Abstract: Rapid and real-time monitoring of dissolved gas content in transformer oil can effectively provide early warning and diagnosis of potential faults in transformers, thus the title research had been conducted. A set of on-line monitoring system for dissolved gases in oil was built using a highly breathable Teflon AF2400 membrane module and a highly sensitive photoacoustic spectroscopy detector, with the minimum detection time of 2.5 min. A theoretical degassing model was constructed and the accuracy of the model was demonstrated from the perspectives of gas chamber volume, oil temperature, and volume fraction of common transformer fault gases (CO, C₂H₂, C₂H₄, C₂H₆, CH₄) in oil. The model could predict the equilibrium volume fraction of faulty gas degassing within 20 min, and this time was further reduced by reducing the gas chamber volume or increasing the oil temperature. When the volume fraction of faulty gas in transformer oil changed sharply, the relative deviation of the equilibrium volume fraction between the model built and the actual detection was less than 7.0% within 20 min.