中文摘要：

为建立基于SF6分解特征组分原理的电气设备在线监测与故障诊断理论和方法，用针．板放电模型模拟了金属突出物绝缘缺陷产生局部放电（PD），使含不同微量H2O的SF6／H2O混合气体发生分解，定量测定了CF4、CO2、SO2F2和SOF24种特征分解组分的体积分数，得到了各特征组分体积分数、相对产气速率和能量特征比值φ（SOF）2/φ（SO2F2）随微量H2O体积分数φ（H2O）的变化规律，初步了研究φ（H2O）对各特征组分生成规律的影响机制。结果显示，φ（H2O）的增加会抑制CF4的生成，而对CO2的绝对产量和相对产气率的影响均无明显规律；SO2F2和SOF2的累积体积分数均与φ（H2O）正相关，并大致呈线性增长趋势；此外，在放电初期，SO2F2与SOF2的相对产气率均大致随着φ（H2O）的增大而减小，但在放电稳定期，却独立于φ（H2O）且趋于相同；φ（H2O）对特征组分产量比值φ（SOF）2／φ（SO2F2）也有显著的影响。实验结果为提出准确的基于SF6分解特征组分原理的在线监测与故障诊断方法提供参考。

英文摘要：

In order to establish the theory and method for on-line electrical monitoring and fault diagnosis of electric equipment based on SF6 decomposition characteristics, we conducted SF6 decomposition experiments with SF6 mixed with different H2O trace, in which a model of needle-plate electrodes was used to simulate metal protraction insulation fault and generates partial discharge （PD）. We measured the concentration of four main characteristic components, i.e. CF4, CO2, SO2F2 as well as SOF2, and analyzed the influence of H2O volume fraction φ（H2O） on the proportions and rela- tive production rates of these components, as well as the influence on a feature concentration ratio φ（SOF）2/φ（SO2F2）. Besides, we preliminarily studied the mechanism of φ（H2O） influencing the characteristic components＇ formation. The results show that increasingφ（H2O） reduces CF4 generation, but has no obvious effect on the absolute yield or the relative production rate of CO2. There is an almost linear positive relationship between φ（H2O） and the cumulative volumes of SO2F2 and SOF2; plus the relative production rates of SO2F2 and SOF2 roughly reduce with increasing φ（H2O） at the be- ginning of a discharge, but independent of φ（H2O）when the discharge is stable. It is also found that φ（H2O） influences φ（SOF）2/φ（SO2F2） significantly. The conclusions could be basis for providing accurate on-monitoring and fault diagnosis methods based on SF6 decomposition characteristics.