Topology optimization is regarded as an important design method for structural lightweighting and performance improvement, and it is widely used in manufacturing. However, different optimization methods show significant differences in computational efficiency, convergence, and applicability of results, so performance analysis and comparison of topology optimization methods are necessary. In this paper, the solid isotropic material with penalization method, the evolutionary structural optimization method, the level set method, the moving morphable component method, and neural-network-based topology optimization method are compared from three dimensions: theoretical framework, parameter sensitivity, and numerical stability. Then, numerical experiments on typical two-dimensional structures are carried out to quantitatively evaluate the optimization performance of these methods. Finally, the comprehensive analysis indicates that current methods face difficulties in balancing computational efficiency, structural clarity, constraint-handling ability, and global convergence. Future research can focus on hybrid strategies of different methods, such as hybrid strategies of the solid isotropic material with penalization method and the evolutionary structural optimization method, the combination of neural networks with traditional methods.
He et al. (2025) studied this question.