基于EFPI传感器的SF₆/N₂混合气体GIS局放模式识别研究

Research on Partial Discharge Pattern Recognition of SF₆/N₂ Mixed Gas GIS Based on EFPI Sensor

摘要: 非本征法布里帕罗干涉(Extrinsic Fabry-Perot Interferometer, EFPI)的光纤超声传感器具有灵敏度高、抗干扰能力强等优点，将其应用于以新型SF₆/N₂混合气体作为绝缘介质的GIS内部局部放电的模式识别中，这对于GIS绝缘状态评估具有重要意义。分别在充有0.4 MPa纯SF₆气体、20% SF₆/80% N₂混合气体、纯N₂气体的GIS腔体内设置了4种放电模型，在单次超声脉冲波形特征参数提取的基础上，利用支持向量机算法(Support Vector Machine, SVM)对局放超声信号进行了模式识别，三种绝缘气体环境下的识别正确率均能达到85%以上。由于SF₆对超声波的强吸收效应，随着SF₆气体浓度的升高，超声波的衰减更为严重，不同类型局部放电的识别正确率也随之降低。

Abstract: Extrinsic Fabry-Perot interferometer (EFPI) optical fiber ultrasonic sensor has the advantages of high sensitivity and strong anti-interference ability. The application of it to the pattern recognition of partial discharge in the GIS with the new SF₆/N₂ mixed gas as the insulating medium is of great significance for the evaluation of the insulation state of the GIS. Four discharge models are set up in the GIS cavity filled with 0.4 MPa pure SF₆ gas, 20% SF₆/80% N₂ mixed gas, and pure N₂ gas respectively. Based on the extraction of the characteristic parameters of the single ultrasonic pulse waveform, support vector machine (SVM) was used to identify the partial discharge ultrasonic signal, and the recognition accuracy rate can reach 85% in three insulating gas environments. Due to the strong absorption effect of SF₆ on ultrasonic waves, as the concentration of SF₆ gas increases, the attenuation of ultrasonic waves becomes more serious, and the recognition accuracy of different types of partial discharges also decreases.