The Research Of Efficient Solver For Structural Topology Optimization With Explicit Boundary

With the increasingly demand for lowcost and excellent efficiency in the intensively competitive manufacturing industries,requirements of engineering dose not only satisfy finding the material distribution with the best trade-off between stiffness and volume by traditional approaches implicitly anymore.It is necessary to find an explicit topology approach,the topology descriptions of which contains specific geometric information,to integrate CAD/CAE.In this way,redundant and multiple structure modifications between CAE and CAD can be avoided by boundary extraction.In addition,the bottleneck of dimension curse also restricts the development of explicit structural topology optimization.Therefore,in this paper the principal purpose is devoted to improve computational efficiency of MMC/MMV based topology optimization.The primary studied contents are described as follows:For 2D structure optimization,an Iterative Reanalysis Approximation(IRA)is integrated with the Moving Morphable Components(MMCs)based topology optimization,aiming to improve computational efficiency for static anal.The IRA stems from the multigrid(MG)and an exact reanalysis solver,(IFU).Because of the non-monotonous behavior in minimum compliance topology optimization,a hybrid optimizer based on the Method of Moving Asymptotes(MMA)approach and the Globally Convergent version of the Method of Moving Asymptotes(GCMMA)is suggested to improve the convergence ratio and avoid local optimum.After sufficient numerical experiments,it indicates that the MMC based topology optimization relies too much on empirical parameters.At the same time,the finally obtained results might be only a local optimum.Consequently,a Machine Learning(ML)-based loop-locked parameter optimization method by PSO for the MMC based topology optimization is proposed in this study.In order to expand the research into 3D structural topology optimization,the MMV framework constructed by nonuniform rational B-spline(NURBS),is introduced into topology optimization.In the same way,IRA is integrated with the Moving Morphable Voids(MMV)based topology optimization to reduce computational cost.