Train Velocity Estimation by Extended Kalman Filter

A train longitudinal velocity is needed for some control devices. The train velocity is typically determined from a locomotive wheelsets velocity. This solution can cause problems when the wheelsets are driven or braked. In these cases, the wheelset velocity is higher or lower than the train longitudinal velocity due to a slip or skid velocity. The problem is typically solved by some type of averaging or filtration of all available wheelsets velocities. These algorithms need to know all wheelsets velocities at the same time. This task can be difficult to fulfil in some types of distributed computer systems due to communication delays between computers. The problem is solved by using an extended Kalman filter that estimates the train velocity from one wheelset velocity in the paper. The filter and its properties are described and designed in the paper. The measured data are used for the filter function check.