Enhanced short-term traffic flow forecasting using a meteorological spatio-temporal transformer with multi-head attention and whale optimization

Rapid urbanization and increasing vehicle ownership have intensified traffic congestion, creating serious social, economic, and environmental challenges. Accurate short-term traffic flow prediction is therefore essential for effective urban traffic management and congestion mitigation. While traditional models struggle with the non-linear, dynamic nature of traffic data, our approach proposes an optimized synthesis named the Meteorological Spatio-Temporal Transformer Network with Multi-Head Attention and Whale Optimization Algorithm (MSTT-MHA-WOA). This framework enhances prediction by uniquely integrating non-linear meteorological features and, critically, by using the Whale Optimization Algorithm (WOA) for autonomous hyper parameter tuning. We evaluate the model's robust and consistent performance across six diverse, real-world traffic scenarios-a crucial test of its generalization capability under complex conditions.