Evaluation of Urban Traffic Flow with Neural Network Algorithms in Intelligent Traffic Control Systems

Traffic congestion in road transportation presents a significant issue for both developed and developing nations. This research leverages artificial intelligence (AI) to assess, retime, and analyze pre-timed traffic signals in the Oredo local government area of Edo State, where vehicle volume has recently surged. Traffic data, including vehicle flow, composition, and movement, was manually recorded at 15-minute intervals. Machine learning techniques, particularly an Artificial Neural Network (ANN) combined with SUMO (Simulation of Urban Mobility), were applied to optimize signal timing and reduce intersection control delay. The analysis demonstrated a 15% reduction in average control delay, from 44.4 seconds to 36.6 seconds, and improved the level of service (LOS) from 'D' to 'C' during peak hours. The ANN model achieved a low root mean squared error (RMSE) of 2.26 × 10^- 13, indicating high accuracy. The optimal model, which featured four hidden layers with 64 neurons each and was trained for 250 epochs, achieved the best validation performance of 0.01748667. These results demonstrate the effectiveness of AI-based models in improving traffic signal performance and mitigating congestion at intersections.