摘要
将两种不同的模糊神经网络分别应用于地铁列车控制中的站间运行阶段和定位停车阶段,仿真取得了令人满意的结果。为了降低控制系统的复杂度,提高系统的泛化能力,采用了在误差函数中引入正则项的方法。同时采用了基于扩展的自适应神经-模糊推理系统(简称ANFIS)获取模糊规则数和训练隶属度函数中心和宽度的方法。实验结果表明,将以上方法应用于地铁列车运行控制是可行的,可以保证较好的舒适性、速度跟随性和停车准确性。
Two kinds of Fuzzy Neural Network have been applied in the running phase and the parking phase of metro train,and the results of the simulation are satisfying.In order to reduce the complexity of the control system and improv-ing it's ability of generalization,a method of adding a regularization term into the error function has been adopted.In the same time ,the method of acquiring the number of the fuzzy regulation and training the center and width of membership function based on adaptive neural fuzzy induce system has been adopted.The result of the experiment shows that the application in the metro train control of above method is feasible,and it can guarantee the comfortable of passenger,fol-lowing of velocity and the accurate of parking.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第2期201-204,共4页
Computer Engineering and Applications
基金
铁道部科技发展基金资助项目(编号:B00(82))
同济大学工科基金项目
