Adaptive signal light timing for regional traffic optimization based on graph convolutional network empowered traffic forecasting

With the acceleration of urbanization, urban traffic congestion is becoming more and more serious, in which the timing of signal lights for regional traffic optimization is particularly important. Since existing signal lights-based traffic optimization technologies, especially green wave, do not take the regional traffic follow into consideration, therefore not being efficient. Therefore, we propose Adaptive Signal Light Timing for Regional Traffic Optimization based on Graph Convolutional Network Empowered Traffic Forecasting. First, we propose a multi-intersection traffic flow prediction model, namely, A-GCN+ with an improved prediction accuracy of 6.3%, which utilizes the attention-aggregated graph convolutional neural networks (A-GCN) and temporal convolutional networks (TCN) to extract spatial and temporal features of the traffic flow. Second, we propose a dynamic regional traffic signal coordination optimization control method, which utilizes the predicted intersection approach traffic flow from A-GCN+ and combines it with the improved whale optimization algorithm (IWOA) to obtain the optimal solution for the regional average vehicle delay model. Finally, we propose a bidirectional green wave automatic control method for the main line, which utilizes the optimized results of dynamic regional traffic signal timing and employs a multi-strategy fusion graphical method to obtain the dynamic main line bidirectional green wave. Experimental results show that compared to the traditional graphical method, the multi-strategy fusion graphical method increases the green wave bandwidth by 20%. The mainline bidirectional green wave adaptive coordinated control method improves main line traffic efficiency by 32.3% and regional network traffic efficiency by 8.7%.