1.中国电力科学研究院有限公司(南京),江苏省南京市 210003;2.华南理工大学电力学院,广东省广州市 510640;3.东北大学佛山研究生院,广东省佛山市 528311
在电网调度决策领域应用人工智能(AI)技术已成为研究热潮,但目前AI技术在泛化性、安全性、可解释性等方面的缺陷成为阻碍其实际应用的瓶颈。作为新一代人工智能五大技术方向之一,混合增强智能将人的认知引入人工智能系统中,与机器智能共同形成的混合智能形态,有望成为突破当前AI应用瓶颈的重要方法体系。文中重点研究混合增强智能调度的基础框架——人机协同知识演化机理和方法,为将当前“机器辅助调度”模式提升到“混合增强智能调度”模式奠定理论基础。首先,分析了当前AI技术在电网调度决策中的应用现状;其次,阐述了知识演化的内涵,并提炼科学问题;再则,探讨了知识演化的实现思路,并归纳为“一个架构、二个通道、一个推理机制”;最后,分别从知识架构、知识获取、知识解释和知识推理对支撑知识演化的4项关键技术进行了研究,提出了研究参考思路和实现方法。
国家自然科学基金资助项目(U2066212)。
姚建国(1963—),男,硕士,教授级高级工程师,主要研究方向:电力系统自动化、电网智能调度。E-mail:yaojianguo@epri.sgcc.com.cn
余涛(1974—),男,通信作者,博士,教授,博士生导师,主要研究方向:复杂电力系统的非线性控制理论、优化及机器学习。E-mail:taoyu1@scut.edu.cn
杨胜春(1973—),男,博士,研究员级高级工程师,主要研究方向:电网调度自动化、需求响应。E-mail:yangshengchun@epri.sgcc.com.cn
1.China Electric Power Research Institute (Nanjing), Nanjing 210003, China;2.School of Electric Power Engineering, South China University of Technology, Guangzhou 510640, China;3.Foshan Graduate School of Innovation, Northeastern University, Foshan 528311, China
There has been a research upsurge of applying artificial intelligence (AI) to the decision-making of power systems dispatch. However, the defects of existing AI technologies in generalization, safety and interpretability have become the bottleneck that hinders its practical application. As one of five technological directions of the new generation AI, the hybrid-augmented intelligence introduces human cognition to AI systems, which forms a hybrid intelligence pattern together with machine intelligence. It is expected to offer an important methodology to break through the application bottleneck of current AI. This paper establishes a theoretical basis for promoting the current machine-aided dispatch mode to hybrid-augmented intelligent dispatch, and investigates the fundamental framework of hybrid-augmented intelligent dispatch which is further described as a mechanism and a method of human-machine collaborative knowledge evolution. Firstly, the existing research on applying AI to the power system dispatch is analyzed. Secondly, the connotation of collaborative knowledge evolution is clarified and its scientific problems are proposed. Thirdly, the realization idea of collaborative knowledge evolution is discussed and is further summarized as “one framework, two passageways, and one reasoning mechanism”. Finally, this paper investigates the four key technologies of collaborative knowledge evolution, including knowledge framework, knowledge acquisition, knowledge explanation, and knowledge reasoning. The research reference ideas and realization solutions for the four technologies are also provided.
[1] | 姚建国,余涛,杨胜春,等.提升电网调度中人工智能可用性的混合增强智能知识演化技术[J].电力系统自动化,2022,46(20):1-12. DOI:10.7500/AEPS20220110004. YAO Jianguo, YU Tao, YANG Shengchun, et al. Knowledge Evolution Technology Based on Hybrid-augmented Intelligence for Improving Practicability of Artificial Intelligence in Power Grid Dispatch[J]. Automation of Electric Power Systems, 2022, 46(20):1-12. DOI:10.7500/AEPS20220110004. |