Detection of winding deformations in power transformers from inrush current characteristics using the Chen-Lee...

A short-circuit fault in a transmission line unbalances the electromagnetic stress or ferromagnetic resonance, which deforms the internal windings of a power transformer. In addition, aged power transformers that have been operating for >25 years may have loosened windings, which can weaken the material or prolong exposure to short-circuit faults and ultimately reduce the mechanical strength. As the time and frequency of the inrush current waveforms exhibit localized characteristics, the time series of one-dimensional normal and test inrush current waveforms can be converted using the dynamic error equation of the Chen-Lee chaotic system to yield a one-dimensional time series of dynamic error values. The features for inner-product calculation and angular transformation can then be extracted using Gramian Angular Summation Field. This process effectively captures the differences in time-frequency characteristics and dynamic structures of the time-series data of the normal and test inrush currents while retaining key information for dimensionality reduction and feature extraction. Moreover, it utilizes synchronized chaotic dynamics to detect nuanced mechanical distinctions between undamaged and deformed windings that are subsequently converted into time-conserved images. In contrast to conventional frequency response analyzers or artificial intelligence-driven diagnostic techniques, our method does not require offline calibration or specialized testing apparatus. It provides a basis for predictive diagnostics to identify potential failures in power supply services.