A Method for Discriminating the Moisture Status of OIP Bushing based on Dissado-Hill and GWO-HMM Model

An accurate discrimination on moisture status (MS) of oil-impregnated paper (OIP) bushings is crucial for the maintenance and replacement schedule of bushings. Based on frequency-domain spectroscopy (FDS) measurement and Dissado-Hill (DH) relaxation model, this paper proposes a hybrid approach of hidden Markov model and gray wolf optimization (GWO-HMM) for MS estimation of bushings subjected to the ununiform moisture distribution and dynamic time-series modeling. First, simulation models of moisture diffusion and FDS of the OIP bushing were constructed using finite element modelling (FEM) approach. Then, the GWO algorithm was employed to explore dielectric parameters influenced by moisture in DH model. Then, GWO-HMMs was further adopted as a classification tool to discriminate the MS. The GWO-HMMs was applied to estimate the MS of bushings using both simulation and experimental data. Classification results confirm that the average identification accuracies of the proposed method are 98.08% and 97.61% over these two datasets, which demonstrates the effectiveness of the proposed moisture estimate method for OIP bushings.