灰度纹理与油气特征融合的油纸绝缘老化状态评估

针对单一局部放电特征辨析老化信息量存在精度不足的问题，提出一种基于局部放电灰度纹理特征与油气特征的D-S证据融合识别方法，对油纸绝缘老化状态进行分析。首先，通过试验构造人工油纸绝缘内部气隙模型，模拟变压器内部实际运行环境，采集油纸绝缘7个老化层级的局部放电信号及油中气体含量，提取各老化层级的灰度纹理特征、统计特征及油气特征，利用支持向量机进行识别。将灰度纹理特征和油气特征的识别结果输入到D-S证据融合框架中进一步识别分析，识别结果与单一特征及其他两两特征融合结果比较。试验结果表明：基于灰度纹理特征与油气特征的D-S证据融合方法对油纸绝缘老化状态的识别效果更佳。

Aging State Evaluation of Oil-Paper Insulation Based on Fusion of Gray Texture and Oil Gas Features

Aiming at the problem of insufficient precision of aging information obtained by single partial discharge(PD) feature discrimination, a D-S evidence fusion recognition method based on PD gray texture and oil-gas features is proposed to analyze the aging state of oil-paper insulation. Firstly, an air gap model of artificial oil-paper insulation is constructed through experiments to simulate the actual operation environment inside the transformer. Next, the PD signals of seven aging levels of oil-paper insulation and the corresponding gas content in oil are collected. Then, the gray texture features, statistical features and oil-gas features of each aging level are extracted, and subsequently identified using the support vector machine. The identification results of gray texture features and oil-gas features are then input into the D-S evidence fusion framework for further recognition and analysis. Finally, the recognition results are compared with the single feature and other two-feature fusion results. The experimental results show that the D-S evidence fusion method based on gray texture feature and oil-gas feature can better identify the aging state of oil-paper insulation.