Upgrades are continuously made to expressway traffic perception technology as part of the ongoing development of intelligent transportation, resulting in a large amount of data, including vehicle position, passing time, and speed. This real-time, accurate, and efficient source data is used to estimate short-term traffic flow on expressways. Aiming to enhance predictive accuracy and more accurately capture the intricate spatial and temporal features of traffic flow, this paper presents a convolutional long short-term memory neural network model based on attention mechanism to predict traffic flow. To extract spatial-temporal features rather than just single temporal features, convolutional long short-term memory neural networks combine the advantages of long short-term memory neural networks to extract temporal features and convolutional neural networks to extract spatial features. Attention mechanism is added to capture the influence of past feature states of time series data on traffic flow. On this basis, the error correction mechanism is introduced to further enhance the prediction accuracy of the model. The experimental results indicate that the proposed model outperforms other current techniques in terms of prediction accuracy.