Model-predictive control and reinforcement learning in multi-energy system case studies
نویسندگان
چکیده
Model predictive control (MPC) offers an optimal technique to establish and ensure that the total operation cost of multi-energy systems remains at a minimum while fulfilling all system constraints. However, this method presumes adequate model underlying dynamics, which is prone modelling errors not necessarily adaptive. This has associated initial ongoing project-specific engineering cost. In paper, we present on- off-policy multi-objective reinforcement learning (RL) approach does assume priori , benchmarking against linear MPC (LMPC — reflect current practice, though non-linear performs better) - both derived from general problem, highlighting their differences similarities. simple (MES) configuration case study, show twin delayed deep deterministic policy gradient (TD3) RL agent potential match outperform perfect foresight LMPC benchmark (101.5%). realistic LMPC, i.e. imperfect predictions, only achieves 98%. While in more complex MES configuration, agent’s performance generally lower (94.6%), yet still better than (88.9%). studies, agents outperformed after training period 2 years using quarterly interactions with environment. We conclude viable for given constraint handling pre-training, avoid unsafe long periods, as proposed fundamental future work. • An integrated strategy enables efficient use energy costs. Model-predictive have common mathematical ground. Reinforcement learning-based management require models. can model-predictive training. Safety fast convergence remain challenges learning.
منابع مشابه
Reinforcement Learning Based PID Control of Wind Energy Conversion Systems
In this paper an adaptive PID controller for Wind Energy Conversion Systems (WECS) has been developed. Theadaptation technique applied to this controller is based on Reinforcement Learning (RL) theory. Nonlinearcharacteristics of wind variations as plant input, wind turbine structure and generator operational behaviordemand for high quality adaptive controller to ensure both robust stability an...
متن کاملModel Predictive Control and Reinforcement Learning as Two Complementary Frameworks
Model predictive control (MPC) and reinforcement learning (RL) are two popular families of methods to control system dynamics. In their traditional setting, they formulate the control problem as a discrete-time optimal control problem and compute a suboptimal control policy. We present in this paper in a unified framework these two families of methods. We run for MPC and RL algorithms simulatio...
متن کاملData-Efficient Reinforcement Learning with Probabilistic Model Predictive Control
Trial-and-error based reinforcement learning (RL) has seen rapid advancements in recent times, especially with the advent of deep neural networks. However, the majority of autonomous RL algorithms require a large number of interactions with the environment. A large number of interactions may be impractical in many real-world applications, such as robotics, and many practical systems have to obe...
متن کاملreinforcement learning based pid control of wind energy conversion systems
in this paper an adaptive pid controller for wind energy conversion systems (wecs) has been developed. theadaptation technique applied to this controller is based on reinforcement learning (rl) theory. nonlinearcharacteristics of wind variations as plant input, wind turbine structure and generator operational behaviordemand for high quality adaptive controller to ensure both robust stability an...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Applied Energy
سال: 2021
ISSN: ['0306-2619', '1872-9118']
DOI: https://doi.org/10.1016/j.apenergy.2021.117634