Nonlinear Observer Design for Guidance and Traction of Railway Vehicles

Abstract

State observer design turned out to be crucial in several recent railway vehicles projects on active control and condition monitoring at DLR, but may also be useful for emerging technologies like the cyber physical system and the digital twin approach or the realization of predictive maintenance concepts. With this background and motivation the paper presents a process in three steps: (i) an initial analysis results in a physical model, (ii) the subsequent transfer to state space representation facilitates the prove of observability and (iii) the observer synthesis supports the design of the observer feedback law. Results from two projects, one related to longitudinal or traction dynamics, the other associated to the guidance task of independently rotating wheels, demonstrate the application of the observer design process and offers a comparison of observer estimates with measurements.

Identification of a dynamic friction model for railway disc brakes

“…This knowledge is indispensable for the system optimization and further development of friction brake systems, such as electro-mechanic and controlled friction brakes. Furthermore, enhanced friction models can be applied in online control strategies which are based on the real time estimation of frictional properties, as presented in [22].…”

Lateral Guidance Control Using Information of Preceding Wheel Pairs

“…Nevertheless, the concept is predicated on information of the track characteristics and the states of the running gear, which is partly not available in recent applications. Particularly, a measurement of y T is recently not implemented in reality, but the progress of the observer development for the states of the DIRW is promising [3]. Most certainly, the availability of track data will rise and more vehicle states can be observed for the realization of an enhanced control.…”

“…According to the concept of feedback linearization x has to be known for the feedback and the input in the two inverted equations of motion. Prospectively, the signals can be attained from observes in reality [3]. In this work, x is considered ideally.…”

“…In reality, the lateral irregularity cannot be measured in a straight-forward way, but can be received from observers in future [3]. To evaluate the potential of the control nevertheless, y T is obtained ideally at the preceding running gear and pre-processed for the control of the trailing one.…”

Improved curving performance of an innovative two-axle vehicle: a reasonable feedforward active steering approach

“…Instead, an extended Kalman filter is used by Heckmann et al in [15] for state estimation of an IRW vehicle non-linear model. Subsequently, in [20], an unscented Kalman filter is compared with the extended Kalman filter showing a reduced estimation error.…”