Abstract: Potential transformers in substations are prone to resonance overvoltages during operation, which may cause measurement distortion and insulation breakdown. This paper presents an in-depth study of suppression techniques tailored to substation operation and maintenance. A wavelet-based suppression strategy is proposed, combining multi-scale adaptive decomposition with energy-reliability weighting: an adaptive scale-frequency mapping precisely identifies the resonant bands, while a multi-stage RC damping network and a first-order digital feedback filter jointly implement dynamic suppression of impulse and switching transients. A field trial on a substation voltage transformer demonstrates that the method significantly improves peak suppression rate and shortens oscillation decay time, with excellent online deployment performance.