Detection of Decomposition Products of SF6/N2 Gas Based on Laser Spectrum Detection and AC-ISODATA-SA Fault Clustering Algorithm

Fault diagnosis method of SF6/N2 gas insulation equipment based on convolution neural network is established. five decomposition products (SO2F2, SOF2, HF, H2S, CS2) are used as characteristic input, and four typical defect types of metal protrusion, free metal particle, insulator air gap and pollution are used as output. For the identification of four typical defect types under PD: metal protrusion, free metal particle, insulator air gap and pollution, the characteristic components are selected by using the maximum correlation and minimum redundancy criterion, the weight is optimized by using Gaussian mixture model, and finally three groups of characteristic quantities (
$\text{CS}2+\text{SOF}2$
), (4SO
2
F
2
) and (
$\mathrm{H}2\mathrm{S}+ \text{SOF}2$
) are selected to construct “AC-ISODATA-SA algorithm diagnosis method” which can be used to identify four typical defect types under PD. The accuracy of the method test using test data is 92.50%. Four kinds of optical path cells made of aluminum alloy, stainless steel, PTFE and PVDF are designed for gas adsorption experiments. The analysis of the experimental results shows that there are differences in adsorption saturation time of optical path cells made of different materials, but it does not affect the accuracy of measurement results. In practical application, the adsorption of SO2F2 and SOF2 gas by cell materials can not be considered. The ICL laser and high-power QCL laser with the output center wavelength in the alternative detection band were investigated and purchased, and the tuning characteristics of the purchased laser were studied. With help of TracePro ray tracing software, the theoretical research and the simulation of Herriott multiple reflection gas absorption cell are carried out. Thermal viscous acoustic theory simulation of photoacoustic cell is carried out by using multi physical field FEM. Two photoacoustic cells are designed, and the resonance conditions, the intrinsic oscillation frequency, sound pressure, temperature and other parameters are obtained by simulation. TDLAS experimental system and photoacoustic spectrum system are built respectively, and the detection experiments of SO2F2 and SOF2 were carried out.