A Multiagent Fuzzy Reinforcement Learning Approach for Economic Power Dispatch Considering Multiple Plug-In Electric Vehicle Loads

Abstract

Economic/environmental power dispatch goals were to determine the optimal power generation from thermal plants to satisfy the given demand while diminishing operational costs and also minimizing emissions due to burning of fossil fuels. In this paper, economic and environmental scheduling problems are mathematically formulated by considering various system constraints like the valve point loading effect, ramp rate constraints, plug-in electric vehicle charging constraints, and generator’s capacity constraints. In this work, multiagent fuzzy reinforcement learning (MAFRL) is implemented for an effective elucidation for the economic/environmental power dispatch (EEPD) problem with multiple charging scenarios of plug-in electric vehicle and valve point loading of concern thermal power generators. Here, the EEPD is framed as a multiagent fuzzy reinforcement learning (MAFRL) tasks in which specific plug-in electric vehicles and power generating units act as multiple players for minimizing emissions due to fossil fuels and costs of power generation also satisfying charging constraints of plug-in electric vehicles and several thermal unit’s constraints. To prove the superiority of multiagent fuzzy RL, two standard test functions containing five and fifteen thermal generating units integrated with multiple charging scenarios of plug-in electric vehicles have projected. Numerical consequences and assessment with a numerous current solution methodologies show the potentiality of multiagent fuzzy reinforcement learning method in solving the EEPD problem.