Abstract:In view of the shortcomings of the existing online fault detection methods such as poor classification performance and low detection rate,a multi fault online detection method based on DAG-SVM was proposed.The original fault information of each component of the underground conveyor was extracted,the fault signals were denoised and normalized to get the highfrequency eigenvector.The DAG-SVM fault classification method was used to construct multiple classifiers according to the type and number of the fault eigenvector.The fault categories were accurately identified through pair comparison,and the evolution trend of fault samples was predicted.The simulation results show that the hyperplane determined by the proposed method is more reasonable,the classification accuracy of the proposed method is high,and the comprehensive online detection rate of multiple faults reaches 99.47%,which is significantly better than the existing detection methods.