半月刊

ISSN 1000-1026

CN 32-1180/TP

+高级检索 English
基于大语言模型的电力系统通用人工智能展望:理论与应用
作者:
作者单位:

1.香港中文大学(深圳)理工学院,广东省深圳市 518172;2.深圳市人工智能与机器人研究院,广东省深圳市 518038;3.浙江大学电气工程学院,浙江省杭州市 310027;4.广东电网电力调度控制中心,广东省广州市 510030;5.南洋理工大学电气与电子工程学院,新加坡 639798,新加坡;6.南瑞集团有限公司(国网电力科学研究院有限公司),江苏省南京市 211106

摘要:

大语言模型(LLM)是一种利用大规模文本语料库进行预训练和微调的深度学习语言模型。目前,在通识问答、文本生成和科学推理等方面已展现出强大的能力。在此背景下,文中探索了基于LLM构建面向电力系统的通用人工智能技术,并展望其在电力系统中的潜在应用。首先,介绍了LLM的基本原理、神经网络架构以及训练方法,特别是与传统人工智能模型相比,LLM在逻辑推理、编程和代码理解以及数学推理方面的突破。然后,展望了LLM在电力系统负荷与新能源发电出力预测、电力系统规划、电力系统运行、电力系统故障诊断与系统恢复、电力市场等领域的潜在应用。最后,阐述了基于LLM构建电力系统通用人工智能技术所面临的挑战,包括电力系统数据的质量与可获取性、输出结果可解释性以及隐私保护问题。

关键词:

基金项目:

国家自然科学基金资助项目(72331009);深圳市科创委科技资助项目(ZDSYS20220606100601002);深圳市人工智能与机器人研究院资助项目。

通信作者:

作者简介:

赵俊华(1980—),男,副教授,博士生导师,主要研究方向:电力系统分析与计算、智能电网、电力市场、低碳转型、人工智能。E-mail:zhaojunhua@cuhk.edu.cn
文福拴(1965—),男,通信作者,教授,博士生导师,主要研究方向:电力系统故障诊断与系统恢复、电力经济与电力市场、智能电网与电动汽车。E-mail:fushuan.wen@gmail.com
黄建伟(1978—),男,教授,博士生导师,主要研究方向:群体智能、网络优化和经济学。E-mail:jianweihuang@cuhk.edu.cn


Prospect of Artificial General Intelligence for Power Systems Based on Large Language Model: Theory and Applications
Author:
Affiliation:

1.School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen 518172, China;2.Shenzhen Institute of Artificial Intelligence and Robotics for Society, Shenzhen 518038, China;3.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;4.Guangdong Power Grid Dispatching and Control Center, Guangzhou 510030, China;5.School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798, Singapore;6.NARI Group Corporation (State Grid Electric Power Research Institute), Nanjing 211106, China

Abstract:

The large language model (LLM) is a deep learning language model that utilizes large-scale text corpora for pre-training and fine-tuning. Nowadays, it has demonstrated powerful capabilities in generalized quizzing, text generation and scientific reasoning. In this context, this paper explores the construction of artificial general intelligence techniques for power systems based on LLM and prospects its potential applications in power systems. Firstly, the basic principles, neural network architecture, and training methods of LLM are introduced, with a particular focus on its breakthroughs in logical reasoning, programming and code understanding, and mathematical reasoning compared with traditional artificial intelligence models. Then, this paper prospects the potential applications of LLM in the areas of load forecasting and renewable energy generation prediction in power systems, power system planning, power system operation, fault diagnosis and system restoration in power systems, and electricity markets. Finally, the challenges in building an artificial general intelligence technology for the power system based on LLM are elaborated upon, including data quality and accessibility in the power system domain, interpretability of output results, and privacy protection concerns.

Keywords:

Foundation:
This work is supported by National Natural Science Foundation of China (No. 72331009), Shenzhen Municipal Science and Technology Innovation Committee (No. ZDSYS20220606100601002), and Shenzhen Institute of Artificial Intelligence and Robotics for Society (AIRS).
引用本文
[1]赵俊华,文福拴,黄建伟,等.基于大语言模型的电力系统通用人工智能展望:理论与应用[J].电力系统自动化,2024,48(6):13-28. DOI:10.7500/AEPS20230804001.
ZHAO Junhua, WEN Fushuan, HUANG Jianwei, et al. Prospect of Artificial General Intelligence for Power Systems Based on Large Language Model: Theory and Applications[J]. Automation of Electric Power Systems, 2024, 48(6):13-28. DOI:10.7500/AEPS20230804001.
复制
支撑数据及附录
分享
历史
  • 收稿日期:2023-08-04
  • 最后修改日期:2023-08-29
  • 录用日期:2023-10-23
  • 在线发布日期: 2024-03-15
  • 出版日期: