From Expertise to Intelligence: NISCO Deploys World's First Swarm-Intelligence Steel Model at Scale

Nanjing Iron & Steel Co., Ltd. (NISCO) has partnered with Huawei and ecosystem allies to commercially launch the world's first steel model utilizing swarm intelligence - the YuanYe Steel Large Model (YuanYe) - across numerous scenarios at scale. This transition enables NISCO to shift from experience-based operations to data-driven intelligence collaboration.

As a traditional manufacturer, NISCO has established a strong digital and intelligent foundation by incorporating industry-leading products like Ascend, Kunpeng, Huawei Cloud Stack, and Pangu models. AI is now integrated throughout production stages - ironmaking, steelmaking, rolling, and quality inspection - resulting in green and high-end AI-powered super steel plants.

The YuanYe model stands out by combining multiple AI models with steel industry metallurgical mechanism models into a dual-drive, multi-model cluster architecture with a right brain for probabilistic computing and a left brain for precise decision-making. With millions of OT data points aggregated from 26 business domains, NISCO achieved real-time digital twin mapping within its intelligent operations center, providing solid data and scenario foundations for YuanYe.

Now, 100% of NISCO's key equipment is connected and digitally controlled, and its new product R&D cycles have been shortened by 50%, reducing annual costs by over CNY100m (approx. US$15m). By October 2026, NISCO had launched 25 business agents covering visual quality inspection, process optimization, production scheduling, and operations management, setting a new standard for digital and intelligent steel transformation.

The YuanYe model revolutionizes traditional operations, delivering significant safety, efficiency, and quality improvements. In the ironmaking stage, AI makes blast furnaces transparent by leveraging internal sensor data and metallurgical mechanism analysis for real-time dynamic simulations, predicting temperatures and identifying issues two hours ahead with over 90% accuracy. This shifts furnace operations from reactive to proactive, increasing prime-grade molten iron yield by over 10% while cutting energy consumption.

In the rolling stage, AI predicts and codifies expert experience to enhance operational stability and quality. The rolling-force prediction model enables millisecond-level responses, achieving ±0.2 mm thickness tolerance and 1.9% thickness deviation, setting a world-leading standard. Enhanced dimensional precision and physical-chemical stability meet diverse steel specifications for applications ranging from mega stadiums to LNG storage tanks.

Quality inspection benefits from the synergy between robots and AI models, doubling metallographic inspection efficiency. The integrated smart metallographic inspection system automates sample grinding, polishing, image acquisition, and grading, improving accuracy to over 95% and reducing inspection time from 10 minutes to 3-5 minutes, freeing operators for anomaly analysis and high-level decision-making.

In energy and environmental control, NISCO developed the S-Link Intelligent Agent for Gas and Flue Gas Sulfur Control using the YuanYe architecture. Combining mechanism inversion, time-series forecasting, and game-theoretic models, the agent issues sulfur-risk warnings one hour in advance with over 95% accuracy, reducing downtime caused by emission exceedance by over 25% and delivering annual economic benefits exceeding CNY600,000 (approx. US$90,000).