With the deepening of the concept of sustainable development, green building has become an important transformation direction for the construction industry. This paper takes green building as the core object of its research, and investigates the feasibility and regulation of its renewable energy system. The DDPG algorithm based on continuous action control in reinforcement learning algorithm is proposed to optimize and regulate the renewable energy system of green buildings, which specifies the state space, action space and their corresponding simpler constraint requirements in the DDPG regulation model, and the setting of the reward function is consistent with that of the deep Q-network algorithm (DQN), etc., and takes the user’s comprehensive energy cost and the utilization of the storage system as a benchmark for the system regulation. Measurement benchmark. Simulation experiments are being conducted to evaluate the effectiveness of the system regulation strategy proposed in this paper, which is based on the DDPG algorithm, for optimizing renewable energy systems in green buildings. Comparing Scheme 1, which only uses ON/OFF strategy, and Scheme 2, which is regulated without coordination, the operating costs of this paper’s regulation scheme are reduced by 24.63% and 5.07%, respectively, and the operating costs at 0.9, 1.8, and 2.4°F conditions are also the lowest of 2181.3, 2284.4, and 2284 yuan, while having smaller temperature deviations.