Analog Circuit Fault Diagnosis Based on One-Dimensional ResNet

With the increase of digitization and automation level in power plants, the development of control circuit boards has become a necessary means to maintain the normal operation of the internal electrical system in power plants. Such control circuit boards can ensure the accurate operation of the power supply system. However, research on relevant maintenance and fault diagnosis is scarce. The usual maintenance method relies on planned overhaul processes, and the replaced control circuit board is directly stored, which requires a high level of technical proficiency from the operators and leads to waste. In recent years, residual neural networks have been widely used in the field of fault diagnosis. A fault diagnosis method for control circuit boards based on one-dimensional residual neural networks is proposed. The time domain and frequency domain signals of analog circuit output are represented as one-dimensional data sequences, and the features of different fault modes are learned by residual neural network, which can effectively detect and diagnose the faults of control circuit boards.