Locating the Oil Leakage on Power Equipment via Ultraviolet-Induced Hyperspectral Imaging Technology

Abstract

Oil leakage will cause abnormal temperature rise and discharge of oil-filled power equipment, which seriously threatens insulation reliability. In this paper, based on the synchronous fluorescence spectra (SFS) and excitation emission matrix (EEM) of mineral and vegetable oils, the ultraviolet-induced hyperspectral imaging (UV-induced HSI) measurement system is developed. Then, the fluorescence-reflectance spectra of oil traces and non-oil areas are determined via the proposed measurement system. Next, the influence factors of fluorescence-reflectance intensity, including UV source light power, detection angle degree and detection distance are discussed. By analyzing the fluorescence-reflectance intensity differences in fluorescence bands, principal component analysis (PCA) is applied for date dimension reduction and anomaly detection to realize oil leakage trace visualization. Finally, the validation test results prove that this method exhibits a general applicability for tracing mineral and vegetable oil leakage traces on different surface materials in indoor and outdoor conditions. This approach will improve the detection rate of oil leakage defects, and promote the intelligent level of operation and maintenance of power equipment.