通过城市轨道交通发展过程阶段变化对其相应的客流变化特点进行分析；基于变点发掘的方法划分出轨道交通客流的不同变化模式，并分析导致客流模式变化的影响因素；通过权重变化有效地结合不同交通状态模式下的变参数自回归求和滑动平均(ARIMA)模型和全局BP神经网络模型，构建城市轨道交通客流组合预测模型；结合某大城市轨道交通线路的实际客流数据，对本文模型的适用性和准确性进行验证。研究结果表明：同一模式区间预测比全局搜索预测更容易获得较高精度预测值，组合模型预测结果相对优于单一模型预测结果。%According to the urban rail transit development changes, and characteristics of their respective changes in passenger flow were analyzed. Based on the method of the change-point detection, the variation of the passenger flow in rail transit was classified, and the influence factors on the variation of the passenger flow pattern were analyzed. Changing the weights of variable parameters ARIMA model in the interval of different modes and the BP neural network model, a combined forecasting model of urban transit passenger flow during the network operation was proposed. The validity and accuracy of this model was proved by utilizing the real data of urban rail transit. The results show that prediction accuracy of the same pattern is higher than global searching, and combined model prediction results are better than the single model’s.
展开▼
