Method for diagnosing defects of insulated tubular busbars based on improved RF model

In this paper, a method for diagnosing defects of the insulated tubular busbars based on the improved RF model is proposed to improve the stability of the power grid operation. With the determined parameters in the RF model waiting to be optimized, the initial RF model was established. PSO algorithm was used to iteratively update the particle speed and coordinates to obtain the optimal parameters of RF model which are the number of subtrees and the number of split features in RF model. Finally, the optimal RF identification model was established with the optimized parameters and used to diagnose the defects of the insulated tubular busbar. In this paper, experiments were conducted on a 40.5kV insulated tubular busbar with different types of defects. Three kinds of sensors (UHF, HFCT and AE) were used to collect partial discharge signals of the insulated tubular busbar. Then statistical characteristics including the average value, effective value, peak value and kurtosis were extracted from the signals and normalized. The experiment results show that the PSO-RF model can accurately identify the defect of the insulated tubular busbar, which will effectively help staff to accurately obtain the operating status and facilitate the maintenance at the first time when the failure occurs.