A surrogate model for real-time dynamic simulation of dielectric elastomer actuators via long short-term memory networks

Abstract

A surrogate model for dynamical behavior of dielectric elastomer actuators was proposed, the reduced-order framework was based on long short-term memory (LSTM) networks, and the combination of proper orthogonal decomposition with the Galerkin projection. To capture the latent dynamics of actuators, both the displacement and velocity fields were employed in this framework with the enforced electric potential signals, and two stacked architectures of LSTM networks were proposed to predict the dynamical evolution. The performance of the proposed surrogate model was demonstrated through two bimorph actuator concepts. The results show that the proposed model is a promising, reliable, and computationally efficient approach for the real-time simulation.