Problems when modelling a Shape Memory Alloy (SMA) system

Current have been stuck trying to implement an agnostic SMA simulink simulation so I can develop a control for position. In trying to solve the problem I keep getting errors, large osciallations or large values outside what should be seen.

Any help or suggestions would be appreated as I have been stuck on this for quite a while. I've updated my two attempts to github (see below.

https://github.com/Jconnor20/Shape-Memory-Alloy-Controller

Answer 1

[Shreeya]

Hello

I understand that you are having certain difficulties in modelling SMA based actuators in MATLAB/Simulink. Given the properties of these materials, black box modelling can be a good solution to predict the behaviour. The modelling is performed in two stages:

• The contraction/ heating phase: This phase requires input signals for predicting the output. This prediction can be done using a Hammerstein-Wiener/NARX model or a concatenation of the two approaches. Please refer to the below documentation to understand more about these model implementations:

Hammerstein-Wiener model: https://in.mathworks.com/help/ident/hammerstein-wiener-models.html

NARX model: https://www.mathworks.com/help/deeplearning/ug/design-time-series-narx-feedback-neural-networks.html

• The recovery/cooling phase: During this phase, since the material acquires its original shape back, no input data is provided, and the model estimation can be done using the Curve Fitting Toolbox. Please refer to the below documentation to understand more about Curve Fitting Toolbox: https://www.mathworks.com/products/curvefitting.html

This workflow has been derived from the publication linked below:

https://www.researchgate.net/publication/338769441_Flexible_shape_memory_alloy_actuators_for_soft_robotics_Modelling_and_control[HS1]

Hope this helps!

[Jack Connor]

Thank you for your input Shreeya.

I have read through the paper you have proposed as a solution. I have a few doubts about the method used to simulate the SMA wire, and I am not sure how well it would work for an agnostic system that has a lot of interact between the two wires of either side of the joint.