Diagnosis of the aging stage of oil-paper insulation is an important technical means for preventing aging failure of oil-paper insulation equipment. This article proposes a Raman spectroscopy diagnostic model to assess the aging status of transformer oilpaper insulation by analyzing the Raman spectra of the insulating oil. Oil-paper samples in different aging stages were obtained from electrical and thermal accelerated aging tests in the lab. The Raman spectra of insulating oil were collected in the range from 400 to 3000 cm−1, which covers the Raman signals of the aging products of oilpaper insulation. Spectral peak contour features were extracted based on Voigt fitting of the Raman spectra and studied using the grid search optimized one-against-one multi-class support vector machine (OAO MCSVM) for diagnosis of the aging stage. The results show that the diagnostic accuracy of the diagnostic model based on the spectral peak contour features was 92% (46/50). Spectral peak contour features demonstrate an ability to reflect the aging process, and the proposed diagnostic method shows potential for monitoring oil-paper insulation.

中文翻译：

---

通过拉曼光谱特性识别变压器油纸绝缘老化阶段

油纸绝缘老化阶段的诊断是防止油纸绝缘设备老化失效的重要技术手段。本文提出了一种拉曼光谱诊断模型，通过分析绝缘油的拉曼光谱来评估变压器油纸绝缘的老化状态。不同老化阶段的油纸样品是通过实验室的电和热加速老化试验获得的。绝缘油的拉曼光谱采集范围为400~3000 cm-1，涵盖了油纸绝缘老化产物的拉曼信号。基于拉曼光谱的 Voigt 拟合提取光谱峰值轮廓特征，并使用网格搜索优化的一对一多类支持向量机 (OAO MCSVM) 研究老化阶段的诊断。结果表明，基于光谱峰轮廓特征的诊断模型的诊断准确率为92%（46/50）。光谱峰值轮廓特征表明能够反映老化过程，并且所提出的诊断方法显示出监测油纸绝缘的潜力。