Accurately predicting line dynamic thermal rating (DTR) is crucial for selecting conductors in AC-to-DC transmission lines. Achieving this prediction relies on effectively reducing the fluctuation characteristics of DTR data. In this paper, the SSA-VMD algorithm is used to decompose DTR data. Then, the decomposition components of DTR data are accurately predicted using the time series feature extraction ability of the time convolutional network (TCN) model to realize the accurate prediction of DTR data. Finally, this paper uses a 110kV overhead transmission line as an example to validate the DTR data. The results show that the prediction accuracy of the proposed method is significantly improved compared with other DTR data prediction methods.