The success of compliant mechanism design by structural topology optimization approach depends,to a large extent,on its structural geometry representation scheme.In this work,a novel representation scheme based on pairs of curves is presented.In the representation,the structure is characterized by a set of input/output(I/O) regions.While it is still unknown how the rest of the design space will be occupied by the structure,the I/O regions must exist somewhere because any structure must have parts which interact with its surroundings by way of at least one loading region,one support region,and one output region.For a valid structural design,pairs of Bezier curves are used to connect I/O regions in order to form one single connected load-bearing structure.The boundary is explicitly described,so the need for smoothening of the blurred and jagged edges can be avoided by developing such a representation scheme to directly generate smooth boundary structures.With the scheme,shape and topology can be optimized simultaneously,and the obtained topology solutions have no check-board phenomena nor intermediate zones.A multi-objective genetic algorithm is then applied to couple with the representation scheme for defining and encoding the structural geometry in the form of graph.The solution framework is integrated with a nonlinear fixed grid finite element method(FG-FEM) code for large-displacement analyses of the compliant structures.Simulation results from a displacement inverter indicated that the proposed representation scheme is appropriate.
