Oil-paper insulation state diagnosis based on neighborhood rough set and clustering cloud model

In this paper, a hierarchical evaluation model of oil paper insulation state based on neighborhood rough set and cluster cloud model is proposed, aiming at the problems of unreasonable weight distribution of dielectric response evaluation of multi-feature quantity and inaccurate evaluation of system randomness. Firstly, five strong characteristic variables representing insulation state were extracted from the recovery voltage polarization spectrum and extended Debye model (EDM), and the insulation state evaluation system of oil paper was established. Secondly, the combination weighting method is used to synthesize neighborhood rough set (NRS) method and improved analytic hierarchy process (IAHP) to improve the sensitivity difference of multi-feature quantity in the reaction insulation state and avoid the loss of weight data information. Finally, the membership degree selector of the cluster cloud model is constructed by taking the randomness and fuzziness of the insulation state grade boundary into account. The example demonstrates that the method can not only accurately depict the current insulation condition of the transformer, but also indicate the trend of deterioration. This offers a new approach for the comprehensive assessment of the oil-paper insulation state.