Chaotic Neural Network-Based Hysteresis Modeling With Dynamic Operator for Magnetic Shape Memory Alloy Actuator

The magnetic shape memory alloy (MSMA) is a new family of smart materials, which exhibits great strain deformation and high energy density. Based on these properties, the MSMA has excellent potential to represent an available means for developing a novel generation of actuators in the micro-positioning application. However, the MSMA-based actuator suffers from the inherent hysteresis and it has become a bottleneck in the industrial application. A hybrid hysteresis model, which consists of a simple dynamic hysteresis operator (SDHO) and chaotic neural network (CNN), is proposed in this article. This developed model possesses a concise construction and distinguished generalization capability. By conducting comparative experiments, the proposed approach has a superior ability to predict the hysteresis behaviors under various input signals.