Metamodel Based Design Automation: Applied on Multidisciplinary Design Optimization of Industrial Robots

Abstract

Intricate and complex dependencies between multiple disciplines require iterative intensive optimization processes. To this end, multidisciplinary design optimization (MDO) has been established as a convincing concurrent technique to manage inherited complexities. This paper presents a high level CAD and CAE design automation methodology which enables fast, efficient concept generation for MDO. To increase the evaluation speed, global metamodels are introduced to replace computationally expensive CAD and CAE models. In addition, various techniques are applied to drastically decrease the number of samplings required to create the metamodels. In the final part of the paper, a multi-level optimization strategy is proposed to find the optimal concept. As proof of concept, a real world design problem, from ABB industrial robotics, is presented.