Indirect Intelligent Sliding Mode Control of Antagonistic Shape Memory Alloy Actuators Using Hysteretic Recurrent Neural Networks

This paper presents the development of an indirect intelligent sliding mode control (IISMC) system for use with antagonistic shape memory alloy (SMA) actuators. The controller manipulates input voltages to a pair of offset antagonistic SMA tendons, resulting in heat-induced bending of a flexible beam. Hysteresis compensation is achieved using a pair of hysteretic recurrent neural networks, which map the nonlinear, hysteretic relationships between SMA temperatures and bending angle for each tendon. The sliding mode control law regulates tendon temperatures based on reference and measured bending angle. Additionally, the IISMC incorporates an antislack component to increase tendon responsiveness. Experimental results demonstrate precise tracking of periodic reference trajectories. The controller is robust to model uncertainty, disturbances, and parameter variations, with bending angle tracking results superior to those of an optimized proportional + integral controller.