Reinforcement Learning for Characterizing Hysteresis Behavior of Shape Memory Alloys

Shape Memory Alloy actuators are useful in a variety of applications where active materials provide a morphing or shape change capability. Achieving a desired strain or position is done by modulating temperature in the Shape Memory Alloy through an applied voltage difference. Characterization of this temperature-strain relationship is currently done using constitutive models or trial and error experimentation with specimens, which are both time and labor intensive. Shape Memory Alloys also contain both major and minor hysteresis loops, and characterization of the minor hysteresis loops, which map the behavior of a wire that is not fully actuated, is not possible using the constitutive method. Understanding these behaviors and determining a control policy for them is crucial for practical applications. Numerical simulation using Reinforcement Learning was used in an earlier paper for determining the temperature-strain relationship and characterizing the major and minor hysteresis loops, and determining a control policy relating applied voltage to desired strain. This paper verifies earlier Shape Memory Alloy characterization simulation results with an experimental hardware apparatus and improved reinforcement learning algorithms. Results presented in the paper verify the temperature-strain major hysteresis loop simulation results, and also determine the relationships of the minor hysteresis loops.

Reinforcement Learning for Characterizing Hysteresis Behavior of Shape Memory Alloys Conference Paper