The Research Of Photoacoustic Spectroscopy Detection For Sf6decomposition Characteristic Components Under Partial Discharge

SF₆has excellent insulating and arc-quenching performance, Gas InsulatedSwitchgear （hereinafter referred to as GIS） which uses SF₆to insulating medium hasmany advantages which include high insulation strength、stable operation、covers lessarea、less maintenance workload, so it apply into the power system more and morewidely. However, GIS equipment will inevitably has some insulation fault which canlead to the generation of partial discharge （hereinafter referred to as PD）.PD cancause SF₆to decompose, it will generate such as SO₂、CO₂、CF₄、SOF2、SO₂F2、SOF₄and HF etc characteristic gas under the action of micro oxygen and water. We canjudge fault cause、discharge level and dangerous level of the PD through checking thefeatures gas content and changing rule,it has become hot spot of the domestic andforeign research. The gas photoacoustic spectroscopy has a good application prospectin SF₆decomposition components detection because of the advantages which includehigh detection sensitivity、wide detection range、not needing measured gas and carriergas、being especially suitable for online monitoring.The paper deeply analyses hot produce and stimulation of sound field in gasphotoacoustic spectroscopy detection based on gas photoacoustic spectroscopydetection principle and gas molecular IR theory. It provides basis for the theoryanalysis of photoacoustic detection results.The paper gives the feature spectrum ofSO₂、 CO₂、 CF₄by SF₆decomposing.Using the photoacoustic spectroscopy device,we got the relation lines between photoacoustic spectroscopy （PAS） signal andconcentration of gas components and the minimum detection limits of SO₂、CO₂、CF₄.At the same time, the paper got photoacoustic signal temperature property of threekinds of single component and cross response property of mixed components of SO₂、CO₂and CF₄. This paper made theory analysis to photoacoustic signal detectionaccuracy of trace gases components which was influenced by temperature changes.The research showed that the photoacoustic signal of SO₂（7.35μm）、CO₂（4.26μm）、CF₄（7.78μm） had negative temperature property and the cross response ratio of mixedgas had positive temperature property in range of the experiment temperature.Therefore, this paper put forward one kind of temperature linear correction model andalgorithm which can be used in photoacoustic spectroscopy detection of trace gases,and corrected the photoacoustic signal influenced by temperature of SO₂、CO₂、 CF₄.The maximum error between correction result and standard photoacoustic signalunder reference temperature was less than4%.It laid feasible temperature correctionfoundation for manufacture SF₆on-line monitoring device.According to it exists overlap that the infrared absorption band of some SF₆decomposition components under PD, and it would appear serious crossover responsewhen use Photoacoustic technology to detect these components. Therefore, this papercombines Principal Components Analysis （hereinafter referred to as PCA） with RadialBasis Function （hereinafter referred to as RBF） neural network so as to be used in theoutput signal array processing of Photoacoustic detection. The purpose of this methodis to improve the decline of detection precision of RBF when the input space existscorrelation. It achieves accurate detection of concentration of SO₂、CO₂、CF₄mixed gas.The results show that PCA-RBF neural network is effective in eliminating correlationbetween sample and sample, the highest average detection error down to about3%.Itprovides effective method of data processing so as to make Photoacoustic spectroscopyto be used in the detection of SF₆decomposition components under partial discharge.Finally, the paper gives performance evaluation and application prospect ofphotoacoustic method through the comparison with GC and detector tube method.