Neural network-based nonlinear model predictive control with anti-dead-zone function for magnetic shape memory alloy actuator

Magnetic shape memory alloy-based actuator (MSMA-BA) has the advantages of large strain and high resolution. However, the inherent hysteresis characteristics accompanied by the dead zone in MSMA seriously degrade the positioning accuracy of MSMA-BA. In this study, a gated recurrent neural network (GRNN)-based nonlinear model predictive control (NMPC) method is designed to achieve precise trajectory tracking control of the MSMA-BA. First, a GRNN-based nonlinear auto-regressive moving average with exogenous inputs (NARMAX) model is designed to predict the various nonlinear characteristics of MSMA-BA. Based on the established model, an NMPC method with an anti-dead-zone function is designed. The introduced anti-dead-zone function enables the proposed NMPC algorithm to accelerate the response speed within the dead zone and prevents violent oscillations in the system. The ability of the NMPC to address the hysteresis characteristics accompanied by the dead zone is enhanced. Additionally, the convergence of the proposed NMPC method is analyzed using the Lyapunov stability theory. Extensive experiments are conducted on the MSMA-BA to validate the effectiveness of the proposed method.