To meet the needs of managers and travelers to comprehensively grasp the traffic status of the expressway network, this paper employed expressway network data and toll data as the basis to build a traffic status discrimination model of the expressway network based on multi-source feature fusion, which was adopted for discriminating the traffic status of the entire expressway network. Firstly, the expressway network was divided into sections, and static features and dynamic features of the sections were extracted to comprehensively characterize road section features based on multi-source feature fusion. Then, a semi-automated data annotation method based on custom rules combined with manual annotation was utilized for data annotation to ensure the accuracy of the original data annotation. Finally, given the data imbalance, a dual random forest model consisting of two layers of single random forest models was proposed, with the second layer of the random forest model employed to improve the classification accuracy of the first layer of the random forest model. To demonstrate the effectiveness and accuracy of the proposed method, this paper validated the model by collecting expressway network data and toll data of Liaoning Province. The results show that the accuracy of the smooth status is 100.0%, the accuracy of the basic smooth status is 94.0%, and the accuracy of mild congestion is 98.0%. Additionally, the accuracy of moderate congestion is 94.0%, the accuracy of severe congestion is 97.0%, and the average accuracy of the traffic status is 96.6%.