基于神经网络客流预测的高峰期公交时刻表优化

为了加强公交发车时刻与高峰期客流需求波动间的协调性,需要依据实时客流需求进行时刻表优化.根据IC卡采集到的上车乘客数据,分别采用BP神经网络和RBF神经网络算法预测计算得到断面客流量.兼顾优化决策和评价模型,设计完善了基于客流预测的公交时刻表动态优化流程.计算文山市公交线路客流数据,发现案例中采用RBF神经网络预测得到的断面流量精度较BP神经网络高出4.9%.基于RBF神经网络和BP神经网络预测客流需求优化的公交时刻表与现状运行时刻表相比,乘客出行成本分别降低了4.11%和1.35%,企业运营成本分别降低了7.06%和4.60%.定量验证了动态优化方法的可行性和有效性. 

An Optimization of Bus Timetable During Peak Periods Based on Forecasts Passenger Flow Using Neural Network

To enhance coordination between departure time and fluctuation of passenger flow demand, timetable of buses need to be optimized based on real-time passenger demand.Based on boarding data of passengers collected from IC card, sectional passenger flows can be predicted and calculated individually using BP neural network and RBF neural network.Based on prediction of traffic flows, the optimization decisions and an evaluation model are employed to design a dynamic optimization process of bus timetable.The data of passenger flows of bus lines in Wenshan city is used and analyzed.The results show that accuracy of sectional passenger flows from RBF neural network is higher than which of BP neural network.Comparing to the former timetable, the travel cost can be reduced by 4.11% and 1.35% based on the optimized timetable using RBF neural network and BP neural network, respectively.The operating costs of enterprises can be reduced 7.06% and 4.60%, respectively.The feasibility and effectiveness of the proposed dynamic optimization method is verified.