A Simulation Environment of Solar-Wind Powered Electric Vehicle Car Park for Reinforcement Learning and Optimization

Li, Handong, Dai, Xuewu, Kotter, Richard, Aslam, Nauman and Cao, Yue (2022) A Simulation Environment of Solar-Wind Powered Electric Vehicle Car Park for Reinforcement Learning and Optimization. In: The 3rd International Symposium on New Energy and Electrical Technology ISNEET 2022. Lecture Notes in Electrical Engineering . Springer, Cham, pp. 1-6. (In Press)

Preview

Text LiHD2022_ISNEET_SP_Powered_EV_Car_Park_Environment_Design.pdf - Accepted Version Download (1MB) | Preview

Abstract

The transportation sector is the second greatest contributor to carbon emissions in the UK and Newcastle upon Tyne, accounting for around 33 of total emissions in 2020. In accordance with the United Kingdom’s goal to reach net zero by2050 (and the city of Newcastle upon Tyne’s ambition to do so by 2030), electric vehicles (EVs) play a crucial role in achieving net zero road transportation. However, if the electricity used to charge EVs is derived from fossil fuels, this does not necessarilyimply a reduction of overall emissions nationally or globally. To achieve optimal EV charging, a deeper comprehension of the unpredictability of (on-site renewable energy sources) ORES energy output is required. In this paper, the predicted renewableenergy generated is used as the actual value for the reinforcement learning algorithm simulation environment. Such a model is able to represent the relationship between the power generation and the wind speed as well as solar irradiation, which arecharacterized by significant uncertainties due to weather changes in both the short-time (hourly) and long-term (seasonally). The uncertainty analysis shows that the uncertainties in wind speed at Newcastle upon Tyne can be modelled as a Weibull distribution with parameters A = 19.98 and B = 1.91. As for energy demand,this paper integrates information from an Oslo (Norway) car parking garage based set of EV charging stations with EVs’ demand statistics. The charging habits of EV users range from 800 minutes to 1,000 minutes of parking time, and from 5 kWh to20 kWh in terms of charging energy. The maximum connection frequency for EV charging is 20 minutes. In addition, this paper develops methods for stochastic EV charging and parking space occupancy employing actual data. On the basis of the aforesaid renewable energy generation and the EV charging status, it ispossible to develop a decision algorithm to optimal renewable energy efficiency.

Actions (login required)

View Item

Downloads