A Data-Driven Method for Residential Transformers Winding Connections and Capacities Recognition in the Distribution Management System

Due to manual registration in the Distribution Management System (DMS), recording errors in the winding connection and capacity of residential trans-formers are inevitable. This paper proposes a linear regression classification method, which can estimate the real winding connection and capacity of residential transformers by exploring the historical operating data sets of residential transformers in the DMS. The linear regression classifications of three-phase outlet voltages are used for winding connections recognition. The inherent law is revealed that Dyn11 and Yyn0 residential transformers are with different zero-sequence current paths. That is different capacities Yyn0 residential transformer is with different zero-sequences, and short circuit impedance of the Dyn11 residential transformers is a fixed value. Ac-cording to the identification of winding connection, the capacity identification of Yyn0 and Dyn11 residential transformers can be obtained. Simulations on a practical line with 10 residential transformers show the effective-ness of the proposed method.