Electric vehicle demand response and charging scheduling method based on deep reinforcement learning

The invention relates to an electric vehicle demand response and charging scheduling method based on deep reinforcement learning, the method can be executed by one or more processors, and the method comprises the following steps: the one or more processors construct an electric vehicle demand response and electric vehicle aggregator operation scheduling model; the one or more processors classify the vehicle based on the charging flexibility of the vehicle; the one or more processors construct an operation process of the electric vehicle aggregator into a Markov decision process capable of ensuring that constraint conditions are satisfied, and determine a state St, an action at and a reward Rt corresponding to a moment t; the one or more processors solve the Markov decision process, and a robust agent containing mapping from any state St to action at is obtained through training; and the one or more processors deploy the robust intelligent agent into the electric vehicle charging station so as to perform optim