Hybrid approach for traffic flow detection using metaheuristics optimization and machine learning algorithms

Etem T., Baker M. R., Buyrukoglu S.

JOURNAL OF THE CHINESE INSTITUTE OF ENGINEERS, 2025 (SCI-Expanded, Scopus)

Özet

This paper introduces a novel hybrid framework for real-time traffic flow detection that marries advanced metaheuristic optimization with a comprehensive suite of machine learning and deep learning classifiers. The approach directly targets the challenges posed by the high-dimensional, spatial-temporally variable nature of traffic data that typically undermine conventional traffic management systems. By systematically employing a set of metaheuristic algorithms, Binary Dragonfly Algorithm (BDA), Binary Gray Wolf Optimizer (BGWO), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Whale Optimization Algorithm (WOA) show that the framework effectively reduces feature dimensionality and isolates key traffic indicators that enhance classification accuracy and computational efficiency. The study further integrates diverse classifiers, including Adaptive Boosting, Decision Tree, Random Forest, Artificial Neural Networks, Logistic Regression, Extreme Gradient Boosting, and Long Short-Term Memory networks, to rigorously compare performance and validate the hybrid design. Experimental evaluations on real-world traffic datasets demonstrate significant improvements in detection accuracy and recall, underscoring the potential of this methodology to support advanced traffic monitoring, anomaly detection, and predictive urban planning.