An Autonomous Division Method of Traffic Signal Control Subregion Based on Complexity Balance of Cooperative Signal Control Algorithm

As the number of weighbridges increases and researchers delve deeper into regional traffic control, applying signal control schemes by dividing signal control sub-regions to reasonably manage the timing and sequence of traffic lights at various intersections within the network has become a crucial method for reducing traffic congestion. However, the evaluation index of the existing method is still mainly based on the correlation degree between intersections, and the cost of signal control is not considered. So, this paper proposes an optimized sub-region division method for balancing the complexity of DQN-based intelligent traffic control algorithms. First, the network architecture of DQN agents in the control unit is determined based on the size of the sub-region, training DQN control agents of different sizes to provide strategic foundations for subsequent implementation of control schemes. Second, the average queue length and average vehicle speed in the selected area are used as performance metrics for control, with the training time and number of parameters serving as indicators of algorithmic complexity. Finally, genetic algorithm is used to solve the problem of optimal partition results with complexity constraints, and the initial subregions are divided, and then dynamically adjusted according to the rules. Experiments show that compared to basic association-based division methods, the optimized division method can achieve better control performance while meeting complexity constraints, reducing the overall road network queue length by approximately 12%.