Sigmoid-based hysteresis modeling and high-speed tracking control of SMA-artificial muscle

This paper presents a novel compact shape memory alloy (SMA) based artificial muscle (AM) with self-sensing, silent operation and high force-to-weight ratio functions. The AM has great potential to develop light and compact active ankle-foot orthosis (AAFO) for ankle-foot rehabilitation. However, its performance is severely limited by SMA's slow-speed response and hysteresis properties. Firstly, an effective forced vessel cooling method is introduced to improve the AM response speed without compromising its compactness. Then SMA hysteresis properties are deeply explored and modeled under different stress and driving frequency conditions, whereby the hysteresis is well compensated when the hysteresis model based feedforward controller is proposed to the AM. Finally, a newly AM actuated AAFO is proposed to demonstrate the actuating capability of the AM. Experimental results show that the AM response speed and accuracy are greatly improved with tracking frequency increased from 0.08 to 1 Hz, RMS tracking errors reduced by up to 82%, which basically satisfies the initial ankle-foot rehabilitation application.