On March 31, the “Artificial Intelligence Innovation Application in Guangxi’s Traditional Manufacturing Industry” event, along with the Launch Ceremony for Liuzhou Steel’s Large Model, took place in Nanning. During this significant event, Liuzhou Steel Group, in partnership with Huawei and China Mobile Guangxi, jointly introduced “Xuantie”—the first large model tailored for Guangxi’s steel industry. Utilizing Huawei’s Pangu large model as its pre-training foundation, Xuantie is designed to excel in three core domains: “AI + Human, AI + Data, AI + Manufacturing.” It establishes a “20+N” scenario-based model system that spans six critical stages of steel production: “pre-ironmaking, steelmaking, steel rolling, logistics, environmental protection, and safety.”

Currently, the Xuantie large model has been successfully deployed across multiple core production scenarios, delivering outstanding outcomes. In the realm of intelligent ore blending and raw material optimization, the system has dramatically shortened the traditional 72-hour ore blending decision-making process to a mere 16 minutes. The intelligent hot metal transportation system, powered by 5G technology, has achieved complete automation in the hot metal transportation process. The equipment predictive maintenance platform is capable of providing 72-hour advance warnings for potential equipment failures. The complex scenario machine vision set has amassed over 60 rapidly reusable visual agents, which have been extensively deployed in more than 100 production scenarios, achieving a recognition accuracy rate surpassing 99%.

Furthermore, by implementing the intelligent steelmaking scheduling model, Liuzhou Steel Group has enhanced coordination across various production processes, resulting in an 8.5% boost in production efficiency. Simultaneously, it has improved product quality, reduced raw material consumption, and lowered the production cost of crude steel by 5 yuan per ton.

Looking forward, Liuzhou Steel Group will prioritize the comprehensive implementation of five key scenarios, including the full-process intelligent steelmaking system, 5G + hot rolling smart factory, and full-process quality intelligent management and control platform. The company aims to fully deploy mature scenarios within the next three years, with the goal of achieving over 80% coverage in full-process intelligence.