Frequency‐Domain Subspace Identification for Noninvasive Transformer Winding Parameter Estimation and Fault Diagnostics

ABSTRACT A novel frequency‐domain subspace identification method enables precise estimation of transformer winding ladder network parameters directly from terminal measurements. Unlike traditional approaches requiring complex design data or intrusive simulations, this technique offers a practical solution for assessing operational transformers. By analyzing driving‐point impedance frequency response data, the method accurately determines critical parameters, including series and ground capacitances, resistances, and self and mutual inductances. Experimental validation demonstrates exceptional accuracy, with parameter estimations deviating less than 1% from finite element method simulations and analytical formulas. The technique's utility extends beyond parameter identification, showing promising capabilities in fault diagnosis by detecting mechanical and electrical faults. This innovative approach provides a noninvasive, computationally efficient method for transformer diagnostics and condition monitoring, offering significant advantages over existing parameter estimation techniques.