Abstract: Based on practical issues found in field investigation, the reasons of over-speed coupling of new heavy axle load cars were analyzed deeply, and so were that for inadequate rolling of light-load cars under unfavorable condition. The reason found to be that, the traditional exit-speed-control model for hump skating has unreasonable basic resistance value for exit speed calculation, and ignore the environmental impact. To resolve these problems, the concept of the unit co-resistance was put forward, and a dynamic interval speed-control model has established based on the energy conservation law. Moreover, based on uncertain information processing ability of fuzzy logic and self-learning ability of neural network, a target speed-control model based on fuzzy neural networks was established. Finally, a hump was taken as an example to validate the models, which provided reference for speed controlling of hump car-unit rolling.

Key words: railway transportation, hump, dynamic exit speed-control model, fuzzy neural networks, heavy axle load car, unit joint resistance

摘要： 从编组站驼峰解体作业中出现的问题出发，在深入分析重载大轴重货车车场内超速连挂和轻载车辆逆向大风条件下溜放不到位这一矛盾问题的基础上，指出其根本原因是驼峰自动化系统的出口定速模型在车组溜放出口定速中单位基本阻力取值不合理，和没有考虑车组溜放时环境条件变化. 基于此，提出了单位合阻力的概念，根据车组溜放过程中的能量守恒定律，建立了间隔制动出口动态定速模型. 利用模糊逻辑的不确定信息处理能力，兼以神经网络的自学习能力，建立了基于模糊神经网络的目的制动出口定速模型. 最后，通过驼峰仿真实验，验证了模型的有效性，为驼峰车组溜放速度控制提供了理论参考.

关键词: 铁路运输, 驼峰, 出口动态定速模型, 模糊神经网络, 大轴重货车, 单位合阻力