Progressive Identification of Distribution Network Topology Based on User-Side Internet of Things Device Measurement Data

Abstract

The flexibility requirements of network topologies in advanced distribution networks (ADNs) have led to increasingly complex distribution network topologies, posing a significant challenge for topology identification. The integration of a vast array of internet of things (IoT) devices at existing distribution network terminals facilitates the real-time collection and transmission of physical measurement data, thus providing new ideas for identifying distribution network topologies. This paper introduces a distribution network topology progressive identification method grounded in measurement data captured by IoT devices. First, the study employs similarity mining of IoT measurement data to identify the topological connection relationships within the low-voltage distribution network. Second, by integrating real-time measurement data with historical data, the medium-voltage-side data are consolidated, and the topology of the medium-voltage distribution network is identified by an inverse power flow model based on a fixed parameter ratio. Furthermore, to address the limitations of low recognition accuracy and non-unique recognition outcomes associated with the aforementioned methods, this study examines the regional similarity among distribution networks and proposes a hyperparameter to increase the recognition accuracy. Finally, different case studies show that the proposed method maintains a topological identification accuracy of more than 80%, and shows a wide range of applicability in different types and sizes of distribution networks.