中文摘要：

笔者采用数字图像处理算法提取了紫外图像中的放电光斑区域并定义了“光斑面积”参数，试验研究了10、35、110kv复合绝缘子高压侧电晕放电时的紫外图像和放电量的变化特性，得到了光斑面积与视在放电量之间的关系曲线。获得了光斑面积随仪器增益和观测距离变化时的试验数据并利用MATLAB的CFFOOL工具箱函数进行了拟合分析．研究表明光斑面积与增益之间近似满足于指数变化规律，光斑面积与距离之间近似满足于幂函数变化规律。在此基础上论文给出了50％、60％、70％、80％4种增益，不同放电强度下的光斑面积随距离的变化曲线．利用最小二乘支持向量机（least squares support vector machine, LS-SVM）回归算法建立了放电量强度预测模型．该模型具有良好的预测精度。

英文摘要：

In this paper, the discharge facular region in the ultraviolet（UV） images was extract with digital image processing algorithm, and the facular area parameter was defined, in the high voltage laboratory, the corona discharge character at the high voltage side of 10 kV, 35 kV and 110 kV composite insulator were studied, the relation curves of facular area and apparent discharge was obtained. The corresponding experiment data was fitted with the CFTOOL function toolbox in MATLAB, research showed that the facular area to gain had approximate exponential relationship, and the facular area to the observe distance had approximate power function relationship. On that basis, this paper established the curves of facular area with the distance under different discharge intensity with the gain was 50% ,60% ,70% and 80%, then a discharge quantification estimation model is established with the LS-SVM （least squares support vector machine, LS-SVM） regression algorithm, testing showed that it has good prediction accuracy.