A Classifier of Railway Power Supply Equipment Concern Importance Based on Ensemble Learning

Abstract

Abstract It is of great significance to evaluate the condition of equipment and identify the degree of important concern of equipment for formulating reasonable condition maintenance strategy and ensuring the safe and stable operation of railway power grid. In view of the characteristics that the overload operation state and maintenance interval of equipment are closely related to the equipment state level, the evaluation feature vector composed of five indexes, including overload time of heavy overload equipment, whether the central power supply equipment is in the city center, unrepaired time of equipment, whether it is the power supply equipment and the voltage level of equipment is maintained, is constructed in this subject. In addition, based on the Stacking ensemble learning framework, a hybrid ensemble learning framework algorithm is proposed, which combines the single Bagging and Boosting framework algorithms, and realizes the mapping between feature vectors and importance levels. The data of actual railway power grid equipment are substituted into the model for analysis, and the model is compared with the random forest model based on single ensemble learning algorithm. The simulation results show that compared with other ensemble learning algorithms, the accuracy of equipment state evaluation is significantly improved by using the combination model of stacking ensemble learning method.