A roller monitoring method based on a distributed fiber-optic acoustic sensor (DAS) is investigated to identify roller faults in a belt conveyor. Based on the principle of using a DAS to detect the fault vibration characteristic signal of an idler, a belt conveyor test system with a belt length of 30 m is constructed to simulate different fault conditions, such as idler jamming, no bearing, and fracture for experimental data collection. By analyzing the collected sound signals in the time and frequency domains, the characteristic quantities under different faults are extracted to identify the faults in the idler. The experimental results show that the DAS can detect the fault vibration characteristics of the belt conveyor idler and has application prospect in the fault monitoring of industrial belt conveyor idler.