Structural topology optimisation based on a multi-agent model

Abstract

In this study, we propose to equip the basic vector-based multi-agent system with global structural awareness by introducing strain energy density gradient vectors to achieve structural topology optimisation. This multi-agent based topology optimisation method can conveniently generate diverse and competitive structural designs with different parameters of the agent swarm behavior rules (separation, alignment and cohesion). Several numerical examples are introduced to explore the relationship between parameters and resulting design diversities. It is demonstrated that the proposed method can produce a variety of designs that possess not only high structural performance but also distinctly different topologies. Taking advantages of the computational multi-agent system, this method has potential to be easily integrated with agent system algorithms in other research fields to generate diverse designs with high structural performance, elegant geometries and interesting features for architects, engineers and other designers.