Intelligent contamination level recognition algorithm for substation insulators based on XGBoost

In order to meet the accuracy requirements of insulator pollution monitoring in substations, an intelligent identification algorithm for insulator pollution levels in substations based on XGBoost is proposed. This study characterizes the pollution state by grayscale processing and segmentation of insulator images, and extracting multidimensional features such as red component, energy feature, and deficit distance. Build a detection model based on XGBoost algorithm and introduce Bayesian optimization (BO) to dynamically adjust hyperparameters. The intelligent recognition of pollution level is ultimately achieved through the probability output by the model. The experimental results show that the average recognition accuracy of the model reaches 95.4%, and the standard deviation of each fold accuracy is ± 1.2%; The iterative curve loss value stabilizes at around 0.1 at 80 seconds, which is 0.15-0.3 lower than the comparison method. Provided reliable technical support for intelligent inspection and preventive maintenance of the power grid.