The friction coefficient between a railway wheel and rail surface is a crucial factor in maintaining high acceleration and braking performance of railway vehicles thus monitoring this friction coefficient is important. Restricted by the difficulty in directly measuring the friction coefficient, the creep force or creepage, indirect methods using state observers are used more frequently. This paper presents an approach using a Kalman filter to estimate the creep force and creepage between the wheel and rail and then to identify the friction coefficient using the estimated creep force-creepage relationship. A mathematic model including an AC motor, wheel and roller is built to simulate the driving system. The parameters are based on a test rig at Manchester Metropolitan University. The Kalman filter is designed to estimate the friction coefficient based on the measurements of the simulation model. Series of residuals are calculated through the comparison between the estimated creep force and theoretical values of different friction coefficient. Root mean square values of the residuals are used in the friction coefficient identification.