At 河北永洋特钢集团有限公司, steel workers now have an artificial intelligence assistant on the job. During converter operations, real-time data on flames, sounds, and exhaust gases is collected and analyzed by models to assist in determining the optimal steelmaking endpoint. This transformation is not an isolated case.
In Handan, AI is venturing beyond laboratories into production workshops, generating practical applications across industries like steel and chemicals. The 2026 government work report calls for implementing an "AI+" initiative over the next five years, cultivating more than 200 application scenarios and promoting deep integration between AI and industrial innovation.
Leveraging strengths in industry, data, and application scenarios, the Handan Municipal Data Bureau and the Municipal Bureau of Industry and Information Technology are jointly advancing the "AI+" initiative. Their focus is on deploying vertical large language models in sectors such as steel and chemicals, bridging the gap between technical supply and industrial demand, while strictly enforcing compliance filing for generative AI services to support the real economy. To date, the city has recognized 214 provincial-level vertical models and intelligent agents; in key areas including steel, chemicals, and government services, 164 vertical scenario models and 50 intelligent agents have been deployed; and eight large models have completed national cyberspace filing, ranking first in the province.
A number of effective application scenarios are enabling AI to integrate into every critical part of Handan's industrial development. At 河北永洋特钢集团有限公司, AI is now deeply embedded in core steelmaking processes. Previously, several critical judgments in converter operations relied heavily on the on-site experience of operators. Now, a converter steelmaking large-scale model utilizes multimodal sensing—flame vision, audio, and exhaust gas—to assess production conditions in real-time, achieving one-click intelligent steelmaking throughout the entire process. The endpoint hit rate exceeds 96%, and the production cost per ton of steel has been reduced by 5 to 15 yuan. At the intelligent control center of 河北永洋特钢集团有限公司, staff monitor real-time converter production data on a large screen.
"We used to rely more on experience, but now we depend increasingly on data and models," said Xi Caihong, head of the information section at Yongyang Special Steel. On June 3 this year, Yongyang Special Steel's "converter steelmaking large model" and "energy-carbon management large model" passed national cyberspace filing. The former focuses on smelting production; the latter on energy dispatch—intelligently scheduling by predicting gas generation and consumption. Currently, gas recovery reaches 140 cubic meters per ton of steel, with an annual increase of 17 million kWh in generated electricity.
The value of AI extends beyond core smelting. In energy management, the steel industry extreme-efficiency large model developed by 金谷物联科技（河北）有限公司 targets the circulating water systems of steel enterprises. Through applications like intelligent water quality regulation and equipment energy optimization, it resolves issues of chemical waste and high equipment energy consumption. After deployment at the oxygen plant of one steel enterprise, energy savings reached 30% for high-pressure circulating water, 2%, 4%, and 40% for air compressors, oxygen compressors, and cooling towers respectively, saving 4.19 million kWh annually in a single plant, translating to 2.51 million yuan in annual electricity cost reductions. AI is shifting from "supporting decision-making" to directly participating in critical production and energy processes.
Beyond steel, AI is making inroads in the chemical industry with predictive equipment maintenance. At 邯郸市裕泰化工集团有限公司, an industrial equipment predictive maintenance large model acts as a "smart doctor" for production equipment. Developed jointly by 邯郸泓联智宇科技有限公司 and Yutai Chemical, it continuously analyzes critical equipment operational data to detect anomalies and provide fault warnings. Currently, the model covers 107 pieces of key equipment, reducing abnormal production line failure rates by 80%, non-planned downtime by 40%, and saving over 5 million yuan annually in equipment faults, downtime, and maintenance costs. In front of Yutai Chemical's 5G+ Industrial Internet intelligent management and control platform, staff rely on AI large models to analyze equipment operational data in real-time, proactively warning of potential faults.
The collaboration between the two parties goes beyond equipment maintenance. This June, Yutai Chemical's SCR denitrification precise ammonia injection renovation project passed completion acceptance. Through intelligent analysis and precise control, the project ensures stable and compliant nitrogen oxide emissions while reducing ammonia consumption by 10% to 20%. AI application scenarios are expanding from equipment maintenance to environmental governance.
Enterprises provide real-world application scenarios, tech companies deliver solutions, and actual production validates model effectiveness. This industry-application synergy model is being replicated and promoted across numerous industrial enterprises in Handan. At 磁县鑫盛煤化工有限公司, AI is further advancing into the field of safety production. The company has deployed a safety production control large model that integrates personnel positioning, video recognition, electronic fencing, and intelligent inspection functions. It provides real-time location tracking for all 851 registered workers; via AI video recognition, "three violations" behaviors—smoking, not wearing helmets, and unauthorized phone use—have dropped by about 85%. Intelligent inspection efficiency has improved by 50%, on-site staffing has been optimized by about 20%, equipment failure rates have fallen by 30%, and safety management costs have been reduced by roughly 35%. Replacing "man-to-man, man-to-equipment" oversight, AI is revolutionizing traditional safety management methods.
In environmental governance, the intelligent dust removal large model developed by 河北金翼科技有限公司 has been implemented in sintering and machine tail dust removal projects at multiple steel companies in Wu'an. Cumulative electricity savings across running projects exceed 3 million kWh annually; the benchmark project achieves an average electricity saving of 22.3% for dust removal high-voltage power supplies, over 1.2 million kWh per plant per year, and annual electricity cost savings exceeding 700,000 yuan. Additionally, fault prediction reduces downtime and spare parts losses. From safety to environmental protection, AI is delving into the fine details of traditional industrial governance.
The changes brought by AI extend to the daily work of frontline staff. At 天津铁厂有限公司, the "Tiantie System Query AI Assistant Large Model" integrates policies, standards, and company regulations in production, safety, and environmental fields. Frontline employees can query in natural language for quick access to process requirements. "In the past, we had to search through extensive process documentation; now we can directly ask the large model. It's much more convenient for frontline staff to reference and learn," said one employee at Tianjin Iron Works. At 新兴铸管股份有限公司, the "New Worker Safety Training Large Model" enhances employee safety education. Trained on 60 million words of industry data and 20,000 company safety regulations, the model achieves over 95% accuracy in violation recognition. AI is also integrating into office and learning environments, changing how employees access knowledge and conduct training.
Each of these deployed applications represents Handan's practical efforts to bridge technical supply with industrial demand. Building on its foundational steel and chemical industries, the Municipal Data Bureau and the Municipal Bureau of Industry and Information Technology continue advancing the "AI+" initiative, guiding enterprises to identify application scenarios from production pain points while local AI companies tackle technological challenges closely aligned with industrial needs. Many projects have evolved from initial pilots into practical tools, gradually establishing a distinctive development path where "technology finds scenarios" replaces "enterprises find technology," and moving from single-point pilots to multi-scenario expansion. Ma Ling, deputy director of the Handan Municipal Data Bureau, stated that the next steps involve maintaining high-level promotion, matching supply with demand, and demonstrating best practices. This will facilitate the deployment of more vertical large models and intelligent agents, promote the replication of mature applications, and empower the digital transformation of local industries.
AI is changing not just production models. As AI models move deeper into plant and equipment zones, digitalization of local industries is moving away from merely pursuing technological implementation, placing greater emphasis on actual, tangible results. A range of practical and effective AI application scenarios in Handan are injecting strong momentum into the transformation and upgrading of traditional industries like steel and chemicals.
Comments