The Analysis of Route Choice Behavior by Fuzzy Reasoning Approach

提出了一种基于模糊推理的路径选择模型，并将其应用于交通信息服务中。本文的主要研究成果按研究过程的先后顺序归纳如下：1）对城市道路网络中的路径选择现象进行了初步的研究。利用这些数据建立了模糊推理的路径选择模型。通过收集与交通信息相关的路径选择因素，并利用语言表达式建立基于知识的路径选择规则。参考T-范数和反模糊化的相关研究成果，建立了原始模型。2）对模糊推理模型进行了实用修正。为了提高路径流量预测的计算效率，引入了简化模型。并引入多级模糊推理来描述顺序决策步骤。提出了一种合适的模糊推理模型（FL模型）， ...更多信息 3）讨论了FL模型的不足之处，即隶属函数的估计问题。由于FL模型的计算过程是非线性的，因此可以采用遗传算法进行参数估计。考虑了真实的值与估计值之间的误差，得到了路径选择的参数。4）建立了基于语言表达的模糊推理路径选择描述模型。然而，对于路径流的实际估计，建议使用清晰值估计。引入误差反向传播的神经网络模型进行修正。将模糊推理模型与神经网络结合，提出了一种混合模型。5）将几种类似的技术归纳为“软计算”，并讨论了它们在构建出行行为模型方面的优势。最后，对模型在交通信息管理中的有效应用进行了探讨。对交通信息的传输和信息服务进行了详细的分析。6）对研究成果进行了总结，并在几次会议上进行了报告，建议进一步讨论。本文还拟在国际、国内有关土木工程的会议上发表。少

The route choice model with fuzzy reasoning is considered to be developed and applied to the traffic information service. The main research results are summarized as follows in the order of the research procedure :1) The primitive survey to describe the route choice phenomena on the urban road network is done. The route choice model with fuzzy reasoning is provided using the data. The factors of route choice associated with traffic information are gathered and the knowledge based rules for the route choice are created with the linguistic expressions. The primitive model is established with referring to the related research results about T-norms and defuzzification.2) The fuzzy reasoning model is practically modified. The simplified model is introduced to promote the efficiency of calculation in forecasting the path flows. And also the multi-stage fuzzy reasoning is introduced to describe sequential decision steps. The proper type of fuzzy reasoning model (FL model) is proposed with the … More se improvements and the impacts of traffic information to the flows are considered.3) The estimation of the membership functions is discussed as a disadvantage of the FL model. The genetic algorithm may be used to estimate the parameters because the FL model involves non-linear calculating process. The parameters are obtained with considering the errors between real values and estimated values.4) The model with fuzzy reasoning for route choice description is established with linguistic expressions. However, the crisp value estimation is recommended for the practical estimation of the path flows. The neural network model with error back propagation is introduced to modify at this point. The fuzzy reasoning model with neural network in conclusions of the rules are proposed as a hybrid model.5) The several similar techniques are summarized as "soft computing" and their advantages to construct the travel behavior models are discussed. Furthermore, the effective use of the models to traffic information management is mentioned. The transforming the traffic information and information service are analyzed precisely. The result shows the proper way of information provision on urban transport networks.6) The research results are summarized and reported in the several conferences and the further discussions are recommended. The some remarks are also planned to be reported the international and domestic conferences related with civil engineering. Less