Research on TOD Based on Isomap and K-means Clustering Algorithm

Urban traffic system is a complex, changeable, non-linear, unstructured and random system with spatial-temporal variations. Time of day (TOD) control is one of the important control methods in urban traffic signal-control system. The main problem of running time of day controller is to program the traffic intervals rationally. However, the traditional means, artificial method has some shortcomings. Based on manifold learning and k-means algorithm, a new method was put forward to ascertain the timing periods. Experiments show that the traffic-flow data´s intrinsic dimensions could be discovered by the Isomap algorithm, and the redundant information could be eliminated effectively. K-means algorithm could be used to part the time periods, and to ascertain the lengths of traffic interval satisfactorily at the same time. The new algorithm developed gets its advantages over the traditional one.