基于"机理+大模型"的智能体水电站协同调度决策方法OA
Research on agent-assisted decision-making system based on mechanism models and large language models
随着电力市场不断发展,面对现货交易与水情调度的强约束,传统人工调度已难以满足精细化运行需求,而随着大语言模型(大模型)的飞速发展,其在逻辑推理上具有显著优势.提出基于"机理+大模型"的智能体水电站协同调度决策方法,用于水位预测、闸门调度、现货市场策略寻优.通过机理模型负责水位、闸门计算,大模型为决策核心,消除了大模型"幻觉",确保了决策安全.该系统在保障大坝安全前提下,显著提升了运行效率和经济效益.
With the continuous development of the electricity market,traditional manual scheduling can no lon-ger meet the requirements of refined operation under the strong constraints of spot trading and hydrological scheduling.Meanwhile,the rapid development of large language models(LLM)brings significant advantages in logical reasoning.This paper proposes an agent-assisted decision-making system based on a large language model and a mechanism model for water level prediction,gate scheduling,and spot market strategy optimiza-tion.The mechanism model handles water level and gate calculations,while the large language model serves as the decision-making core,eliminating the hallucination of large models and ensuring decision safety.Under the premise of ensuring dam safety,the system significantly improves operational efficiency and economic bene-fits.
石洪宇;李阳;王小龙;姜虹志;邓博;赖星
四川华能泸定水电有限公司,四川 泸定 626100四川华能泸定水电有限公司,四川 泸定 626100四川华能泸定水电有限公司,四川 泸定 626100四川华能泸定水电有限公司,四川 泸定 626100四川华能泸定水电有限公司,四川 泸定 626100四川华能泸定水电有限公司,四川 泸定 626100
建筑与水利
电力市场大语言模型智能体辅助决策大坝安全
electricity marketlarge language modelagentassisted decision-makingdam safety
《四川水力发电》 2026 (4)
92-96,5
评论