Adaptive Traffic Signal Control Using a Multi-Tier Analytical and Fuzzy Logic Approach

Abstract

Urban intersections in rapidly growing cities frequently experience persistent oversaturation, resulting in excessive delay and queue growth. This study develops and evaluates an adaptive Mamdani fuzzy logic traffic signal controller for oversaturated urban intersections. The controller uses arrival flow and queue length to determine bounded green-time extension through nonlinear fuzzy control mapping. The methodology integrates HCM 2022 deterministic analysis, SIDRA Intersection independent assessment, and closed-loop stochastic simulation in MATLAB/Simulink. The fuzzy controller is mathematically formulated using min-max inference and centroid defuzzification, with boundedness and continuity established analytically and congestion-responsive behaviour assessed numerically. Genetic algorithm (GA) optimisation is used to calibrate the membership function parameters. Five independent GA runs are performed using 800 training samples, while 200 independent samples are withheld for validation. Traffic control performance is subsequently evaluated using 30 Monte Carlo replications. Compared with fixed-time control, the adaptive fuzzy controller achieves more than a 35% reduction in average control delay, together with reductions in queue-related measures and improved throughput characteristics. The results demonstrate that the proposed framework combines interpretable fuzzy control, traffic engineering validation, stochastic evaluation, and multi-run parameter optimisation. The findings indicate that the evaluated adaptive fuzzy controller can improve intersection performance under the tested oversaturated conditions.