State Diagnosis of Composite Post Insulator Surface Based on the Clustering Analysis

With the development of the power system, composite post insulators are widely used in China. However, the contamination flashover of post insulators caused by the loss of the hydrophobicity of silicone rubber remains a threat to the grid, especially in EHV and UHV substation. The statue variations of the composite post insulator surface are studied in this paper. Based on the cluster analysis method, the characteristics of the leakage current was also investigated, and a continuous Hopfield neural network threshold optimization model is constructed to select the class-space distance threshold. Sixteen parameters are extracted to characterize the surface state of the composite post insulator. The discharge model spectrum diagram of composite post insulator surface is also established, aiming for the diagnosis of the degree of hydrophobicity loss and salt deposit density over the insulator surface. The results show that when the hydrophobic performance is deteriorated, the proportion of weak discharges is slightly reduced, and the intermittent arc discharges are partially converted into continuous arc discharges. Besides, when the pollution degree is increased, the weak arcs were partially converted into intermittent arc discharges, and the proportion of intermittent arc discharge increased. Therefore, it is indicated that these parameters could be used to recognize the degree of hydrophobicity loss over the surface of the composite post insulator. This work can enrich the recognition methods of surface statue of composite post insulators and may provide useful reference for performance evaluation of composite post insulators in service.