An optimization similarity fuzzy inference method for traffic signal control at an isolated intersection

Abstract

Managing urban traffic is challenging because traffic patterns change unpredictably. Although fuzzy logic-based traffic signal control (TSC) systems like Mamdani and Sugeno work well, they struggle to adjust effectively to real-time traffic changes. This study introduces the Optimization Similarity Fuzzy Inference (OSFI) method, which improves traffic signal control at isolated intersections by continuously adjusting fuzzy rules based on the similarity between actual and desired outcomes. Unlike traditional models, OSFI uses truth tables to dynamically adjust signal timing and phase sequencing based on real-time factors such as vehicle arrival rates and queue lengths. Simulation results show that OSFI reduces average vehicle delays by 1.11–5.73% compared to Mamdani controllers and 0.69–4.84% compared to Sugeno controllers, with traffic throughput improvements of up to 18.75% during heavy traffic. These findings demonstrate OSFI's ability to consistently improve traffic flow. Future research will focus on expanding OSFI to control networks of intersections and testing its real-world performance to address current challenges related to scalability and efficiency.