Abstract: In order to improve the efficiency of underground transportation, this paper proposes a method combining wavelet packet decomposition and support vector machine to carry out on-line monitoring and fault diagnosis of belt conveyor. Because the number of rollers may be large, in order to reduce the number of sensors, an acceleration sensor is used to collect the vibration signals of multiple rollers. The vibration signal is decomposed by wavelet packet decomposition, and the energy of each frequency band is extracted as the feature. Then, these features are used to train the support vector machine to realize the detection of the roller fault. Firstly, the proposed fault diagnosis method is tested on the test bench, and then the on-line monitoring and fault diagnosis system of belt conveyor is designed. Through the field test, it is found that the position of the fault roller can be located under a limited number of sensors, which is of great significance to the operation of the actual belt conveyor.