Detection of Winding Faults Based on a Characterization of the Nonlinear Dynamics of Transformers

The electrical-mechanical dynamics of a transformer may be described by a model with cascade-connected linear and nonlinear parts. This paper shows that mechanical faults have a significant influence on the linear part of the model. A change in the system order of the linear part of the model can be used as an indicator of variance in a transformer's physical characteristics. A technique based on the Lipschitz criterion is developed to estimate the system order, which is then used for detecting changes in the vibration of the system caused by mechanical faults in the windings. The proposed method is tested on a three-phase power transformer, demonstrating promising detection of faults in transformer windings.