����ģ��������� �źŽ���ڷ���ˮƽģ��

回顾了现有信号交叉口机动车服务水平确定方法，提出了基于模糊神经网络的信号交叉口服务水平模型，设计了基于可视化仿真的驾驶员感受量化调查方法。该模型首先将评价指标输入模糊化，以便利用模糊理论模拟驾驶员的感受形成过程，然后利用人工神经网络驱动模糊推理，从而预测驾驶员对交叉口运行状况的感受，以及交叉口服务水平。本文以混合交通条件下信号交叉口转向车流为例，利用调查数据标定、验证模型。实验结果表明,该模型能够有效预测驾驶员对交叉口运行状况的感受和交叉口服务水平。通行能力、饱和度与驾驶员感受打分值之间的相关性较弱，但方差分析表明这两项指标是感受打分值及服务水平的显著影响因素。延误与感受打分值显著相关。

Signalized Intersection Level-of-Service Model Based on Fuzzy Neural Networks

This paper reviewed the existing methods for signalized intersection level-of-service (LOS). A methodology for evaluating signalized intersection LOS was developed based on fuzzy neural network model. A quantitative surveying method for the driver perceptions was also designed using visualized simulations as an auxiliary tool. The proposed methodology uses fuzzy theory to simulate how the drivers perceive a signalized intersection. It also takes advantages of neural network model to memorize the fuzzy rules to predict the driver perceptions, as well as LOS accordingly. The LOS of turning movements in mixed traffic conditions was taken for example to illustrate the methodology. Surveyed data were used to calibrate and validate the proposed model. The numerical results indicate that the proposed model has great capability to predict the driver perceptions toward LOS. The driver perceptions do not significantly correlate with vehicular capacity or volume-to-capacity ratios, but are found to be significantly influenced by these two parameters through analysis of variance. The driver perceptions depend strongly on average delay.