Abstract:We present a globally convergent method to accelerate density-based topology optimization using projection-based reduced-order models (ROMs) and trust-region methods. To accelerate topology optimization, we replace the large-scale finite element simulation, which dominates the computational cost, with ROMs that reduce the cost of objective function and gradient evaluations by orders of magnitude. To guarantee convergence, we first introduce a trust-region method that employs generalized trust-region constraints and prove it is globally convergent. We then devise a class of globally convergent ROM-accelerated topology optimization methods informed by two theories: the aforementioned trust-region theory, which identifies the ROM accuracy conditions required to guarantee the method converges to a critical point of the original topology optimization problem; a posteriori error estimation theory for projection-based ROMs, which informs ROM construction procedure to meet the accuracy conditions. This leads to trust-region methods that construct and update the ROM on-the-fly during optimization; the methods are guaranteed to converge to a critical point of the original, unreduced topology optimization problem, regardless of starting point. Numerical experiments on three different structural topology optimization problems demonstrate the proposed reduced topology optimization methods accelerate convergence to the optimal design by up to an order of magnitude.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to accelerate density - based topological optimization by using projection - based reduced - order models (ROMs) and trust - region methods. Specifically, the article aims to: 1. **Reduce computational cost**: Traditional topological optimization methods rely on large - scale finite - element simulations, which are very computationally expensive. The article proposes to replace large - scale finite - element simulations with reduced - order models (ROMs), thereby significantly reducing the cost of objective function and gradient evaluations. 2. **Ensure global convergence**: In order to ensure the convergence of the optimization method, the article introduces a trust - region method based on generalized trust - region constraints and proves its global convergence. In addition, the article also proposes several ROM - accelerated topological optimization methods with global convergence. These methods can dynamically construct and update ROMs during the optimization process. 3. **Handle convex - constraint problems**: The article generalizes the error - aware trust - region method to problems with cheap convex constraints, such as box constraints on design variables and volume fraction limitations, which are common constraint conditions in topological optimization. 4. **Improve optimization efficiency**: Verified by numerical experiments, the proposed ROM - based topological optimization method can accelerate up to an order of magnitude more than standard topological optimization methods (such as the method of moving asymptotes) in different structural topological optimization problems. ### Key Technologies and Methods - **Reduced - Order Models (ROMs)**: Use projection techniques to construct low - dimensional reduced - order models to approximate the solutions of finite - element problems. - **Trust - Region Methods**: By introducing generalized trust - region constraints, ensure that each step in the optimization process is carried out within a reasonable range, thereby ensuring global convergence. - **Error Estimation Theory**: Based on a posteriori error estimation theory, dynamically adjust and update reduced - order models to ensure that they meet the required accuracy conditions. ### Mathematical Formulas The mathematical formulas involved in the article include but are not limited to: - **Volume Constraint**: \[ \sum_{e = 1}^{N_e}\rho_e|\Omega_e|\leq V \] where \(\rho_e\) is the density of element \(e\), \(|\Omega_e|\) is the volume of element \(e\), and \(V\) is the upper limit of the volume constraint. - **Stiffness Matrix and Load Vector**: \[ K(\rho)=\sum_{e = 1}^{N_e}\alpha(\rho_e)P_eK_eP_e^T,\quad f = \sum_{e = 1}^{N_e}P_ef_e \] where \(\alpha(\rho_e)=\rho_l+(1 - \rho_l)\rho_e^p\) is the material property model, and \(K_e\) and \(f_e\) are the element stiffness matrix and load vector respectively. - **Helmholtz Filter**: \[ - r^2\Delta\phi+\phi=\psi \] where \(\phi\) is the filtered density field, \(\psi\) is the unfiltered density field, and \(r\) is the characteristic radius of the filter. Through these technologies and methods, the article successfully develops an efficient and globally convergent topological optimization method, which significantly improves optimization efficiency and reduces computational cost.

A globally convergent method to accelerate topology optimization using on-the-fly model reduction

Globally Convergent Composite-Step Trust-Region Framework for Model-Based Real-Time Optimization

Dynamic graph-based convergence acceleration for topology optimization in unstructured meshes

A Novel 3-D Topology Optimization Methodology Based on the Min-Cut Theorem

Discrete Variable Topology Optimization Using Multi-Cut Formulation and Adaptive Trust Regions

On-the-fly model reduction for large-scale structural topology optimization using principal components analysis

2D Minimum Compliance Topology Optimization Based on a Region Partitioning Strategy

The Path-Engulfment Method for Topology Optimization of Structures

An efficient topology optimization method based on adaptive reanalysis with projection reduction

An augmented Lagrangian trust-region method with inexact gradient evaluations to accelerate constrained optimization problems using model hyperreduction

A New Topology Optimization Methodology Based on Constraint Maximum-Weight Connected Graph Theorem

Multi-grid reduced-order topology optimization

An efficient and easy-to-extend Matlab code of the Moving Morphable Component (MMC) method for three-dimensional topology optimization

An Efficient Gradient Projection Method for Structural Topology Optimization

An adaptive and scalable artificial neural network-based model-order-reduction method for large-scale topology optimization designs

Convergence and worst-case complexity of adaptive Riemannian trust-region methods for optimization on manifolds

Trust Region Based Moving Asymptotes Method: A Stabilized Optimizer for Stress-Constrained Topology Optimization

Enhancing high-fidelity nonlinear solver with reduced order model

A nonmonotone spectral projected gradient method for large-scale topology optimization problems

Latent Crossover for Data-Driven Multifidelity Topology Design

A novel unsteady aerodynamic Reduced-Order Modeling method for transonic aeroelastic optimization