Traffic speed is an important index to measure the traffic status, and real-time and accurate traffic speed prediction is an important part of building an intelligent transportation system. A new traffic speed prediction model based on the combination of attention mechanism and graph convolutional neural network is proposed to address the problems of randomness, nonlinearity and spatio-temporal correlation of traffic speed. Finally, the proposed model is combined with five other benchmark models to predict traffic speed on two publicly available traffic speed datasets. The experimental results show that the accuracy of the proposed model is 75.1% and 86.6% on the two datasets, which is about 3% higher than the accuracy of the advanced benchmark model. This indicates that the proposed model has high accuracy and stability, and can provide scientific basis for traffic management.