Use of load generated by a shape memory alloy for its position control with a neural network estimator

Abstract

This paper presents the position control of a shape memory alloy (SMA) wire actuator by measuring the load generated by the SMA and using a neural network as an estimator for displacement. The experimentation facility is designed and developed for the position control of the SMA. Gain scheduled proportional integral (GSPI) and pulse width modulated GSPI controllers are designed and implemented. The closed-loop performance of the position control system using the displacement estimated from the load measurement is the same as the performance obtained with the direct position feedback. The position control of the SMA actuator using the SMA load with the neural network estimator is an easily implementable method that does not affect the frequency of operation of the SMA actuator.