论文部分内容阅读
提出一种将再励学习与遗传算法相结合的遗传再励学习方法对交通信号进行自组织控制。再励学习是针对每一个道路交叉口交通流的优化，修正每个信号灯周期的绿性比；而遗传算法产生局部学习过程的全局优化标准，即是修正信号灯周期的大小。这种方法克服了现有的控制方法需要大量数据传输通讯、准确的交通模型等缺陷，将局部优化和全局优化统一起来。通过计算机仿真实验表明了方法的有效性。
A self-organization control of traffic signal is proposed by using a genetic re-learning method combining re-energizing learning and genetic algorithm. Reinforcement learning is aiming at the optimization of traffic flow at each road junction and correcting the green ratio of each signal light cycle. The global optimization criterion of local learning process generated by genetic algorithm is to correct the size of signal light cycle. This method overcomes the defects that the existing control method needs a large amount of data transmission and communication, an accurate traffic model and the like, and unifies the local optimization and the global optimization. Computer simulation shows that the method is effective.