Fault-tolerant control based on algebraic derivative estimation applied on a magnetically supported plate

•A non-linear fault-tolerant control of a magnetic levitation system is introduced.•Algebraic derivative estimators facilitate the model-based residual generation.•Robustness against miscalibration, noise, and mechanical vibrations is achieved.•Fault diagnosis parameters are chosen systematically from simple measurements.•The proposed algorithms are validated experimentally.

A fault-tolerant control method based on algebraic derivative estimation is introduced. It is applied on an electromagnetically supported plate as an example of a nonlinear and an open-loop unstable system. The design of the closed loop control is facilitated assuming that relevant faults are identified sufficiently precisely and fast. This is justified by a novel robust model-based fault identification scheme which relies on algebraic methods for numerical differentiation. Derivative estimation parameters and fault-detection thresholds are chosen systematically based on the properties of the measurements. The experimental results show the practical usefulness of the presented methods.