Abstract: The paper introduces the neural network and fuzzy logic methods into on-ramp metering of urban freeway. In which, the advantages of these two methods are considered, namely, the learning ability, optimization ability, and interconnection structure of the neural network, and the human-like thought manner and professional knowledge integration of fuzzy logic method. The suitable input and output variables are selected by optimization, and are get fuzzification and unfuzzification. Then, the corresponding fuzzy inference rules are established, and the relation generating method and inference synthesis algorithm are also developed. The membership function styles and parameters are determined by the adaptive neuro training method. An example was also given to illustrate the proposed method. The results indicate an increase of operating efficiency and a decrease of accident rate with the application of on-ramp metering of neuro-fuzzy.

Key words: urban traffic, urban freeway, on-ramp metering, neuro-fuzzy, membership function

摘要： 综合考虑神经网络的学习能力、优化能力及连接式结构和模糊逻辑类似于人思维方式并易于嵌入专家知识的特点，将神经网络和模糊逻辑算法共同应用于城市快速路入口匝道驶入控制系统中. 通过优化选择输入输出变量并对其进行模糊化和反模糊化处理，建立相应的模糊推理规则、关系生成方法及推理合成算法，并利用神经自适应训练方法确定隶属函数的形式和参数，最后给出应用示例. 研究结果表明，利用神经模糊原理进行快速路入口匝道驶入控制能够有效提高匝道连接段的利用效率，减少交通事故.

关键词: 城市交通, 城市快速路, 入口匝道控制, 神经模糊, 隶属函数