Adjustable and weight-based adaptive traffic signal control for multiple intersections

Traffic signal management, which is considered one of the biggest problems of the 21st century; It is an important problem due to its negative effects on the environment, economy and public health. To overcome this challenge, many methods, both deterministic-based and artificial intelligence-based, have been proposed for traffic signal control at isolated and coordinated intersections in urban areas. Deterministic systems are limited in their ability to account for complex traffic conditions and variables and have limited ability to respond flexibly to traffic conditions in real time. Additionally, the manual adjustments required to manage large-scale traffic networks are time-consuming and costly. Artificial intelligence-based systems, on the other hand, cannot perform as desired if they are dependent on the training data of the model and if these data are incomplete or misleading. Maximum Weighted Flow method is one of the recommended methods to make urban traffic more fluid by optimizing traffic flow. While the Classical Maximum Weighted Flow method ensures that the flow in the network is maximum by using artificial intelligence, deterministic approaches are also used. However, the complexity and computational costs in the classical Maximum Weighted Flow method reduce the practical applicability of the method. In this thesis study, a new approach consisting of two strategies is proposed to increase the effectiveness of the classical Maximum Weighted Flow method and make the technique more applicable. The first proposed strategy simplifies the process by calculating the approach time of vehicles to the intersection according to their instantaneous speed in a certain location range. The second proposed strategy uses regression to identify the optimal vehicle waiting coefficient. The effectiveness of the proposed approach on traffic signal control is tested in the SUMO simulation environment in isolated and coordinated intersection systems using real data. Additionally, the proposed New Maximum Weighted Flow method was compared with Static, Excited, Delay-based, Webster, MaxPressure and Classical Maximum Weighted Flow methods. As a result, the proposed method outperforms existing algorithms in isolated and coordinated intersection systems.