Study On Transformer Dissolved Gas-in-oil Analysis Technology Based On Near-infrared Photoacoustic Spectroscopy

A continuous online system is necessary to monitor the health status of oil immersion transformers in order to ensure the power equipment operates reliably. Dissolved gas-in-oil analysis (DGA) for transformers is the most effective way to prevent malfunction and diagnose faults of transformers. The multi-gas photoacoustic spectroscopy system, which combined with wavelength scanning and frequency modulation technique, consists of near-infrared tunable fiber laser and resonant photoacoustic cell. This system has achieved to measure four transformer fault characteristic gases (C2H2, H2O, CO and CO2) at the same time, and the limit of detection sensitivity can reach4ppb,19ppm,39ppm and34ppm respectively.Two methods of gas separation are applied to the process of multi-gas photoacoustic spectroscopy system used in the transformer DGA technology. A ceramic membrane tube, whose mechanical properties and penetration effect are both excellent, is used as a gas separation device first. And then the traditional headspace method is improved. A comparison between them shows that both two methods have obvious advantages against the organic film method for each equilibrium time is about3and4hours respectively.A transformer DGA detection platform is established on the photoacoustic spectroscopy system just mentioned. The platform consists of high-voltage discharge device, gas separation device form oil, cycle devices for oil and air and control systems. The system can produce C2H2through simulate discharging by this platform, and then make online monitoring of the gas. In order to verify the feasibility of the experiment platform, standard gases whose concentration is known are flown into oil to simulate the process of releasing gases from transformer faults. Finally, the result turned to be well. It also shows the advantages and bright futures of the photoacoustic spectroscopy that applied to the transformer DGA monitoring, and realizes a real sense of continuous online measurements with a high sensitivity.