本报讯 （记者袁传玺）10月19日，在"人工智能与未来能源系统"为主题的远景闭门科技会上，远景科 技集团（以下简称"远景"）董事长张雷发表演讲，系统阐述人工智能作为划时代生产力的革命性影响， 并清晰定义了"物理人工智能"这一前沿概念。他指出，AI不是"工具"而是"主体"，未来能源系统绝非设 备的简单堆叠，而将进化成"智能体"生态系统。能源行业的竞争核心将从传统的"物质资产"转向未来 的"人工智能资产"。 具备全局感知、系统洞察和持续进化能力的"天枢"能源大模型，成功应用于构建新型电力系统，提升储 能和风机收益，优化电力交易和资产投资决策，扩大绿电消纳。该能源大模型基于海量的天气、设备、 电网和市场数据，能够运用图神经网络、时空模型和多模态Transformer等先进算法，并通过云、边、端 协同实现实时控制。 张雷认为，未来能源企业的核心竞争力，将从装机容量和资产规模，转向"人工智能资产"的规模。他呼 吁行业关注："你的大模型智商有多高？智能体数量有多少？算力有多大？"这场从"物理资产"到"智能 资产"的深刻转变，正在重塑全球能源格局。 图说：远景科技集团董事长张雷 远景"物理人工智能与未来能源系统"闭门科技会 ...

Core Viewpoint - Envision Energy has launched the Galileo AI Wind Turbine, which aims to address major pain points in the wind power industry by providing more flexible and precise power generation strategies and higher reliability, marking a new phase in the application of physical artificial intelligence in the sector [1][3]. Summary by Sections Addressing Industry Pain Points - The Galileo AI Wind Turbine offers a validated solution to three major pain points in the wind power industry: inaccurate forecasting (power/load/consumption/electricity price), poor turbine performance, and high safety and quality risks. The implementation of the "Tianshu" energy model intelligent control platform has led to over a 20% increase in revenue for wind farms equipped with AI compared to those without [3][4]. Enhancing Forecast Accuracy - The "Tianji" meteorological model utilizes advanced computing power and a model with over 10 billion parameters to achieve significant breakthroughs. It integrates multi-modal data from satellites, radar, and ground stations, along with data from over 800 GW of global energy assets, to generate precise forecasts for the next 15-30 days within just three minutes [5][6]. Improving Power Generation Capability - The core of the Galileo AI Wind Turbine is a neural network with over 100 million parameters, functioning as a "super brain" for the turbine. This system, supported by high-performance chips, enables real-time online reasoning to handle complex, non-linear problems that traditional control logic struggles with. The intelligent control platform allows for real-time adjustments and self-healing capabilities, enhancing overall efficiency [7][8]. Increasing Warning Accuracy - The development of a high-fidelity digital twin platform is crucial for improving warning accuracy in the wind power sector. By leveraging AI computing power and extensive operational data, the integration of multi-modal information has significantly enhanced prediction accuracy. For instance, early detection of blade failures through sound and strain monitoring has improved maintenance scheduling, resulting in substantial operational gains [9][10]. Future of AI in Wind Power - The transition from traditional wind turbines to intelligent systems capable of understanding weather changes and market dynamics represents a significant evolution in the industry. The potential for further advancements in artificial intelligence applications within wind power remains vast [11].