Detection of SF6 decomposition gases using ultraviolet absorption spectroscopy based on Markov transition fields

Partial discharge caused by insulation faults in gas-insulated switchgear can lead to the decomposition of the insulating medium， SF6 gas. The byproducts of this decomposition， fluorosulfides， can react with moisture and oxygen in the equipment， generating compounds like SO2， CS2， and H2S. The concentrations of these three characteristic gases can reflect the type and severity of insulation faults in the equipment. To address this， the paper proposes a method based on Markov transfer fields （MTF） and deep learning to analyze the ultraviolet spectra of these characteristic gases and identify trace gas mixtures with overlapping absorption peaks. This method first converts one-dimensional sequence data into two-dimensional images， presenting the ultraviolet spectral data of the mixed gases in a more intuitive image format. Subsequently， these images are input into a deep learning model for classification and recognition. Experimental results demonstrate that this method achieves a higher recognition accuracy compared to other models.