The Experts below are selected from a list of 151668 Experts worldwide ranked by ideXlab platform
Finn Donal - One of the best experts on this subject based on the ideXlab platform.
-
A Centralised Soft Actor Critic Deep Reinforcement Learning Approach to District Demand Side Management through CityLearn
2020Co-Authors: Kathirgamanathan Anjukan, Twardowski Kacper, Mangina Eleni, Finn DonalAbstract:Reinforcement learning is a promising model-free and adaptive controller for demand Side Management, as part of the future smart grid, at the district level. This paper presents the results of the algorithm that was submitted for the CityLearn Challenge, which was hosted in early 2020 with the aim of designing and tuning a reinforcement learning agent to flatten and smooth the aggregated curve of electrical demand of a district of diverse buildings. The proposed solution secured second place in the challenge using a centralised 'Soft Actor Critic' deep reinforcement learning agent that was able to handle continuous action spaces. The controller was able to achieve an averaged score of 0.967 on the challenge dataset comprising of different buildings and climates. This highlights the potential application of deep reinforcement learning as a plug-and-play style controller, that is capable of handling different climates and a heterogenous building stock, for district demand Side Management of buildings.Comment: Accepted for ACM BuildSys2020 RLEM'20 Worksho
-
A Centralised Soft Actor Critic Deep Reinforcement Learning Approach to District Demand Side Management through CityLearn
'Association for Computing Machinery (ACM)', 2020Co-Authors: Kathirgamanathan Anjukan, Twardowski Kacper, Mangina Eleni, Finn DonalAbstract:The 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities (RLEM 2020), New York, United States of America, 17 November 2020Reinforcement learning is a promising model-free and adaptive controller for demand Side Management, as part of the future smart grid, at the district level. This paper presents the results of the algorithm that was submitted for the CityLearn Challenge, which was hosted in early 2020 with the aim of designing and tuning a reinforcement learning agent to flatten and smooth the aggregated curve of electrical demand of a district of diverse buildings. The proposed solution secured second place in the challenge using a centralised 'Soft Actor Critic' deep reinforcement learning agent that was able to handle continuous action spaces. The controller was able to achieve an averaged score of 0.967 on the challenge dataset comprising of different buildings and climates. This highlights the potential application of deep reinforcement learning as a plug-and-play style controller, that is capable of handling different climates and a heterogenous building stock, for district demand Side Management of buildings.Science Foundation IrelandESIPP UC
Rejean Samson - One of the best experts on this subject based on the ideXlab platform.
-
real time environmental assessment of electricity use a tool for sustainable demand Side Management programs
International Journal of Life Cycle Assessment, 2018Co-Authors: Alexandre Milovanoff, Thomas Dandres, Caroline Gaudreault, Mohamed Cheriet, Rejean SamsonAbstract:Demand-Side Management is a promising way to increase the integration of renewable energy sources by adapting part of the demand to balance power systems. However, the main challenges of evaluating the environmental performances of such programs are the temporal variation of electricity generation and the distinction between generation and electricity use by including imports and exports in real-time. In this paper, we assessed the environmental impacts of electricity use in France by developing consumption factors based on historical hourly data of imports, exports, and electricity generation of France, Germany, Great Britain, Italy, Belgium, and Spain. We applied a life cycle approach with four environmental indicators: climate change, human health, ecosystem quality, and resources. The developed dynamic consumption factors were used to assess the environmental performances of demand-Side Management programs through optimized changes in consumption patterns defined by the flexibility of 1 kWh every day in 2012–2014. Between 2012 and 2014, dynamic consumption factors in France were higher on average than generation factors by 21.8% for the climate change indicator. Moreover, the dynamic conSideration of electricity generation of exporting countries is essential to avoid underestimating the impacts of electricity imports and therefore electricity use. The demand response programs showed a range of mitigation up to 38.5% for the climate change indicator. In addition, an environmental optimization cost 1.4 € per kg CO2 eq. for 12% mitigation of emissions as compared to an economic optimization. Finally, embedding the costs of some environmental impacts in the electricity price with a carbon price enhanced the efficiency of economic demand response strategies on the GHG emissions mitigation. The main scientific contribution of this paper is the development of more accurate dynamic electricity consumption factors. The dynamic consumption factors are relevant in LCAs of industrial processes or operational building phases, especially when consumption varies over time and when the power system participates in a wide market with exports and imports such as in France. In the case of demand-Side Management programs, dynamic consumption factors could prevent an environmentally damaging energy from being imported, despite the economic interest of system operators. However, the approach used in this study was attributional and did not assess the local grid responses of load shifting programs. Therefore, a more comprehensive model could be created to assess the local short-term dynamic consequences of located prospective consumptions and the global long-term consequences of demand-Side Management programs.
