Duhem Hysteresis Modeling of Magnetic Shape Memory Alloy Actuator via Takagi-Sugeno Fuzzy Neural Network

Abstract

The magnetic shape memory alloy (MSMA)-based actuator is a promising candidate in the micro positioning field by virtues of its large stroke and small volume. However, the inherent hysteresis nonlinearity between the input current and the output displacement seriously limited the application of the MSMA-based actuator. In this paper, the hysteresis, which is related to the input frequency and working condition (such as load), is analyzed. Then a mathematical modeling method using Duhem model (DM) and Takagi-Sugeno fuzzy neural network (TSFNN) is introduced to describe the hysteresis behavior. The mathematical expression of the DM is explicit and simple; and the TSFNN, which has the advantages of both fuzzy system and NN structure, is used to identify the DM parameter. Hence, the proposed TSFNN-DM method has the merits of self adjustment and clear expression. To certify the validity of the developed model, comparative experiments with the modeling methods in other literatures are executed. Experimental results confirm that the TSFNN-DM has the better modeling performance to depict the hysteresis under the different input frequencies and loads than other modeling methods in previous studies.