In this paper, a method of aging diagnosis under different oil-paper ratios is proposed by investigating the accelerated thermal aging data with different oil-paper ratios. Aged oil-paper insulation samples with oil-paper ratios of 10:1, 15:1 and 20:1 are obtained by accelerated thermal aging tests. The Raman spectra of the transformer oil samples are measured by a Raman spectrometer, and screened by random forest algorithm to extract the features. A modified model is established to correcting the features of the aged oil samples under the conditions of 15:1 and 20:1 oil-paper ratio to features under the condition of 10:1 oil-paper ratio. An grid search optimized support vector machine (SVM) classification model is established, and verified by the corrected features. The experimental results show that the proposed method can effectively identify the aging of oil-paper insulation with different oil-paper ratios.

中文翻译：

---

不同油纸比油纸绝缘的拉曼光谱特性及老化诊断

本文通过研究不同油纸比下的加速热老化数据，提出了一种不同油纸比下的老化诊断方法。通过加速热老化试验得到油纸比为10:1、15:1和20:1的老化油纸绝缘样品。变压器油样品的拉曼光谱由拉曼光谱仪测量，并通过随机森林算法筛选以提取特征。建立修正模型将15:1和20:1油纸比条件下老化油样的特征修正为10:1油纸比条件下的特征。建立了网格搜索优化的支持向量机（SVM）分类模型，并通过修正后的特征进行验证。