A Novel Self-Healing Strategy for Distribution Network with Distributed Generators Considering Uncertain Power-Quality Constraints

Abstract

Self-healing of smart distribution networks with distributed generators, which are usually operated as independent islands after fault, can improve power-supply reliability. As a hot research topic, a self-healing scheme is usually treated as the output of a nonlinear optimizuoation model. However, existing strategies have two main shortcomings. The first, high-optimization dimension, results in low-optimization efficiency. The second, the effects of power-quality issues, which are more serious on islands and may further threaten the safe operation of islands, is usually neglected. To quickly obtain a reliable self-healing scheme, a novel strategy is proposed. As the first step, the distribution network after a fault occurrence can be divided into several trouble-free subnets via the connectivity analysis; each subnet is called an initial island. Further, for each initial island, a two-step optimization model of self-healing, which consists of load-shedding optimization and network reconfiguration optimization, is proposed to obtain the self-healing strategy with lower searching space as well as higher solving efficiency. In detail, in load-shedding optimization, by means of heuristic differential evolution algorithm, larger total recovery capacity is achieved by considering the droop characteristic of distributed generators (DGs) within the permissible change in frequency. In network-reconfiguration optimization, based on the improved hybrid particle swarm optimization algorithm, a comprehensive set of power-quality constraints, including constraint of change in frequency, uncertain constraints of node voltage total harmonic distortion (THD), and negative sequence components of DGs, is developed to guarantee the reliability of each island. To evaluate whether the constraints are satisfied during the optimization procedure, an improved flexible power-flow algorithm is developed to calculate the power flow of each island under change in frequency. Further, 2m+1-point estimate method is employed for uncertainty analyses of the distributions of harmonic and negative sequence components caused by the uncertainty of corresponding sources. Finally, via a 94-node practical distribution network, the effectiveness and advantages of the proposed strategy in safety, recovery capacity, and optimization efficiency are verified.