Nonlinear Model Order Reduction of a MEMS Actuator by a Trajectory Piecewise-Linear Approximation

Abstract:
This contribution presents how to generate a highly efficient yet accurate model of a nonlinear microactuator. A finite element model with nonlinear mechanical contact provides a reference solution, utilizing effective material data to capture the effect of perforation. Based on this model, the combination of two methods generates a surrogate model: projectionbased model order reduction by proper orthogonal decomposition and a trajectory piece-wise linear approximation. The former exploits patterns in deformation to drastically decrease the model’s dimension; the latter efficiently handles nonlinearities, approximating them as a combination of linearized terms. The resulting reduced order model achieves a relative error below 1×10(exp −4) and a speedup factor of more than 250. Therefore, feedback-controlled operation or efficient design studies become feasible. The workflow is suited for industrial simulation software.