Adaptive topology identification with dynamic update mechanisms for flexible interconnected distribution networks

Abstract

The flexible interconnection in the distribution network will arise new challenges in terms of network structure and operation characteristics. The complexity arises as topology changes become less apparent due to the support of soft open points (SOPs) for nodal power and voltage. Additionally, outages may not be reported after switches trip, making detection more difficult. To address these challenges, a two-stage topology identification (TI) framework for flexible interconnected distribution networks is investigated, integrating historical and real-time TI. For historical data samples that include multiple topology categories, all topology categories and the topology type to which each sample belongs are identified based on a modified expectation-maximization (EM) algorithm. Then, the topology classifier is trained to perform the real-time TI based on historical labeled samples, and the credibility of obtained real-time identification results is analyzed. Furthermore, different classifier update mechanisms are designed to accommodate the emergence of new topology categories driven by the integration of renewable energy sources (RESs) and growing load demands in distribution networks. Finally, the feasibility and effectiveness of the proposed framework and models are validated through testing on both IEEE 33-bus and IEEE 118-bus test systems. The framework demonstrates remarkable adaptability across diverse network scenarios, including flexible interconnected networks, RES-penetrated networks, and networks with latent faults. Experimental results and performance evaluations confirm that the proposed framework exhibits superior practical applicability.