Reinforcement Learning based Energy Management System for Sustainable Electrified Urban Transportation Systems with The Integration of Renewable Energy Systems

Main Article Content

Amany El-Zonkoly

Abstract

Targeting the increase access to efficient, safe and low carbon public transportation, the city of Alexandria, Egypt has adopted a fleet of electric buses. In addition, a project of upgrading and electrifying the existing urban railway system is in progress, which will alleviate traffic congestion on roads of this highly populated city, saving time and reducing fuel consumption. Electric vehicles (EVs) parking lots are also considered. In this paper, the integration of roof top photovoltaic (PV) systems and green-hydrogen powered gas turbines are considered as part of the integrated energy system (IES) of the urban metro system of the city of Alexandria. An optimal management of this IES is proposed in order to achieve the maximum benefit of integrating renewable energy resources into urban transportation system (UTS). The proposed energy management algorithm includes demand side management (DSM) of UTS loads and EVs, which increases the difficulty of decision-making process due to the high uncertainty of decision variables and the large search area. In order to deal with such challenge, a modified multi-agent reinforcement learning (MRL) is applied for decision making. The optimization problem is formulated as a finite Markov decision process (FMDP). The mixed uncertainties are introduced as additional states and action scenarios in the MRL algorithm. The simulation results show the economic potential of integrating renewable and sustainable energy resources into the IES of the electrified urban transportation system, in which the average daily cost of energy consumption can be reduced by 38.9%.

Article Details

Section
Articles