Truss topology optimization for mass and reliability considerations - Co-evolutionary multiobjective formulations

Abstract

The paper presents an approach for simultaneous optimization of structural mass and reliability in discrete truss structures. In addition to member sizing, the selection of an optimal topology from a pre-specified ground structure is a feature of the proposed methodology. To allow for a global search, optimization is performed using a multiobjective evolutionary algorithm. System reliability is based on a recently developed computational approach that is efficient and could be integrated within the framework of an evolutionary optimization process. The presence of multiple allowable topologies in the optimization process was handled through co-evolution in competing subpopulations. A unique feature of the algorithm is an automatic reunification of these populations using hypervolume measure-based indicator as reunification criterion to attain greater search efficiency. Numerical experiments demonstrate the computational advantages of the proposed method. These advantages become more pronounced for large-scale optimization problems, where the standard evolutionary approach fails to produce the desired results.