Research on Fault Prediction of the Power Unit Motor of Scraper Conveyors Based on Improved XGBoost

Aiming at the problems of high failure rate and insufficient monitoring accuracy of the power unit motors of scraper conveyors in underground coal mines, a method for evaluating the health status of motors based on the XGBoost algorithm is proposed. A multi - source feature set is constructed by collecting vibration signals, current signals, and temperature signals. The input parameters are optimized by combining feature importance analysis, and an XGBoost health status classification model is established. Experiments show that this method is superior to traditional support vector machines (SVM) and random forest algorithms in terms of fault recognition accuracy and generalization ability, providing new ideas for intelligent operation and maintenance of underground motors.