基于Stackelberg博弈与改进深度神经网络的多源调频协调策略研究  

Research on Multi-source Frequency Regulation Strategies Based on the Stackelberg Game and Improved Deep Neural Network

作　　者：王永文 赵雪锋 李夏叶 詹巍 单怡琳 闫启明 赵泽宇 杨锡运[3] WANG Yongwen;ZHAO Xuefeng;LI Xiaye;ZHAN Wei;SHAN Yilin;YAN Qiming;ZHAO Zeyu;YANG Xiyun(State Power Investment Group Sichuan Electric Power Co.,Ltd.,Chengdu 610213,Sichuan Province,China;State Power Investment Group Southwest Energy Research Institute Co.,Ltd.,Chengdu 610218,Sichuan Province,China;School of Control and Computer Engineering,North China Electric Power University,Changping District,Beijing 102206,China)

机构地区：[1]国家电投集团四川电力有限公司,四川省成都市610213 [2]国家电投集团西南能源研究院有限公司,四川省成都市610218 [3]华北电力大学控制与计算机工程学院,北京市昌平区102206

出　　处：《全球能源互联网》2025年第1期76-86,共11页Journal of Global Energy Interconnection

基　　金：国家电投集团四川电力有限公司科技项目(XNNY-WW-KJ-2021-16)。

摘　　要：随着电网中新能源渗透率的增加,传统火电机组调频已无法满足电能质量需求。针对多源场景中传统自动发电控制系统区域控制误差较大的问题,提出一种基于Stackelberg博弈与改进深度神经网络(Stackelberg game and improved deep neural network,S-DNN)的多源调频协调策略。首先,设计一种改进多层次深度神经网络(deep neural network,DNN),由DNN层、自然梯度提升层、最小二乘支持向量机层顺序递进完成预测、评价、执行动作,输出总调频功率指令。该多层次总调频功率输出模型考虑新能源渗透率对调频系统的动态影响,充分学习历史信息与实时状态中更多的特征,提高了时序调频指令精度。然后基于Stackelberg博弈理论,考虑多源调频特征与协同作用,优化各调频源间的功率分配,提高系统二次调频的经济性。最后,通过算例分析验证了提出的多源调频协调策略的有效性。与传统调频方法相比,所提出的S-DNN多源调频协调策略可有效降低区域控制误差与频率偏差,并降低调频成本。With the increase of new energy penetration in the power grid,the traditional frequency regulation of thermal power units can no longer meet the power quality demand.Aiming at the problem of large area control error in traditional automatic generation control systems in multi-source scenario,a multi-source frequency regulation strategy based on the Stackelberg game and improved deep neural network(S-DNN)is proposed.Firstly,an improved multilevel deep neural network(DNN)is proposed,which consists of a DNN layer,natural gradient boosting layer,and least squares support vector machine layer to sequentially and progressively complete the prediction,evaluation,and execution of actions,and output the total frequency regulation power command.This multilevel total frequency regulation power output model considers the dynamic impact of new energy penetration on the frequency regulation system,fully learns more features from historical information and real-time state,and improves the accuracy of frequency regulation instructions.Then,based on Stackelberg game theory,it considers the characteristics and synergy of multi-source frequency regulation,optimizes the power allocation among frequency regulation sources,and improves the economy of the system’s secondary frequency regulation.Finally,the effectiveness of the proposed multi-source frequency regulation strategy is verified by case analysis.Compared with the traditional frequency regulation method,the proposed S-DNN multi-source frequency regulation strategy can effectively reduce the area control error and frequency deviation,and reduce the frequency regulation cost.

关 键 词：多源系统 二次调频 STACKELBERG博弈 深度神经网络 自然梯度提升 最小二乘支持向量机 

分 类 号：TM73[电气工程—电力系统及自动化]