Actuator characteristics, particularly actuator dynamics, play a pivotal role in mechatronic control systems. The inherent nonlinearity in synchronous machines may reduce the closed-loop performance of any high-level control system, especially in industrial applications, where system sizing is performed with a heavy emphasis on cost-optimisation, forcing the actuator to consistently reside in marginal operating states. The performance of the motor can be described by its reactiveness over the feasible operating points, yielding a nonlinear character. Typically, motor performance decreases with increasing rotor speed, and in turn, the interconnected high-level control system may be impacted by abruptly occurring oscillations or may become unstable. To counter this effect, feedback linearization is proposed for a Permanent Magnet Synchronous Motor (PMSM) with torque control objective. To maintain exact linearizability of the motor and, critically, to account for sudden load variations, torque tracking is ensured indirectly by a derived rotor speed reference. Such a control scheme is intended to turn a nonlinear motor character into a seemingly linear one to which linear high-level control techniques are more easily applicable. This motor control loop is then embedded into an example electromechanical power steering system (EPS). To validate the proposed approach, its superiority is shown over the standard linear motor control in terms of linear character and consistency within the prescribed operating range, significantly improving overall behaviour in a high-level control system.

This paper presents a novel approach for the subdivision of the parameter space in an LPV control framework. The least appealing characteristic of LPV controller design is the conservative nature of the synthesis suggesting a robustness-inclined overtone albeit handling of nonlinear systems by LPV is proven to be efficient with a moderately high performance level. As a possible remedy, parameter-space subdivision is a standard way of concentrating control effort more suited to the specific operating region in which the system resides. LPV, in the overwhelming majority of cases, is utilized either in the name of nonlinearity inherent in the controlled system or as a method for linear systems in which parameters are uncertain. This paper is an effort to combine these ideas in the polytopic LPV framework. First, the parameter-space division lines are drawn according to the presumed evolution of the scheduling variable. A subset of the parameter region is known a priori to behave ’close’ to linear, that is, in a certain range, any deviation from linear behaviour is handled as a parameter uncertainty and the system is controlled by regular linear controllers whereas in other regions an obvious nonlinearity is present motivating the utilization of LPV controllers. These subsegments are then blended in a way that any guaranteed characteristics of the individual controllers are maintained. The proposed approach is demonstrated on a simple torque-controlled pendulum example in a simulation environment.