Application of fuzzy neural network control to automatic train operation

A transit system is subject to widely varying external conditions, such as weather, the time of day, etc. Different situations imply different control purposes and the varying conditions induce fluctuating dynamic characteristics. Therefore it is difficult to automatically control the vehicle satisfactorily. We apply a two-degree-of-freedom fuzzy neural network control system to automate the vehicle operation. The controller's function is to accurately stop the vehicle at a station. We investigate the controller's performance with acceleration error due to changing dynamic characteristics and with a control purpose transition from velocity control to position control.<<ETX>>