Optimized Feedforward Control of a Multistable Magnetic Actuator

Multistable actuators have a great potential to open up new areas or extend the function of existing applications for microactuators. Their advantage lies in the ability to remain in different resting positions without further energy consumption. A challenge lies in finding suitable inputs to guarantee stable transitions between these positions, especially for actuators with more than two resting positions. This is the case because the area between the resting positions is unstable and may be sensitive to disturbances and input variations. In particular, poorly designed inputs or large disturbances may lead to transitions to undesired positions. In this work, we therefore propose a sampling-based optimization approach to obtain feedforward control inputs, which are robust against such stochastic influences. The inputs generated with our algorithm are applied to a magnetic actuator with three resting positions. We show that the number of successful transitions is increased, while the variance of the outcome is lower compared to manually tuned input signals.