Long-term traffic forecasting in smart traffic management: a Taiwan case study based on the informer architecture for highway traffic analysis

This study introduces an innovative traffic speed prediction framework for Taiwan's highways using the Informer model, a Transformer-based deep learning architecture designed for long-sequence time series forecasting. This work is novel in applying the Informer model to highway traffic management in Taiwan, capturing complex traffic patterns and long-term dependencies. Addressing the challenges posed by rapid urbanization and increasing traffic demand, the model offers a new solution to congestion management and road safety enhancement. By effectively analysing historical traffic speed and flow data, it provides accurate and actionable predictions for real-time traffic management. The model's robustness across diverse traffic conditions and peak hours enables highway authorities to optimize lane usage, speed limits, and control strategies, leading to reduced congestion and improved safety. Experimental results confirm the model’s high accuracy compared to actual data, positioning it as a promising tool for advancing smart traffic systems in Taiwan. This approach offers potential for broader application across other road networks, paving the way for intelligent traffic solutions.