-
real time environmental assessment of electricity use a tool for sustainable demand Side Management programs
International Journal of Life Cycle Assessment, 2018Co-Authors: Alexandre Milovanoff, Thomas Dandres, Caroline Gaudreault, Mohamed Cheriet, Rejean SamsonAbstract:Purpose Demand-Side Management is a promising way to increase the integration of renewable energy sources by adapting part of the demand to balance power systems. However, the main challenges of evaluating the environmental performances of such programs are the temporal variation of electricity generation and the distinction between generation and electricity use by including imports and exports in real-time.
Kathirgamanathan Anjukan - One of the best experts on this subject based on the ideXlab platform.
-
A Centralised Soft Actor Critic Deep Reinforcement Learning Approach to District Demand Side Management through CityLearn
2020Co-Authors: Kathirgamanathan Anjukan, Twardowski Kacper, Mangina Eleni, Finn DonalAbstract:Reinforcement learning is a promising model-free and adaptive controller for demand Side Management, as part of the future smart grid, at the district level. This paper presents the results of the algorithm that was submitted for the CityLearn Challenge, which was hosted in early 2020 with the aim of designing and tuning a reinforcement learning agent to flatten and smooth the aggregated curve of electrical demand of a district of diverse buildings. The proposed solution secured second place in the challenge using a centralised 'Soft Actor Critic' deep reinforcement learning agent that was able to handle continuous action spaces. The controller was able to achieve an averaged score of 0.967 on the challenge dataset comprising of different buildings and climates. This highlights the potential application of deep reinforcement learning as a plug-and-play style controller, that is capable of handling different climates and a heterogenous building stock, for district demand Side Management of buildings.Comment: Accepted for ACM BuildSys2020 RLEM'20 Worksho
-
A Centralised Soft Actor Critic Deep Reinforcement Learning Approach to District Demand Side Management through CityLearn
'Association for Computing Machinery (ACM)', 2020Co-Authors: Kathirgamanathan Anjukan, Twardowski Kacper, Mangina Eleni, Finn DonalAbstract:The 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities (RLEM 2020), New York, United States of America, 17 November 2020Reinforcement learning is a promising model-free and adaptive controller for demand Side Management, as part of the future smart grid, at the district level. This paper presents the results of the algorithm that was submitted for the CityLearn Challenge, which was hosted in early 2020 with the aim of designing and tuning a reinforcement learning agent to flatten and smooth the aggregated curve of electrical demand of a district of diverse buildings. The proposed solution secured second place in the challenge using a centralised 'Soft Actor Critic' deep reinforcement learning agent that was able to handle continuous action spaces. The controller was able to achieve an averaged score of 0.967 on the challenge dataset comprising of different buildings and climates. This highlights the potential application of deep reinforcement learning as a plug-and-play style controller, that is capable of handling different climates and a heterogenous building stock, for district demand Side Management of buildings.Science Foundation IrelandESIPP UC
Anibal T De Almeida - One of the best experts on this subject based on the ideXlab platform.
-
the role of demand Side Management in the grid integration of wind power
Applied Energy, 2010Co-Authors: Pedro Moura, Anibal T De AlmeidaAbstract:In a scenario of large scale penetration of renewable production from wind and other intermittent resources, it is fundamental that the electric system has appropriate means to compensate the effects of the variability and randomness of the wind power availability. This concern was traditionally addressed by the promotion of wind resource studies and acting in the supply Side of energy and using energy storage technologies. However, in electric system planning, other options deserve to be evaluated, namely the options related with the energy demand. The most severe problems due to the wind power intermittence happen during the peak load hours. Thus, instead of trying to replace the lost capacity due to the intermittence, other option is to act in the energy demand Side, with the aim of reducing the consumption in such hours. The demand-Side Management technologies are an option that must be conSidered to reduce and manage the wind power intermittence. The present paper analyzes the possible impact of demand-Side Management and demand response, with the aim of enabling the integration of the growing intermittent resources in Portugal.
