Research On Fast Analysis Method Of Pneumatic Soft Actuator Based On POD Reduced Order

Dielectric elastomers(DEs)are materials that possess advantages such as low elastic modulus,low density,high energy density,and fast response time.Therefore,they have significant advantages in flexible actuator devices and sensor devices.Currently,they are widely used in soft robots,actuators,sensors,and other applications.Compared to traditional rigid materials,dielectric elastomers exhibit strong material nonlinearity.Traditional finite element software faces challenges in simulating and analyzing hyperelastic material models with large deformations and contact analysis due to the strong material nonlinearity and geometric nonlinearity.These challenges result in extremely time-consuming finite element analysis,low computational efficiency,and difficulties in convergence.This paper focuses on pneumatic soft actuators and proposes a method using dimensionality reduction to achieve rapid analysis of their large deformations and contact forces.This article focuses on dielectric elastomer flexible actuators as the primary research subject.It first describes the driving mechanism of dielectric elastomer actuators and the electromechanical coupling theory based on the thermodynamic framework for uniform deformation of dielectric elastomers.Three strain energy models,namely the Neo-Hookean model,Gent model,and Mooney-Rivlin model,are introduced.In further work,the Gent hyperelastic model,which considers strain stiffening,is selected.Finite element calculations are performed to analyze the driving force of an octopus tentacle soft actuator.The bending curvature and contact force of the octopus tentacle are predicted,and the results are compared with existing experiments to verify the accuracy of the finite element analysis.To address the computational time-consuming issue of solving the flexible actuator model in Abaqus,the following approach was adopted.Firstly,a sufficient amount of sample data was obtained through Abaqus analysis.Secondly,the Proper Orthogonal Decomposition(POD)method was used to process and extract the eigenvectors from the samples.Subsequently,a Radial Basis Functions(RBF)neural network was employed for interpolation to establish a POD-RBF surrogate model.This model exhibited good training,high prediction accuracy,and stable convergence.Finally,by selecting the octopus tentacle soft actuator as the research object,the POD-RBF method was applied to predict the bending large deformation and contact force of the soft actuator.The prediction of the bending curvature and contact force of the octopus tentacle was achieved under different shape parameters,position parameters,cone angles,and fluid pressure.The results showed that the predictions obtained using the POD-RBF rapid prediction method had small errors compared to experimental and finite element results.Thus,it can be used as a substitute for traditional finite element analysis.Furthermore,the modeling process of the flexible actuator was simplified,and the computation time was reduced.This proposed method provides a new approach for the simulation and computation of flexible actuators,offering valuable insights for the design and manufacturing of soft robots.