This study addresses the critical need for effective traffic signal control in reducing traffic congestion, enhancing road safety, and minimizing environmental impacts. It proposes a real-time dynamic traffic signal control system that incorporates multi-agent systems and considers vehicle type differences. Through comprehensive evaluation experiments, the method proves highly effective in reducing delay time. Experiment 1 optimizes the split control method parameter α in a straightforward setting, while Experiment 2 extends these findings to a complex 4 × 4 intersection, emphasizing the impact of vehicle type differences. Experiment 3 assesses physical decentralization for multiple intersections, showcasing effective split calculations despite occasional execution time fluctuations. The results demonstrate the proposed method’s substantial advantages over static control and methods neglecting vehicle type, exhibiting a notable 5.8% reduction in average delay time and a 12.9% decrease in the average number of waiting times. Despite occasional execution time fluctuations in Experiment 3, the method underscores its adaptability to diverse traffic conditions and configurations, offering promising prospects for advancing traffic signal control strategies.