MULTILEVEL INTELLIGENT FUZZY CONTROL OF OVERSATURATED URBAN TRAFFIC NETWORKS

Some drawbacks of existing automated urban traffic control systems are discussed. It is shown that most of them are based on deterministic algorithms and have a centralized structure. As a result, qualitative and quantitative complexity of transport processes are not taken into account effectively. To overcome this inconvenience, an approach for implementing distributed intelligence systems is proposed. It is based on multilevel and artificial intelligence techniques. Fuzzy sets theory is chosen as a tool for distributed intelligence control systems design and a specific method is presented. It is shown, however, that this method is not suitable for real time control of large scale systems and an alternative method, leading to a closed-loop fuzzy control law, is proposed. It is based on decomposition of the system into subsystems and presentation as a hierarchical two-level control structure. In this case, the controls have two components-local (calculated by the lower level) and global (calculated by the upper level). The method is used for computer simulation of an oversaturated urban traffic network and gives satisfactory results. Its main advantages are the reduction of the number of fuzzy relations and the effective accounting of constraints on state and control variables.