Real-time application of grey system theory in intelligent traffic signal optimization.

This article focuses on the application of Grey System Theory (GST) in Intelligent Traffic Signal Optimization (ITSO) to enhance real-time traffic management. It introduces a model combining GST with deep reinforcement learning algorithms—specifically, the Deep Q-Network (DQN) and Proximal Policy Optimization (PPO)—to dynamically adjust traffic signal timings based on real-time data. Experimental results from S City demonstrate that applying GST significantly reduced the average weekly traffic congestion index from 60.1% to 26.6%, increased average vehicle speeds by about 20 km/h, and decreased pedestrian waiting times at traffic lights. Additionally, GST showed superior robustness in handling outliers, noise, missing data, and nonlinearities, as well as faster data processing and higher real-time prediction accuracy compared to genetic algorithm and simulated annealing methods. The study concludes that GST-based ITSO can effectively improve urban traffic efficiency and suggests further research to enhance the model and integrate it with other intelligent transportation technologies.