| Abstract |
Torque oscillations occur in poorly damped drive trains of electrical high-speed locomotives due to the stochastic
wheel-rail-contact. These reduce the life time of mechanical components. The drive train can be considered as three-mass-spring-system with two frequencies of natural torsion at approximately 20 Hz and 50 Hz. A conventional speed controller is able to damp only the first frequency to a limited extend. With the high-dynamic torque control and single-fed traction motors,
it is possible to develop new control concept based on a state controller while taking the mechanical system into account. It shifts the poles of both natural oscillations in the negative real direction and increases the damping. As the control concept should not need more sensors than the conventional speed controlled AC drive, the missing states have to be reconstructed by an observer or a fllter. While a Luenberger observer is designed for reconstructing the system states after an initial disturbance, the Kalman filter is an optimal state estimator, if the process and the measurements involve permanent random perturbations.
Simulation results with deterministic and stochastic load torques show that the state controller with Kalman filter reduces both oscillations of natural torsion and thereby increases the lifetime of mechanical components in the drive train. |
