hwashin reported fewer defects after introducing predictive maintenance on its battery-case equipment. The ministry did not disclose how the reduction was measured.
Categories
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Government support program
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
The ministry reviewed 42 early sites in its AI Factory leading project and reported average productivity up 30.1% and defect rates down 15.5%. Hwashin reported a 20% fall in its product defect rate after introducing predictive maintenance on its battery-case line; the ministry did not disclose the measurement method.
Korea Data Economy News reports Hwashin's 20% reduction in an analysis of vision inspection and predictive maintenance. It sets the Korean results against an unnamed McKinsey example where a model predicted about a quarter of failures with 85% accuracy, but a false-alarm rate near 10% caused about 1,000 extra maintenance jobs a year and erased the savings.
Reviews AI results at smaller manufacturers and cites Hwashin's 20% lower defect rate from predictive maintenance on battery-case equipment. It notes that results depend on the stage of adoption and the process, and that staffing and keeping systems running after deployment remain difficult.
Full details are in the original sources.