Transformer Oil Chromatography Fault Detection System Based on Raman Technology

In this paper, Raman spectrometer is used to analyze the oil sample of transformer. By optimizing the procedures of sample collection and processing and spectral data collection and processing, a model that can effectively diagnose faults is established. In terms of methods, the automatic sampler is used to collect oil samples at first, and at the same time, the pretreatment step is used to ensure the representativeness and purity of the samples. Then, high-resolution Raman spectrometer is used to collect spectral data, and the quality of spectral data is improved by background subtraction and noise filtering. When building the fault diagnosis model, support vector machine (SVM) is selected as the main classification algorithm, and compared with random forest (RF) and gradient propulsion machine (GBM). The experimental results show that SVM effectively performs the task of oil chromatography fault diagnosis, the accuracy rate is as high as 99.99%, the detection time is between 501ms and 646ms, and the detection delay is less than 106ms, which is better than RF and GBM. These results show that SVM is an effective classifier and has the ability to quickly respond to transformer oil fault detection tasks.