Indoor thermal comfort control method and system based on improved DQN

The invention discloses an improved DQN-based indoor thermal comfort control method and system. The method comprises the following steps: acquiring a physical environment index value influencing an indoor thermal comfort index PMV in real time; establishing an indoor thermal comfort state model by utilizing the physical environment index value; judging the indoor thermal comfort state model through a control strategy to obtain an air conditioning system judgment action, performing action by utilizing an air conditioning system action space to obtain an indoor thermal comfort state at a certain moment, mapping the indoor thermal comfort state with the air conditioning system action at the corresponding moment, and mapping the state sensed in the environment into an independent numerical value as a reward function; and in a deep reinforcement learning network of a DQN control strategy based on dual-experience pool priority experience playback, determining a reward value according to a reward function, carrying