Neural Network Adaptive Control of Magnetic Shape Memory Alloy Actuator With Time Delay Based on Composite NARMAX Model

Magnetic shape memory alloy (MSMA) based actuator plays an important role in the field of micro and nano-fabrication due to its large stroke. However, the inherent hysteresis of MSMA seriously affects the application prospects of MSMA-based actuator in the field of precision positioning. In this study, a composite model is proposed for hysteresis in the MSMA-based actuator by coupling an improved fractional-order Bouc-Wen model to a nonlinear auto-regressive moving average with exogenous inputs (NARMAX) model. Based on the proposed model, a neural network adaptive control method is then used to control the MSMA-based actuator, and the controller design takes into account the effect of time delay on the system performance to improve the positioning accuracy of the MSMA-based actuator. First, we fuse the composite NARMAX model with the canonical form of the nonlinear system to describe the MSMA-based actuator. Then, a control strategy is developed to address the impact of the unknown time delay for controller design and achieve the satisfactory positioning accuracy of the MSMA-based actuator. By using Lyapunov theory, the tracking error is proven to be asymptotic convergence. Experimental studies are provided to validate the effectiveness of the proposed modeling and control schemes.

Automated summary

In this study, a composite model combining an improved fractional-order Bouc-Wen model and a nonlinear auto-regressive moving average with exogenous inputs (NARMAX) model is proposed to describe the hysteresis in MSMA-based actuators. A neural network adaptive control method is used to improve the positioning accuracy of the actuator, considering the effect of time delay on the system performance. Experimental studies validate the effectiveness of the proposed modeling and control schemes.