摘要
给出了一种单交叉口混杂交通控制器和仿真结果。首先,采用Q学习和BP神经元网络根据环境的变化决定最优的相位切换时间,然后增加一个模糊控制器决定最需要切换的相位,即决定相位次序。该方法在PARAMICS交通仿真软件中进行仿真,和定控制以及定相序控制相比,该方法具有明显的优势。
A hybrid traffic controller for an intersection and its simulation results was proposed. Firstly, a method based on Q-learning and BP neural networks was introduced to determine the optimal switching time of a certain phase in order to adapt to the varying traffic condition. Furthermore, the hybrid controller, which includes an added fuzzy controller, was provided to select the optimal phase sequence, namely, choose the optimal and the urgent phase to be switched. The performance of the system was evaluated by PARAMICS traffic simulation software, which is intuitionistic and visual. Compared with conventional fixed-time control, the validity of the method was proved. At the same time,comparing the method of fixed phase sequence with the policy of varying phase sequence, the efficiency of the hybrid control of traffic signal was indicated.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第10期2889-2894,共6页
Journal of System Simulation
基金
北京市基金(4042006)
北京工业大学校青年基金(97002011200401,97002011200501)。
