Deep reinforcement learning for home energy management system control

作     者：Paulo Lissa Conor Deane Michael Schukat Federico Seri Marcus Keane Enda Barrett 

作者机构：College of Science and EngineeringNational University of IrelandGalwayIreland Informatics Research Unit for Sustainable Engineering(IRUSE)GalwayIreland Ryan InstituteNational University of Ireland GalwayIreland 

出 版 物：《Energy and AI》 (能源与人工智能（英文）)

年 卷 期：2021年第3卷第1期

页      面：64-72页

核心收录：

学科分类：0202[经济学-应用经济学] 02[经济学] 020205[经济学-产业经济学] 

基　　金：This research work was funded by the European Union under the RESPOND project with Grant agreement No.768619 

主　　题：Deep reinforcement learning Residential home energy management Demand response Autonomous control 

摘      要：The use of machine learning techniques has been proven to be a viable solution for smart home energy *** techniques autonomously control heating and domestic hot water systems,which are the most relevant loads in a dwelling,helping consumers to reduce energy consumption and also improving their ***,the number of houses equipped with renewable energy resources is increasing,and this is a key ele-ment for energy usage optimization,where coordinating loads and production can bring additional savings and reduce peak *** this regard,we propose the development of a deep reinforcement learning(DRL)algorithm for indoor and domestic hot water temperature control,aiming to reduce energy consumption by optimizing the usage of PV energy ***,a methodology for a new dynamic indoor temperature setpoint definition is presented,thus allowing greater flexibility and *** results show that the proposed DRL al-gorithm combined with the dynamic setpoint achieved on average 8%of energy savings compared to a rule-based algorithm,reaching up to 16%of savings over the summer ***,the users’comfort has not been compromised,as the algorithm is calibrated to not exceed more than 1%of the time out the specified temperature *** analysis shows that further savings could be achieved if the time out of comfort is increased,which could be agreed according to users’*** demand side management,the DRL control shows efficiency by anticipating and delaying actions for a PV self-consumption optimization,performing over 10%of load ***,the renewable energy consumption is 9.5%higher for the DRL-based model compared to the rule-based,which means less energy consumed from the grid.