-
multi objective optimization of a mixed renewable system with demand Side Management
Renewable & Sustainable Energy Reviews, 2010Co-Authors: Pedro Moura, Anibal T De AlmeidaAbstract:The 2001/77/CE European Commission Directive sets the target of 22% of gross electricity generation from renewables for the Europe, by 2010. In a scenario of large scale penetration of renewable production from wind and other intermittent resources, it is fundamental that the electric system has appropriate means to compensate the effects of the variability and randomness of the wind, solar and hydro power availability. The paper proposes a novel multi-objective method to optimize the mix of the renewable system maximizing its contribution to the peak load, while minimizing the combined intermittence, at a minimum cost. In such model the contribution of the large-scale demand-Side Management and demand response technologies are also conSidered.
Zhu Han - One of the best experts on this subject based on the ideXlab platform.
-
distributed demand Side Management with energy storage in smart grid
IEEE Transactions on Parallel and Distributed Systems, 2015Co-Authors: Hung Khanh Nguyen, Ju Bin Song, Zhu HanAbstract:Demand-Side Management, together with the integration of distributed energy storage have an essential role in the process of improving the efficiency and reliability of the power grid. In this paper, we conSider a smart power system in which users are equipped with energy storage devices. Users will request their energy demands from an energy provider who determines their energy payments based on the load profiles of users. By scheduling the energy consumption and storage of users regulated by a central controller, the energy provider tries to minimize the square euclidean distance between the instantaneous energy demand and the average demand of the power system. The users intend to reduce their energy payment by jointly scheduling their appliances and controlling the charging and discharging process for their energy storage devices. We apply game theory to formulate the energy consumption and storage game for the distributed design, in which the players are the users and their strategies are the energy consumption schedules for appliances and storage devices. Based on the game theory setup and proximal decomposition, we also propose two distributed demand Side Management algorithms executed by users in which each user tries to minimize its energy payment, while still preserving the privacy of users as well as minimizing the amount of required signaling with the central controller. In simulation results, we show that the proposed algorithms provide optimality for both energy provider and users.
-
a differential game approach to distributed demand Side Management in smart grid
International Conference on Communications, 2012Co-Authors: Quanyan Zhu, Zhu Han, Tamer BasarAbstract:Smart grid is a visionary user-centric system that will elevate the conventional power grid system to one which functions more cooperatively, responsively, and economically. Dynamic demand Side Management is one of the key issues that enable the implementation of smart grid. In this paper, we use the framework of dynamic games to model the distribution demand Side Management. The market price is characterized as the dynamic state using a sticky price model. A two-layer optimization framework is established. At the lower level, for each player (such as one household), different appliances are scheduled for energy consumption. At the upper level, the dynamic game is used to capture the interaction among different players in their demand responses through the market price. We analyze the N-person nonzero-sum stochastic differential game and characterize its feedback Nash equilibrium. A special case of homogeneous users is investigated in detail and we provide a closed-form solution for the optimal demand response. From the simulation results, we demonstrate the use of demand response strategy from the game-theoretic framework and study the behavior of market price and demand responses to different parameters.
-
demand Side Management to reduce peak to average ratio using game theory in smart grid
International Conference on Computer Communications, 2012Co-Authors: Hung Khanh Nguyen, Ju Bin Song, Zhu HanAbstract:In this paper, we propose a novel demand Side Management technique to reduce the peak load of the system. We conSider a smart power system with distributed users that request their energy demands to an energy provider and the energy provider dynamically updates the energy prices based on the load profiles of the users. The users try to minimize the Peak-to-Average Ratio (PAR) of the power system by charging for their batteries at low-demand periods and discharging the energy at high-demand periods. We also propose a distributed demand Side Management algorithm using a game theoretical approach in which each user tries to minimize its total energy cost. In simulation results, we show that the proposed algorithm will simultaneously minimize the PAR and the total energy cost.
