LPV Control of Systems via Nonlinear Actuator Dynamics Mapping

The nonlinear nature of industrial actuators poses a major challenge in the design of mechatronic components. Variants of DC motors tend to exhibit more or less linear characteristics but are usually burdened with nonlinear marks especially due to friction, cogging, and magnetic current phenomena. Still, for an overwhelming majority of large-scale industrial applications, it is more common to employ sophisticated AC motors with nonlinear dynamics. Moreover, the load is rarely, if ever known, introducing a dramatic uncertainty in the actuator dynamics. This paper is concerned with a controller design scheme, in which nonlinear actuators are described as subsystems embedded into a higher-level control system. In the simplest form, the nonlinear actuator can be represented as a linear system by virtue of coordinate transformation coupled with linearizing feedback. However, the inclusion of technical constraints like measurement timing and disturbance handling results in an actuator description that makes exact linearization a less feasible option. Therefore, an alternative approach is presented in which the nonlinear character is mapped into a series of simple $1^{\text {st }}$ order systems and is subsequently handled by an LPV control structure. The control scheme is completed with an Unscented Kalman Filter as a load estimator. The design technique and validation tests are performed in a simulation environment, heavily based on a real-life motor and a general $2^{\text {nd }}$ order mechatronic system.