At the 2025 China Huadian New Power System Technology Innovation Forum held on November 4, China Huadian officially launched its latest developed "Huadian Zhi" large model. This model has shown outstanding performance in artificial intelligence and forecasting applications, marking another important milestone in China's intelligent transformation of the power industry.

The "Huadian Zhi" large model pioneered a runoff prediction model globally, significantly improving the accuracy of hydropower utilization prediction. This technology has increased the hydropower utilization rate in the Wujiang River Basin from an average of 5.8% over the past decade to 10.8%. The model has also been included in the first batch of strategic high-value scenarios for artificial intelligence by the State-owned Assets Supervision and Administration Commission of the State Council, highlighting its important position in the national energy strategy.

Large Model Metaverse (1)

Image source note: The image is AI-generated, and the image licensing service provider is Midjourney

Additionally, in terms of adaptive collaborative operation, the "Huadian Zhi" large model has achieved seamless start-up and shutdown control for 100,000 kW units and wide-load automatic cruising technology, which is leading in the industry. The development of these technologies not only improves the operational efficiency of power generation equipment, with a load change rate of 2.2%, but also reduces the operation volume by more than 80%. Such innovations have significant implications for reducing operating costs and improving overall efficiency.

In terms of maintenance and inspection, the "Huadian Zhi" large model has also performed well. The model integrates AI and professional analytical decision-making capabilities, building the first technical supervision model in the power sector that covers all specialties and multiple scenarios. Currently, the accuracy of hidden danger classification has exceeded 94%, and the availability of maintenance scheduling is higher than 90%, greatly enhancing the reliability of equipment.

In the bidding and procurement process, this model has also demonstrated strong intelligent support capabilities, achieving intelligent evaluation of complex categories such as wind turbine units and chemical products, shortening the evaluation cycle to the "hour" level, with an accuracy rate of over 90%, greatly improving work efficiency.