In order to improve the accuracy of short-term traffic flow prediction, a short-term traffic flow prediction method based on the combined model of complementary ensemble empirical mode decomposition and gated recurrent unit is proposed. Firstly, the unstable original traffic flow time series data are decomposed into relatively stable multiple modal components by complementary ensemble empirical mode decomposition algorithm; Then, a GRU model is established for each decomposed modal component sequence for one-step prediction. Finally, the predicted value of each component is superimposed to obtain the final prediction result. Using the measured traffic flow data of Shanghai to verify and analyze, the experimental results show that CEEMD-GRU combination model is superior to GRU neural network model, EMD-GRU combination model and EEMD-GRU combination model, and the average prediction accuracy is improved by 33.4%, 25.6% and 18.3% respectively.