Optimal Location and Pricing of Electric Vehicle Charging Stations Using Machine Learning and Stackelberg Game

The widespread adoption of electric vehicles (EVs) requires strategically located and well-priced charging stations (CSs) to facilitate the charging and discharging of EVs. To implement this necessity, a two-stage framework is proposed that involves demand forecasting and an optimization model to optimize the location and pricing of CSs. The first stage employs a Long Short-Term Memory (LSTM) model to forecast the 30-day energy demand for CSs using historical data from New South Wales (NSW), Australia. The energy demand is integrated into a bilevel optimization problem modeled as a Stackelberg game. Considering the energy demand, the leader in the game strategically selects the locations for new CSs, while the followers determine their charging prices at each location to maximize payoffs. In this article, an in-depth theoretical and analytical analysis of the potential new locations is performed to understand the demand and profitability at these locations. Moreover, a detailed game theoretic analysis is presented, considering both static and dynamic games to illustrate their impact on the payoff of CSs. Furthermore, a penalty function is designed at the follower end to limit the charging prices within reasonable bounds to improve the social welfare of the market mechanism. Overall, this paper presents a comprehensive scheme that offers a systematic approach to optimizing the location and pricing decisions for CSs.