Force Control of SMA Springs Using RNNs

Shape memory alloys (SMAs), generally composed of nickel-titanium, have unique properties that enable them to transition from the martensite phase to the austenite phase when heated. During this transformation, SMAs can shorten and revert to their original shape, which makes them valuable as actuato...

Full description

Saved in:
Bibliographic Details
Main Authors: Ahmet Atasoy, Mehmed Ozkan
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11005977/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850219639422844928
author Ahmet Atasoy
Mehmed Ozkan
author_facet Ahmet Atasoy
Mehmed Ozkan
author_sort Ahmet Atasoy
collection DOAJ
description Shape memory alloys (SMAs), generally composed of nickel-titanium, have unique properties that enable them to transition from the martensite phase to the austenite phase when heated. During this transformation, SMAs can shorten and revert to their original shape, which makes them valuable as actuators. SMAs’ biocompatibility, high power-to-weight ratio, low noise, and ability to generate substantial force position them as essential materials in robotic and biomedical applications. However, their control is complicated by inherent nonlinearity and hysteresis. When cooled in the austenite phase, SMAs elongate and deform, with different phase transformation temperatures observed during heating and cooling due to hysteresis. These complexities highlight the need for advanced control techniques. Artificial intelligence models, particularly recurrent neural networks (RNNs), are well-suited for capturing the intricate relationships within SMA behavior. This paper introduces an approach that utilizes RNNs and real-time measurements for force control in SMAs. A compact test setup equipped with sensors collects data on voltage, current, and generated force, enabling the system’s dynamic behavior analysis. RNNs effectively model the nonlinear characteristics of SMA actuators, leveraging their capacity to learn complex temporal relationships from the collected data. This approach is shown to enhance the control of SMAs over conventional control schemes, i.e., Proportional Integral Derivative (PID) control, and expand their applications in various fields.
format Article
id doaj-art-da27bf8219384bb18d0e7960f194af83
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-da27bf8219384bb18d0e7960f194af832025-08-20T02:07:19ZengIEEEIEEE Access2169-35362025-01-011310105110106510.1109/ACCESS.2025.357085611005977Force Control of SMA Springs Using RNNsAhmet Atasoy0https://orcid.org/0000-0003-1797-1420Mehmed Ozkan1https://orcid.org/0000-0003-2492-8113Institute of Biomedical Engineering, Boğaziçi University, İstanbul, TürkiyeInstitute of Biomedical Engineering, Boğaziçi University, İstanbul, TürkiyeShape memory alloys (SMAs), generally composed of nickel-titanium, have unique properties that enable them to transition from the martensite phase to the austenite phase when heated. During this transformation, SMAs can shorten and revert to their original shape, which makes them valuable as actuators. SMAs’ biocompatibility, high power-to-weight ratio, low noise, and ability to generate substantial force position them as essential materials in robotic and biomedical applications. However, their control is complicated by inherent nonlinearity and hysteresis. When cooled in the austenite phase, SMAs elongate and deform, with different phase transformation temperatures observed during heating and cooling due to hysteresis. These complexities highlight the need for advanced control techniques. Artificial intelligence models, particularly recurrent neural networks (RNNs), are well-suited for capturing the intricate relationships within SMA behavior. This paper introduces an approach that utilizes RNNs and real-time measurements for force control in SMAs. A compact test setup equipped with sensors collects data on voltage, current, and generated force, enabling the system’s dynamic behavior analysis. RNNs effectively model the nonlinear characteristics of SMA actuators, leveraging their capacity to learn complex temporal relationships from the collected data. This approach is shown to enhance the control of SMAs over conventional control schemes, i.e., Proportional Integral Derivative (PID) control, and expand their applications in various fields.https://ieeexplore.ieee.org/document/11005977/Non-linear force controllayer-recurrent neural network (LRNN)prosthetic fingerreal-time measurementrecurrent neural network (RNN)resistance measurement
spellingShingle Ahmet Atasoy
Mehmed Ozkan
Force Control of SMA Springs Using RNNs
IEEE Access
Non-linear force control
layer-recurrent neural network (LRNN)
prosthetic finger
real-time measurement
recurrent neural network (RNN)
resistance measurement
title Force Control of SMA Springs Using RNNs
title_full Force Control of SMA Springs Using RNNs
title_fullStr Force Control of SMA Springs Using RNNs
title_full_unstemmed Force Control of SMA Springs Using RNNs
title_short Force Control of SMA Springs Using RNNs
title_sort force control of sma springs using rnns
topic Non-linear force control
layer-recurrent neural network (LRNN)
prosthetic finger
real-time measurement
recurrent neural network (RNN)
resistance measurement
url https://ieeexplore.ieee.org/document/11005977/
work_keys_str_mv AT ahmetatasoy forcecontrolofsmaspringsusingrnns
AT mehmedozkan forcecontrolofsmaspringsusingrnns