Aiming at fault prediction for track circuit is restricted by data, the six-terminal network model is established based on the working principle and transmission line theory of ZPW-2000A track circuit to simulate and analyze three kinds of shunting malfunction signals. A combined analysis method combining variational mode decomposition and convolutional neural network is proposed. The energy spectrum feature is extracted from the original signal by variational mode decomposition, and the deep feature is extracted by convolution neural network. Then, the sensitive feature for fault prediction is obtained by weighted fusion of the two. The simulation experiments indicate that the proposed method can predict the shunting malfunction accurately and effectively, which can achieve a prediction accuracy of 99.87% and provide a new idea for the prediction of the shunting malfunction for track circuit.