Traffic Flow Prediction Using Deep Learning Techniques in Urban Road Networks

Abstract Predicting traffic flow with high accuracy is crucial for enhancing the performance of urban road networks, alleviating congestion, and boosting transportation efficiency. This study investigates advanced deep learning approaches, such as Long Short-Term Memory (LSTM), Graph Neural Networks (GNN), and Transformer-based architectures, to predict traffic flow in urban environments. By utilizing historical traffic records, weather information, and live sensor data, the models effectively learn and represent intricate spatial-temporal relationships in traffic patterns. Experiments on benchmark datasets show that deep learning models surpass traditional statistical approaches in prediction accuracy and scalability. These results underscore the capability of advanced neural network architectures to deliver valuable insights for smart city traffic management.