Abstract: Aiming at the problem that vibration signal features in the fault diagnosis of idler bearing of belt conveyor are extracted difficulty, which leads to low accuracy of fault diagnosis. A fault diagnosis method for idler bearings based on one-dimensional convolutional neural network (1DCNN) and extreme learning machine (ELM) is proposed. First, the collected data is separated according to the specific fault diagnosis task, the Fourier transform is performed, and the health status, fault type and damage degree are expressed by multiple labels. Then, 1DCNN is used to extract fault features, and ELM performs fault classification according to the extracted features. In this method, the parameters are randomly generated, and iterative updating is not needed, which speeds up the calculation speed. Finally, the fault diagnosis experiments were carried out through the bearing data set of Case Western Reserve University and the self-made idler fault data set. The test accuracy reached 100%, and the running time was 2.82 s and 2.42 s, respectively. It can accurately judge the type of idler failure in a short time, which verifies the effectiveness of this method. The superiority of the proposed method is demonstrated by comparing it with methods such as ELM, random forest, K-nearest neighbor, support vector machine, and convolutional neural networks. The results show that the diagnosis effect of the combination of 1DCNN and ELM is better than that of a single method, and it can meet the needs of idler fault diagnosis in the coal mine field.