Abstract: Based on the technical framework of Internet of Things, an embedded intelligent monitoring system for on-load tap-changer(OLTC) is designed. The system mainly extracts feature information from the vibration signal of OLTC, and performs online monitoring and fault diagnosis on the mechanical operation status of OLTC, so as to discover hidden faults in time. The system adopts an embedded system to replace PC, which is small in size, low in cost, low in power consumption, and easy to implement on-site. The system preprocesses the vibration signal in segments, which can compress the original data and highlight the state characteristics. The FFT-based equal-integral-bandwidth method is designed to extract features, which makes the data processing concise and can effectively solve the problem of insufficient computing power in the edge computing of embedded systems. The experimental results show that the system has high fault diagnosis efficiency and accuracy, and has good application prospects.