Application of WNN-GNN-SVM combined algorithm to time series analysis of SF6 decomposed gas signal detected by photoacoustic spectroscopy

Abstract

This paper first analyzes the principle of the photoacoustic spectroscopy detection technology, gives the absorption wavelength and the absorption line intensity of typical SF6 decomposition gas, and establishes the original database of the photoacoustic spectroscopy combined with the existing experimental platform. Furthermore, the principle of WNN-GNN-SVM combined algorithm and its construction method model are proposed, and the combined algorithm is applied to fit and predict the time series curve of sulfur hexafluoride characteristic decomposition gas with 6 peaks. The results show that the photoacoustic spectrum detection technology can effectively obtain the time series of typical sulfur hexafluoride decomposition gas, such as SO2F2, H2S, COS and CS2, and the WNN-GNN-SVM combination algorithm proposed in this paper can effectively select key points of sulfur hexafluoride decomposition gas characteristic peak for the multi spline curve quantitative fitting, and reasonably predict its change trend. In this paper, the optical detection method for typical decomposition gas of sulfur hexafluoride is effectively proposed, and the fitting and prediction post-processing method of the original data is given based on the detection results. The research results of this paper have good engineering application and theoretical research value for the detection and evaluation of the operation state of sulfur hexafluoride gas insulated power equipment.