基于动态Fisher聚类算法的多时段二次划分

现有多时段划分方法大多只考虑时间维度下交通流宏观变化,即大范围趋势变化规律,未能考虑交通微观变化下的波动情况。针对这种情况,深入分析交通流的宏观和微观变化,提出了基于动态Fisher聚类算法的多时段二次划分方法。选取济南济泺路—汽车厂东路交叉口对提出的方法进行了验证,实验结果表明与传统方法相比,该方法能更好的适应实际交通流的变化趋势。

Multi-period quadratic division based on dynamic Fisher clustering algorithm

A large variety of existing multi-period division methods only consider variation of traffic flow under the time dimension macroscopically which refers to the trend change rule in large scale, while the fluctuation of traffic under microscopic variation are ignored. This paper analyzed the macroscopic and microscopic variation of traffic flow deeply and proposed a multi-period quadratic division based on dynamic Fisher clustering algorithm. It selected Shi Jing Shang-Ancient City Street intersection to verify the method proposed in this paper, of which the result had shown that this method can adapt variation trend of real traffic flow better than conventional methods.