Combined heat and power dispatch (CHPD) opens a new window for increasing operational flexibility and reducing wind power curtailment. Electric power and district heating systems are independently controlled by different system operators; therefore

a decentralized solution paradigm is necessary for CHPD

in which only minor boundary information is required to be exchanged via a communication network. However

a nonideal communication environment with noise could lead to divergence or incorrect solutions of decentralized algorithms. To bridge this gap

this paper proposes a stochastic accelerated alternating direction method of multipliers (SA-ADMM) for hedging communication noise in CHPD. This algorithm provides a general framework to address more types of constraint sets and separable objective functions than the existing stochastic ADMM. Different from the single noise sources considered in the existing stochastic approximation methods

communication noise from multiple sources is addressed in both the local calculation and the variable update stages. Case studies of two test systems validate the effectiveness and robustness of the proposed SA-ADMM.