基于近红外光谱法的变压器全组分油中气体检测  

Gas Detection in Transformer Full Component Oil Based on Near Infrared Spectroscopy

作　　者：陈静 刘江东 徐伟[1] CHEN Jing;LIU Jiang-dong;XU Wei(State Grid Yangzhou Power Supply Company,Yangzhou 225000 China)

机构地区：[1]国网扬州供电公司,江苏扬州225000

出　　处：《自动化技术与应用》2024年第12期88-92,共5页Techniques of Automation and Applications

基　　金：国网江苏省电力有限公司科技项目资助(J2022082)。

摘　　要：为实现变压器全组分油中不同类别组分气体浓度的可靠检测,提出基于近红外光谱法的变压器全组分油中气体检测方法。首先,在光源模块利用近红外光谱仪,采集油中混合气体数据,经由数据采集模块的光电探测器及前置和锁放两种放大电路调节后,输出光声信号结果;再通过A/D转换处理该信号,形成数字数据并传送至数据处理模块的上位机中;接着,上位机利用改进极端随机混合树算法,分类混合气体的组分;最后,根据TDLAS理论计算变压器各个组分油中气体浓度结果。测试结果显示:该方法的气体检测灵敏度较高,可检测不同组分气体的最小浓度结果;决定系数指标结果均在0.950以上,可精准获取各组分气体浓度的检测结果,可为变压器运行早期隐性故障诊断提供依据。In order to realize the reliable detection of gas concentration of different kinds of components in the oil,a method based on near infrared spectroscopy for the detection of gas in the whole component oil of transformer is proposed.Firstly,in the light source module,the near-infrared spectrometer is used to collect the mixed gas data in the oil,and the photoacoustic signal results are output after adjustment by the photodetector of the data acquisition module and the amplifier circuit of the front and the lock.Secondly,the signal is processed through A/D conversion to form digital data and sent to the upper computer of the data processing module.Then,the upper computer uses the improved extreme random mixing tree algorithm to classify the components of the gas mixture.Finally,the gas concentration in transformer oil of each component is calculated according to TDLAS theory.The test results show that the gas detection sensitivity of this method is high,and the minimum concentration results of different component gases can be detected.The determination coefficient index results are all above 0.950.Accurate detection results of gas concentration of each component can provide basis for hidden fault diagnosis in the early stage of transformer operation.

关 键 词：近红外光谱法 变压器 全组分油 气体检测 吸收谱线 气体浓度结果 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TM85[自动化与计算机技术—计算机科学与技术]