Multi-objective optimization-driven machine learning for charging and V2G pattern for plug-in hybrid vehicles: Balancing battery aging and power management

Abstract

This research study aims to optimize the economic energy management of a plug-in hybrid vehicle by maximizing revenue through vehicle-to-grid interactions during peak hours, while also prioritizing the reduction of battery aging as an additional objective. The study focuses on the Mitsubishi Outlander vehicle in New York City, analyzing charging and V2G times, rates, and ambient temperatures to understand their impact on lithium-ion battery aging. Using a machine learning approach, a deep neural network is trained with time, rate, and temperature inputs to improve accuracy in predicting battery aging effects. A multi-objective optimization algorithm, NSGA-II, is employed to balance the trade-off between maximizing revenue and minimizing battery aging based on the neural network's predictions. The results indicate that the best time for charging is before starting the travel, with a 3.95 % battery aging. Raising the charging amperage and implementing V2G elevate battery aging to 7.83 %. Furthermore, higher ambient temperatures can accelerate battery aging significantly. A temperature rise from 10 °C to 30 °C could potentially result in a 12.23 % increase in battery aging. Through optimum selection of charging and V2G time, rate, and temperature, the study suggests that an income of $1043 and a battery aging rate of 7.2 % can be achieved within one year.