Research on Partial Discharge Localization Method in Transformers Based on Fiber Optic Fabry-Perot Sensing and SOGWO Algorithm

Insulation failure is the leading cause of transformer faults, with partial discharge (PD) serving as a critical early indicator. This study addresses PD detection and localization in large oil-immersed power transformers using fiber optic FabryPerot (F-P) ultrasonic sensors, which offer advantages over traditional piezoelectric sensors, including electromagnetic immunity, resilience, and internal installation capability. Focusing on full-scale transformer applications, the research proposes a PD localization method integrating multi-frequency joint fiber optic F-P sensing with a Selective Opposition-based Grey Wolf Optimizer (SOGWO) algorithm. An adaptive Wiener filter combined with the Teager Energy Operator is developed for precise ultrasonic arrival time extraction. Experiments conducted on a 110 kV prototype transformer demonstrate the superior sensitivity and noise immunity of the fiber optic F-P sensors compared to external piezoelectric sensors. By accounting for ultrasonic time delay increments ($\xi_{\mathbf{i}}$) caused by internal structures (e.g., windings, cores), the SOGWO-based localization method significantly improves accuracy: positioning errors for two PD points ($x_{1}, x_{2}$) are reduced from approximately 37 cm and 56 cm (without considering $\xi_{i}$) to 3.14 cm and 4.47 cm, respectively. These findings validate the practical potential of fiber optic $\mathrm{F}-\mathrm{P}$ sensing combined with the proposed algorithms for accurate PD localization within complex transformer geometries.