Parameter Estimation Selection for a Sensorless Railway Traction application

This paper presents an approach to solving the speed estimation problem in high-power railway traction applications, including the very low speed range. The full-order Luenberger observer design, based on voltage and current models, is presented as the optimal alternative instead of the conventional method based on observer pole placement. Observer discretization and parameter sensitivity is also analyzed. An adaptive scheme based on Lyapunov theory is proposed for the on-line estimation of stator resistance and the motor speed considering the mismatch of the parameters, measurement noise and inverter model errors. Rotor resistance is also compensated for improved robustness