Hysteresis modeling for a shape memory alloy actuator using adaptive neuro-fuzzy inference system

Abstract

Hysteretic behavior of shape memory alloys (SMA) has become a critically important problem for modeling the SMA actuators. In this paper, an adaptive neuro-fuzzy inference system (ANFIS) is developed to compensate for the hysteretic non-linearity in a mechanism actuates by SMA wires. Experimental data obtained from the mechanism are used to train the ANFIS model. Past output of the system is fed to the model as an input. The trained ANFIS model is validated using several experimental data sets. Compared to other work on the same experimental setup, ANFIS model predicts hysteretic behavior of this system with better performance and lower error bounds.