Gaussian Surrogate Dimension Reduction for Efficient Reliability-Based Design Optimization

What is it about?

Surrogate centric reliability-based design optimization is typically limited by the size of approximated space. The Gaussian Surrogate Dimension Reduction Method (GAUSS-DRM) alleviates this limitation by decomposing linear parameters that are normally distributed into a kriging error estimator. We show that GAUSS-DRM is more efficient than traditional surrogate and gradient-based techniques.

Featured Image