Abstract: To enhance renewable energy utilization and reduce carbon emissions in hydrogen-integrated energy systems (H-IES), this study proposes a deep reinforcement learning-based scheduling strategy. Firstly, an H-IES model incorporating hydrogen energy-related equipment is established, and a reward-penalty ladder-type carbon trading mechanism is introduced to incentivize emission reduction behaviors via stepwise carbon prices. Subsequently, the uncertainty optimization scheduling problem is transformed into a sequential decision-making task using the Markov decision process (MDP). Considering multiple objectives—such as total cost and renewable energy consumption—an adaptive reward function is designed to dynamically adjust their respective weights. Finally, based on the measured data, the agent is trained using the twin delayed deep deterministic policy gradient (TD3) algorithm. Case study results indicate that the proposed method can effectively exploit the carbon emission reduction potential of H-IES, enhance renewable energy utilization, and achieve dual optimization of economic performance and environmental sustainability.