This paper aims at assessing the effect of traffic congestion arising from urbanization of cities, which densifies the traffic flow, and affects productivity, fuel usage, and the environment. In this paper, an understanding of the Intelligent Traffic Congestion Control System (FITCCS-VN) for Vietnamese smart cities is introduced based on ML techniques. The actual traffic information received by the system through IoV and the utilization of VN help the system have better predictions in regard to congestion. The system combines two machine learning models: Here, the two machine learning algorithms of interest are artificial neural networks (ANN) and support vector machines (SVM). Combination of these models allows enriched prediction results and real-time decision making by using advantages of both approaches. The efficiency of the proposed FITCCS-VN system was assessed based on 2,22 traffic instances where 70% were for training and 30% for validation. Cross entropy yielded a prediction accuracy of 95%, with a miss rate of 5%, a higher level of performance than methods of prior work, such as, random forest classifiers and CNN-based model. The paper also shows that integrating multiple ML approaches yields a far higher traffic efficiency and low congestion than basic approaches. For example, future developments of the system can include the incorporation of improved deep learning algorithms as well as an investigation of the use of cloud processing for large data sets. This research focuses the ability of novel fusion-based ML techniques to transform traffic management systems in smart city applications.