基于改进DBSCAN-RNN的电力负荷建模及可调特征提取  

Power Load Modeling and Adjustable Feature Extraction Based on Improved DBSCAN-RNN

作　　者：张露 颜宏文[1] 马瑞[2] ZHANG Lu;YAN Hongwen;MA Rui(School of Computer and Communication Engineering,Changsha University of Science and Technology,Changsha 410141,China;School of Electrical Engineering,Changsha University of Science and Technology,Changsha 410141,China)

机构地区：[1]长沙理工大学计算机与通信工程学院,湖南长沙410141 [2]长沙理工大学电气工程学院,湖南长沙410141

出　　处：《智慧电力》2023年第3期39-45,共7页Smart Power

基　　金：国家自然科学基金资助项目(51977012);国网电子商务有限公司科技项目(8200/2021-72001B)。

摘　　要：针对面向能源消纳的电力负荷实时调控需求,以电热水器为例建立调控模型,提出一种改进DBSCANRNN算法的电力负荷可调特征提取与可调潜力挖掘方法。以改进DBSCAN聚类结果作为RNN输入获得一种深度学习新策略,基于改进DBSCAN-RNN进行电器群设定温度与天气温度、电器负荷功率的建模,考虑用户电器使用习惯,输出输入量对电器实际功率的影响因子以及电器可调功率与真实功率对应的状态方程参数。某市电热水器群实际数据结果表明所提方法可正确有效地获取海量电热水器群聚合负荷模型及其可调功率。To meet the energy consumption oriented demand for real-time power load regulation,this paper establishes a regulation model by taking an electric water heater as an example,and proposes the method of adjustable feature extraction of power load and adjustable potential mining based on the improved DBSCAN-RNN algorithm.Firstly,the improved DBSCAN clustering results are used as RNN input to obtain a new deep learning strategy.Then the setting temperature and air temperature of electric appliance group and the load power of electric appliances are modeled based on the improved DBSCAN RNN algorithm,considering user habits using the electric appliances,the impact factors of input and output on the actual power of the electric appliances,and the state equation parameters corresponding to the adjustable power and real power of the electric appliances.The actual data from an electric water heater group in a city shows that the proposed method can correctly and effectively obtain the aggregate load model and adjustable power of a large number of the electric water heater group.

关 键 词：可调潜力挖掘 改进DBSCAN聚类算法 RNN特征提取 负荷特性建模 

分 类 号：TM734[电气工程—电力系统及自动化]