การวิเคราะห์ประสิทธิภาพการควบคุมสัญญาณไฟจราจรบริเวณทางแยกรินคำโดยใช้วิธีโครงข่ายประสาทเทียม

Abstract

Traffic congestion is a major problem in many cities. The management of traffic signal control system at the intersection is crucial in solving traffic problems. Traffic congestion often occurs at intersections with traffic signals during peak hours, leading to queue accumulation. Currently, the problem is addressed by traffic police officers who manually control the traffic signals. However, evaluating the efficiency of intersections becomes challenging due to the dense traffic volume. This study provides an application of the Artificial Neural Network (ANN) to simulate the police-generated traffic signal control during peak hour at Rin Kham Intersection in Chiang Mai. Specifically, the signal control efficiency is analyzed and compared with other types of traffic signal control using the PTV VISSIM. The results of a neural network model to simulate the police-generated signal control using the data of February 2023, the training model by Levenberg-Marquardt backpropagation method. The neural network structure [16-99-4] with R2 0.9400. Total delay reduced as traffic volume through the junction rose. The green time, which varies based on queue length, green time, and traffic volume, is a factor that influences police management of traffic signals. The results of the traffic simulation using police to control traffic signals compared to the vehicle actuated system traffic signal control show that the traffic model controlling traffic signals by police performed better than the vehicle actuated system traffic signal control. Although the total delay of the intersection was not much different. But there was more than 4.02% of the traffic passing through the intersection. However, the police-controlled signal control that changed according to the traffic conditions resulted in the traffic volume variability and delays