Hysteresis Modeling and Analysis of Magnetic Shape Memory Alloy-Driven Actuator

The accuracy of the micro-positioning system impelled by a magnetic shape memory alloy-driven actuator (MSMADA), is severely restricted by the frequency-dependent hysteresis nonlinearity. Moreover, the actuating accuracy is further affected by various operating factors such as load and temperature. In this study, a Duhem model (DM) identified online by a Takagi-Sugeno fuzzy neural network (TSFNN-DM) is innovatively proposed for describing the frequency-dependent hysteresis nonlinearity of the MSMADA. The DM, which has the explicit function expression, is one of the popular differential equation-based hysteresis models. However, the determination of the DM parameters is difficult and hinders its further applications. The TSFNN, which combines the advantages of easy expressing of the fuzzy inference system and self-adjustment ability of the NN, is employed to identify the DM parameters online. The rationality of the developed method is proved by a Taylor expansion in theory. Plenty of experiments verify that the proposed TSFNN-DM method is an efficient manner to capture the frequency-dependent hysteresis nonlinearity under different working conditions.