Research On Signal Control For Oversaturated State Of Urban Road Traffic Networks

With quick development ofsocical economy in China, the number of motor cars rapidly increases, resulting in serious urban traffic congestion, and then restricting social development, that makes the governments to thinking to put great research efforts on urban traffic and to improving the traffic control systems. Moreover, the developments in science and technology also admit us introducing new control theory approaches and technologies to traffic control field for solving congestion problems. Because of most of traffic signal control systems are designed for the under-saturated network, the control performance of the systems largely decreases when the network is in over-saturated state. Thus, the problem of traffic signal control for the over-saturated network is still a difficult problem in traffic communities. Then, this thesis mainly focuses on the research about traffic signal control of the over-saturated networks, main results of which are as follows:Firstly, the method of dynamic partition of control subzones of the network is provided. The weighed directed graph of the network is established with signalized intersections as vertices, links as directed edges, and the impedance as weighs, which can be algebraically represented by the weighed adjacency matrix, transformed to the usual adjacency matrix by setting weigthed thresholds of links. Furthermore, based on the K-step reachability methods of vertices for structure decomposition of the network, solving the critical vertices of the network (i.e., the critical intersections), from which control subzones are generated by determining K-step reachable subsets of vertices. The variance of the space occupancies of all links in the subzone is chosen for the evaluation of the similarity of traffic states.Secondly, the signal control design approach of consensus of traffic states in the network is proposed. Traffic signal control state-space model is established based on conservation principle of traffic flows in the network, which takes discharging propositions of links as control variables and boundary vehicle flows of the network as input variables; The analytical relation between the steady-state values of links and boundary inputs of the network is established based on nonnegative matrix theory; The set of network links is partitioned by using fuzzy cluster algorithm, and then the solution to the steady-state values of network links is given; Based on the feedback ideas in control theory, the distributed state-consensus signal control law is proposed, which can realize the balance of space distribution of network flows, and then decrease local traffic congestion and enhance network capacity.Thirdly, the coordinated signal control design method of regions is proposed. Under the base of determination of the link-queue thresholds, considering comprehensively the time discharging the current link queue, and the time filling up respectively upstream and downstream links, establishing the relevance between adjacent intersections, from which the similarity degree index of traffic states between vehicle flows is proposed, in which the similarity degree of traffic states between vehicle flows is evaluated quantitatively. Then, the approaches of dynamic phase combination and variable phase sequence are proposed; Based on traffic wave theory, and with consideration of the weighs of transits in phase vehicle flows, the coordinated fuzzy signal control approach is provided for the network, with the objective of decreasing the possibility of occurrence of overflows.Fourth, the boundary control method of control subzones is designed. The definition of the boundary of region is given, containing boundary intersections and links, respectively. Establish traffic control and guidance subsystem optimization models respectively, with the optimization indices of average delay per person and average travel time of the network, respectively; Then, the coordinated optimization objective function of traffic control and guidance is proposed, with the cycle, splits and boundary inputs of the subzone as optimization variables, which is solved by using genetic algorithm. The simulation results validate the effectiveness of the proposed control algorithms.At last, choose the real-world network topology in Beijing, the comprehensive simulation validation is conducted with applications of the dynamic partition method of subzones, the state-consensus signal control method of traffic states, and the methods of the dynamic phase combination and variable phase sequences.