文章摘要
许鹏,孙毅,张健,等.基于人工智能代理的负荷态势感知及调控方法[J].电力系统自动化,2019,43(3):178-186. DOI: 10.7500/AEPS20180329006.
XU Peng,SUN Yi,ZHANG Jian, et al.Load Situation Awareness and Control Method Based on Artificial Intelligence Agent[J].Automation of Electric Power Systems,2019,43(3):178-186. DOI: 10.7500/AEPS20180329006.
基于人工智能代理的负荷态势感知及调控方法
Load Situation Awareness and Control Method Based on Artificial Intelligence Agent
DOI:10.7500/AEPS20180329006
关键词: 负荷态势感知  人工智能  代理  随机森林  时域特征拓展
KeyWords: load situation awareness  artificial intelligence (AI)  Agent  random forest  time-domain feature expand
上网日期:2018-11-16
基金项目:国家自然科学基金资助项目(51777068);国家电网公司科技项目
作者单位E-mail
许鹏 电力节能教育部工程研究中心, 华北电力大学, 北京市 102206  
孙毅 电力节能教育部工程研究中心, 华北电力大学, 北京市 102206 sy@ncepu.edu.cn 
张健 国网鹤岗供电公司, 黑龙江省鹤岗市 154101  
吕艳霞 国网鹤岗供电公司, 黑龙江省鹤岗市 154101  
李彬 电力节能教育部工程研究中心, 华北电力大学, 北京市 102206  
祁兵 电力节能教育部工程研究中心, 华北电力大学, 北京市 102206  
摘要:
      负荷态势感知是需求侧资源管理愈加精细化发展需求下的关键技术之一。针对典型负荷动态态势估计及其调控需求,提出一种基于人工智能代理的负荷态势感知方法,并构建人工智能代理与云端平台协作的实现架构。通过用户侧负荷模型的学习认知,进而建立负荷空间集,并提出时域特征拓展的随机森林算法以实现对于负荷运行状态的评估分析,为负荷调控提供决策依据。进而根据评估结果及电力系统调控需求,在云端平台实现差量计算,并基于态势指标计算结果遴选优化组合,实现负荷精准控制。最后,通过仿真验证了所提方法能够准确预测负荷行为并达到良好的负荷整形效果。
Abstract:
      Load situation awareness is one of the key technologies for demand-side resource management under the situation of more and more precise demands. Aiming at the situation assessment and dynamic regulation for typical loads, a load situation awareness method based on artificial intelligence (AI) Agent is proposed. The architecture for collaboration of AI Agent and cloud platform is constructed. Through the learning and cognizing of the load model at the user side, a spatial set of loads is further built, and temporal feature expanded random forest algorithm is put forward to realize the assessment of load operating state, which provides the decision-making basis for load regulation. Then, according to the results of assessment and the requirements of system optimization, differences between load curve and target curve in the cloud platform are evaluated, and optimal portfolio is selected based on the calculation results of situation indicators, which achieves precise load control. Finally, the results of simulation show that the proposed method can accurately analyze the load situation and achieve better load shaping effect.
查看全文(Free!)   查看附录   查看/发表评论  下载PDF阅读器