Abstract:  Aiming at the problem that it was difficult for the gray elements and zigzag boundaries in the result of density-based topology optimization method to convert the topology optimization result into the computer-aided design (CAD) model and significantly hinder its engineering application, a new method of boundary smooth reconstruction was proposed for variable density topology optimization result. First, the pseudo density values obtained by the density-based method were used to construct a non-uniform rational B-spline (NURBS) surface in the design domain. Then, a bisection iteration algorithm was constructed to determine the position of the "zero level set" surface, so as to ensure that the transformed model met the volume constraint, and the Visual Basic Application (VBA) was employed to compile the iterative program. Finally, the "zero level set" in the level set method was applied as the structure boundary to construct the CAD model and achieve the smooth boundary extraction. The results of the study show that the proposed method reduces the requirement of reducing gray elements for density-based topology optimization, eliminates the zigzag boundary, and realizes the fast conversion from the optimized configuration to the CAD model. The volume fraction error of the topology optimization model obtained by the three numerical studies is less than 1%, the boundary is smooth and the structural performance is good, which shows the validity and versatility of the method.
Key words:  topology optimization; boundary smooth reconstruction; boundary detection; computer-aided design(CAD); non-uniform rational B-splines(NURBS)