Abstract The traction main circuit serves not only to supply power to the EMUs train set but also plays a crucial role in the initiation, speed regulation, and braking of the train. Therefore, diagnosing the operational status of the traction main circuit is of paramount importance. This paper initiates the modeling of the traction main circuit for the CRH2 EMUs train set, followed by an analysis of the primary components’ faults within the traction main circuit. Voltage and current signals from the transformer’s primary side, traction winding, as well as the stator current and speed signals of the traction motor, are collected. Fault characteristics are then extracted using FFT and wavelet analysis methods. Subsequently, employing the Product-based Neural Network (PNN), the transformer, converter, and motor are subjected to fault diagnosis. Experimental results validate the high accuracy of the proposed fault diagnosis method presented in this paper.