Abstract: Distributed scheduling has become a novel dispatch strategy for integrated energy systems (IES) due to its robustness and flexibility. However, frequent information exchanges under a distributed framework can lead to the leakage of private information via communication pathways to external attackers, thereby affecting the optimal operational state of the system. This paper first introduces multi-agent consensus theory into the distributed framework to address this issue, presenting a collaborative economic equation for IES and considering two typical eavesdropping models. Secondly, using the concept of splitting and recombining the real-time status information of each agent, a privacy-preserving distributed dispatch algorithm based on state decomposition (PPDDASD) is proposed, and its convergence and the ability to protect privacy under an eavesdropping attack environment are rigorously proven theoretically. Finally, the feasibility and superiority of the proposed algorithm are validated from different perspectives, such as ideal communication environment, plug-and-play characteristics, and comparison with differential privacy to counter eavesdroppers, based on a case study of the IEEE39-32 thermoelectric coupled energy system.