Research on Transformer Fault Diagnosis Technology based on Multivariate Correction of Fluorescence Spectrum
Jia Xie, Yue Zhao, Yumei Song, Fengxiang Ma, Feng Zhu, Xueqing Chen, Anjing Wang · 2023
The traditional insulating oil condition detection mostly adopts Dissolved Gases Analysis (DGA) technology, which faces serious shortcomings such as long detection period, inability to respond quickly in real time, and low sensitivity when the amount of dissolved gases is small or there are no dissolved gases. In order to solve the above problems, based on the change mechanism of fluorescent substances in faulty oil and the principle of multivariate calculation, a fluorescence multivariate correction transformer fault diagnosis technology is proposed, to establish and optimize the fluorescence multivariate correction model of insulating oil for qualitative and quantitative diagnosis of faults. Taking three commonly used insulating oils as research samples: Karamay oil, Nynas oil Nytro Libra (NL without additives) and Nynas oil Nytro Gemini (NG with additives). The fluorescence spectral data of the oil samples are collected and analyzed under different fault conditions, and the optimal excitation wavelengths and spectral preferred bands are determined. The threshold ranges for the rate of change of the aromatic hydrocarbon concentration are analyzed to complete the qualitative diagnosis, then to construct polynomial fitting models of the aromatic hydrocarbon concentration and fault degree, to complete the fault quantitative diagnosis. The proposed method provides theoretical basis and research foundation for realizing online realtime monitoring, and also lays the foundation for subsequent making the multivariate correction coefficients into filters, realizing the model hardware, and greatly reducing the detection cost.