Multi-variable geometry repair and optimization of passive vibration isolators

Description/Abstract

A range of techniques are considered for the search of a high dimensional design landscape with extensive, unknown and disjointed regions of infeasibility. We present the use of
a hybrid genetic-algorithm / gradient-descent search of the objective function / feasibility problem. The genetic algorithm is used to optimize the vibration isolation of a novel passive structure concept, while the gradient descent method is used to repair infeasible geometries. For a more complicated structure the gradient search fails to find feasible geometries and we resort to a non-dominated sorting multi-objective genetic algorithm which searches the vibration isolation and geometry feasibility simultaneously. Although the complicated geometry is more dificult to optimize, there is potential for significant improvements in vibration isolation.