RODRIGUES, L. F. A.; http://lattes.cnpq.br/6595955809685765; RODRIGUES, Luiz Fernando Alves.
Resumo:
Shape Memory Alloys (SMA) can recover a induced plastic deformation through a simple heating which originates a thermoelastic martensitic phase transformation. This phenomenon is known as Shape Memory Effect (SME). It is especially interesting for actuators development and conception in the form of thin wires or ribbons because these elements have ability to recover around 4 to 8 % of its original length. If the SMA has its displacement by SME restricted, it is possible the generation of large recover forces when compared with its dimensions. On the other hand, the design of smart structures using
SMA actuators can be hard in some applications due intrinsic thermal hysteresis associated with its SME behavior. Because of this, studies are often carried to characterize this hysteresis phenomenon for improve the applications with SMA actuators. In this sense, this work proposes to use Artificial Neural Networks (ANN) technique to identify the thermal hysteresis of NiTi SMA wire actuators under the regime of isobaric loads (constant loads) and different thermal activation modes. Results of hysteretic behavior were collected through two experimental test benchs, one electromechanical which uses resistive heating (Joule Effect) and other thermomechanical through performing heat by forced convection in a liquid medium. It was shown the results of ANN learning and network's ability to estimate the hysteretic behavior of the SMA actuators
from a behavior not used in training. In general, it was found that the ANN were efficient in the simulation of hysteresis loops under load as function of electrical current or temperature.