Research On OD Matrix Estimation & Signal Phase Optimization Under Mixed Traffic Condition Based On Genetic Algorithm

Intelligent transportation system(ITS) applies advanced technologies such as information, computer, operational research and artificial intelligence, etc, in transportation and service control to form an effective transportation system.O-D matrix estimation and signal phase optimization under mixed traffic condition are two of the important optimization problems in the field of ITS. The entropy maximizing (EM) model is a main approach to solve the O-D matrix estimation problem, but the current algorithms solving the EM model are limited by the difficulty of choosing a proper initial solution for searching procedure. According to the current road and traffic conditions, signal control in China is mainly timing control. The current optimization algorithms on traffic lights mainly do not belong to global optimization.Genetic algorithm (GA) is a global random search algorithm, which uses the idea of natural select and natural heredity in living nature for reference, and has advantages while solving the complicated problems which are large space and global search, etc.A mixed evolutionary algorithm (MEA) integrating GA and Levenberg-Marquardt algorithm(LMA) is presented here to solve the EM model. The numerical results show that MEA performs much better than GA and LMA in solving the EM model.According to the situation of China that there are a considerable number of bicycles on the roads, signal phase optimization algorithm under mixed traffic condition is presented. As there are many constricts on signal phase, two methods, namely phase merge optimization based on question drive and improved cooperative co-evolutionary genetic algorithm are designed to search the solution space. Single intersection traffic lights optimization software is designed and implemented, and the output of the software is analysed. In addition, this software has been embedded into the intersection simulation software to help the users design the phase sequence and the corresponding distribution of the traffic lights.The thesis is arranged as follows:In Chapter 1, two of the optimization problems in the field of ITS, namely O-D matrix estimation and traffic lights control, and one optimization algorithm, namely genetic algorithm are introduced. In Chapter 2, basic principle and mathematic description of O-D matrix estimation are described, and then a mixed evolutionary algorithm is presented to solve the O-D matrix estimation problem. Finally, thealgorithm is tested by the simulation experiment and the numerical results show its efficiency. In Chapter 3, the design method, the objective function and the solution space of the signal phase optimization algorithm under mixed traffic condition are described. The constricts of the signal phase and the phase design rules are summarized. In Chapter 4, two methods, namely phase merge optimization algorithm based on question drive and improved cooperative co-evolutionary genetic algorithm are presented to search the solution space. Finally, signal phase distribution is introduced. In Chapter 5, the function, interface, data structure and procedure of single intersection traffic lights optimization software are described, and the output of the software is analyzed. In addition, this software has been embedded into the intersection simulation software to help the users design the phase sequence and the corresponding distribution of the traffic lights. And finally, Chapter 6 is a conclusion with the summary of the thesis and prospect research.