Loading...

Optimizing Smart Energy Hub Operation by Reinforcement Learning Approach

Rayati, Mohammad | 2015

881 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47420 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Ranjbar, Ali Mohammad
  7. Abstract:
  8. Nowadays, demands increment is one of the greatest concerns of energy systems operators. Consequently, enhancing the efficiency of consumption is an inevitable goal, which leads us to a concept named as energy hub (EH). In a simple definition, an EH is a multi-generation system where different energy carriers are converted and stored by various devices such as combined cooling, heating and power (CCHP), battery, and boiler to meet the load demands. In addition, with advent of smart grids (SG), demand side management (DSM) becomes one of the effective techniques with significant role in optimizing energy systems performance. DSM commonly refers to methods implemented by utility companies to reduce or shift the energy consumption at customers’ side. It is worth mentioning that DSM programs fall in short in a system with only must-run loads, i.e. with strict energy consumption scheduling constraints. The current thesis addresses the short coming of DSM by introducing a new concept entitled Smart Energy Hub (S.E. Hub) enabling customers, even those with only must-run loads, to be active in DSM programs. In this thesis, a fully automated demand side management program is proposed based on a reinforcement learning (RL) algorithm to motivate customers for reducing peak load in electricity network. RL is one of machine intelligence techniques proposing to solve the real-world problem with dynamic environment, such as electricity, heating, cooling loads, and prices. RL is the study of programs that adapt a learner to an unknown and dynamic environment by receiving rewards and punishments from the environment. The proposed method has a benefit over most of the earlier works in a sense that it does not require any parameters of the system. The proposed method determines the customers’ satisfaction function, energy carrier prices, and appliances efficiencies based on customers’ historical actions. Simulations are performed for the sample model and results depict how much of each energy carrier the S.E. Hub should consume and how they should be converted to meet the load at S.E. Hub’s outputs. It is also shown that the proposed RL algorithm reduces customer’s energy bill and electrical peak load up to 20 % and 24 %, respectively
  9. Keywords:
  10. Optimization ; Smart Power Grid ; Reinforcement Learning ; Demand Side Mangement ; Energy Hub ; Smart Energy Hub

 Digital Object List

 Bookmark

...see more