Abstract:In traditional topology optimization, the computing time required to iteratively update the material distribution within a design domain strongly depends on the complexity or size of the problem, limiting its application in real engineering contexts. This work proposes a multi-stage machine learning strategy that aims to predict an optimal topology and the related stress fields of interest, either in 2D or 3D, without resorting to any iterative analysis and design process. The overall topology optimization is treated as regression task in a low-dimensional latent space, that encodes the variability of the target designs. First, a fully-connected model is employed to surrogate the functional link between the parametric input space characterizing the design problem and the latent space representation of the corresponding optimal topology. The decoder branch of an autoencoder is then exploited to reconstruct the desired optimal topology from its latent representation. The deep learning models are trained on a dataset generated through a standard method of topology optimization implementing the solid isotropic material with penalization, for varying boundary and loading conditions. The underlying hypothesis behind the proposed strategy is that optimal topologies share enough common patterns to be compressed into small latent space representations without significant information loss. Results relevant to a 2D Messerschmitt-Bölkow-Blohm beam and a 3D bridge case demonstrate the capabilities of the proposed framework to provide accurate optimal topology predictions in a fraction of a second.

High-Speed Link Design Optimization Using Machine Learning SVR-AS Method

A Fast Optimization Method for High-Speed Link Inverse Design with SVR-AS Algorithm

Support Vector Regression-Based Active Subspace (SVR-AS) Modeling of High-Speed Links for Fast and Accurate Sensitivity Analysis

Deep Neural Network-Based Surrogate-Assisted Inverse Optimization for High-Speed Interconnects

The Optimization of the Key Components of the Spinning Machine Conveyor System Based on ISIGHT

Fast Machine-Learning-Driven Supply Noise-Aware Macromodeling for High-Speed Nonlinear Drivers

Comparison of Machine Learning Techniques for Predictive Modeling of High-Speed Links

Machine Learning for high speed channel optimization

Automated Design of Machine Vision Lens Based on the Combination of Particle Swarm Optimization and Damped Least Squares

A Fast Global Optimal Technique Based On Combinations Of Improved Radial Basis Functions And Tabu Search Method

Developments of an Efficient Global Optimal Design Technique – a Combined Approach of MLS and SA Algorithm

Accelerating Chip Design with Machine Learning: from Pre-Silicon to Post-Silicon

A machine learning-assisted structural optimization scheme for fast-tracking topology optimization

Accelerating gradient-based topology optimization design with dual-model artificial neural networks

Accelerating gradient-based topology optimization design with dual-model neural networks

An Efficient Analog Circuit Sizing Method Based on Machine Learning Assisted Global Optimization

Real-time topology optimization via learnable mappings

Steering linkage topology design using angle-based block partitioning symmetric model (APSM)

Efficient Inverse Design Optimization through Multi-fidelity Simulations, Machine Learning, and Search Space Reduction Strategies

A Systematic Optimization Design Method for Complex Mechatronic Products Design and Development

Microarchitectural Design Space Exploration Via Support Vector Machine