Customized Distributed Decision Making for Distribution Networks Using Generalized Fast ADMM

ABSTRACT This paper proposes a customized distributed decision making scheme for distribution networks using the generalized fast alternating direction method of multipliers (GF‐ADMM). In this scheme, an independent distribution system operator (IDSO) maximizes the social welfare by optimally coordinating customers' power usages in a customized and distributed fashion, subject to both network and local customer constraints. This proposed GF‐ADMM‐based strategy leverages the Hessian matrices of customer utility functions to construct a customized matrix for each customer, which the IDSO then uses to update the dual variables in a second‐order Newton‐like manner. This significantly reduces the number of iterations required for convergence, thereby achieving a reduction in total computational and communication cost. The optimality and convergence rate of GF‐ADMM are also established. Numerical experiments on a modified IEEE 123‐bus test system demonstrate the effectiveness and superiority of the proposed method.