Algorithms and Electric Vehicles Revolutionize Smart Charging at Argonne’s Energy Plaza
Argonne National Laboratory, in collaboration with the University of Chicago, is spearheading a groundbreaking project to address the challenges of efficiently charging electric vehicles (EVs) on the power grid. The initiative aims to train an algorithm, using reinforcement learning techniques, to intelligently schedule and manage the charging of a diverse fleet of EVs.
The first phase of the project involves studying the charging patterns of Argonne employees’ vehicles at the laboratory’s Smart Energy Plaza. This facility offers both regular AC chargers and fast DC chargers, giving researchers valuable insights into the charging requirements of different EV owners. By analyzing data and considering factors such as departure times and peak demands on the grid, the algorithm can optimize the charging process, ensuring efficiency and low-cost charging.
When multiple EVs charge simultaneously, it can strain the power station, resulting in increased charges. To mitigate this, the reinforcement learning algorithm incorporates feedback from positive and negative outcomes. For example, if an EV achieves the desired charge level at the designated departure time, it is considered a positive result. On the other hand, drawing power beyond a predetermined peak threshold is viewed as a negative outcome. By learning from these experiences, the algorithm becomes increasingly adept at making intelligent decisions concerning which vehicles to charge and when.
Jason Harper, principal electrical engineer at Argonne, explains that the essence of smart charge scheduling lies in optimization. The charging station constantly balances various tradeoffs to ensure each car is charged as efficiently as possible. Although the project is currently in its early stages at Argonne, there is potential for its expansion beyond the laboratory’s boundaries. At-home charging, for instance, offers more flexibility due to overnight charging possibilities, allowing for better distribution of the charging load.
Harper emphasizes that genuine smart charging entails considering all stakeholders in the ecosystem, including the utility companies, charging station owners, and EV drivers or homeowners. The aim is to meet the needs of everyone involved while remaining mindful of their constraints.
Future iterations of the project will involve simulating a larger charging network using data collected from Argonne’s chargers. Additionally, the team has developed a mobile app named EVrest, enabling users of networked charging stations, initially including Argonne employees, to reserve stations and participate in smart charge scheduling. This platform collects charging behavior data, which will further train AI models for improved smart charge management and vehicle grid integration.
With its potential to revolutionize the EV charging landscape, this collaborative project between Argonne National Laboratory and the University of Chicago holds great promise for achieving efficient, cost-effective, and eco-friendly charging solutions. By leveraging the power of algorithms and reinforcement learning, we are one step closer to a truly transformative electric vehicle revolution.
