求解交通出行预测的反馈四阶段法研究

Study on four step procedure with feedback solving travel forecasting problems

四阶段法为经典的交通出行预测方法,并广泛应用于实际工程中。传统四阶段法将各个阶段独立建模求解,由于其没有考虑各阶段间的相互影响,所以得到的数据缺乏一致性。在传统四阶段法中引入反馈可以将不同阶段紧密结合,保持各阶段数据的一致性,同时可以避免组合模型的复杂建模与求解。介绍了反馈四阶段法的计算过程,并将其运用到不同规模网络进行求解,着重研究权重系数、交通分布参数、出行总量对计算效率和计算结果的影响。最后为反馈四阶段法的实际应用提出建议。

Four step procedure is a classic method in traffic engineering and has been widely used for travel demand forecasting. It takes at each step as a relatively independent process. Because of not considering the interactive effect of the steps, the whole data is lack of consistency. The four processes can be closely connected by the method of introducing feedback into four step procedure. It can keep the consistency of the data and avoid solving a complicated combined model. The paper presents the calculation process of the four step procedure with feedback, and solves different scale networks by using the method. It focuses on the effect of three parameters, which are weight, travel parameters and total travel demand, to the efficiency of algorithm. At last, this paper proposes several suggestions to practical application of the four step procedure.