Electromagnetic Actuator Control: A Linear Parameter-Varying (LPV) Approach

Abstract

This paper deals with system identification and control of a nonlinear electromagnetic actuator, which can be used in many practical applications: electromagnetic valve actuators of combustion engines, artificial heart actuators, magnetic levitation, electromagnetic brakes, etc. The considered practical control problem requires accurate control of the moving armature between two extreme positions. The main objective is to assure small contact velocity, which is known as "soft landing" of the moving armature, and, in this way, low-noise low-component-wear operation. First, due to open-loop instability, system-identification experiments are performed around different equilibrium positions under closed-loop control, and a linear parameter-varying (LPV) model and a bound of plant uncertainty are derived. Next, an LPV controller is designed in a robust control framework (robust gain-scheduled controller). Since the system evolves along quasi-equilibrium positions, quadratic and biquadratic analyses are performed using linear matrix inequalities. Finally, the experimental results show that the controller design problem can be handled successfully, considering an LPV approach. This paper reflects a pragmatic viewpoint: The control structure is simple and easy to implement, and offers good performance and robustness; therefore, it is suitable for industrial applications