Abstract: Oilless equipment has the objective characteristics of very low oil quantity, fully sealed steely structure, and lack of explosion-proof measures. Long-term partial discharge (PD) insulation defects can induce equipment explosions, fires and other safety accidents, while oilless equipment has been lacking effective PD insulation defect detection and diagnosis techniques. Based on this, this paper builds an experimental platform for oil surface gas detection of PD insulation defects in 110 kV positive and vertical current transformers, and experimentally investigates the characteristic oil surface gas variation law and its correlation characteristics with dissolved gas in oil under different PD strengths of oilless equipment. The test results show that the PD characteristic gases H₂, CH₄ and CO are significantly enriched on the oil surface and the concentration increase rate is much higher than that in the oil, so the PD detection based on the oil surface gas has higher sensitivity than the oil chromatography detection. When the PD insulation defect is seriously deteriorated, the dissolved gas concentration in the oil does not increase significantly and the key gas C₂H₂ does not reach the warning value, while the characteristic gas concentration on the oil surface all show a significant and sharp increase, among which the concentration of H₂ and C₂H₂ is the most significant, with the concentration of C₂H₂ reaching 17.41×10⁻⁶. By capturing this change during the deterioration of the discharge defect, the occurrence of breakdown discharge can be effectively warned. The oil surface gas concentration can effectively characterize the generation and development of PD defects within oilless equipment. The results of this paper provide a new solution for the highly sensitive monitoring of PD insulation defects in oilless equipment.