Multi-objective scheduling optimization for electric vehicles based on deep reinforcement learning

This paper proposes a multi-objective scheduling optimization model based on the TD3 deep reinforcement learning algorithm to address the challenges of rising user charging costs and increased grid load fluctuations resulting from the uncoordinated integration of large-scale electric vehicles (EVs) into the grid. The proposed model represents the charging behavior of EVs as a Markov decision process, which characterizes the charging scenario through state, action, and reward functions, and dynamically optimizes the charging strategy with the assistance of intelligent agents. To improve the efficiency and accuracy of solving the optimal scheduling strategy, this paper introduces the TD3 algorithm, which effectively alleviates the problem of overestimation of Q value in traditional methods and enhances the convergence and stability of the strategy. The objective functions of this article consider EV aggregator profits, user charging costs, power grid load fluctuations, and green power consumption to balance the interests of stakeholders. The experimental results show that after optimization, the standard deviation of power grid load decreases by14.3%. The user charging cost decreases from 37.51 to 29.94 Yuan, saving 20.2%. Under multiple uncertain factors, such as vehicle size, user engagement rate, and seasonal factors, the proposed model exhibits good adaptability and robustness.