This paper develops an operational modal analysis application framework for vibration measurement‐based condition monitoring of transformer winding structure. Within the framework, load transient signals generated by turning on and off transformer or changing transformer load are used as the broadband excitations. Frequency domain decomposition (FDD) technique is used to process the measured transformer vibration signals to detect possible mechanical changes to the transformer's winding and core structure. Case studies demonstrate the OMA application framework (broadband excitation and FDD) can achieve higher accuracy of natural frequency detection of windings, which can pave a way for online transformer winding and core condition monitoring.