Abstract: To address the issues of insufficient real-time performance and complex working condition interference in partial discharge (PD) monitoring of power transformers, this study proposes an online monitoring system based on FPGA and ARM architecture. In hardware design, FPGA is used to achieve high-speed signal acquisition and adaptive filtering preprocessing, and ARM processors are combined to complete data storage, feature extraction, and pattern recognition. This device preprocesses the high-frequency signal of the transformer before collection, effectively reducing the noise interference of the signal. Adopting modular design, the FPGA core board is responsible for high-speed data processing, while the ARM core board is responsible for system coordination, scheduling, and data transmission and reception.