Dimensionality reduction in feasibility analysis by latent variable modeling

We propose a systematic methodology to exploit partial least-squares (PLS) regression modeling to reduce the dimensionality of a feasibility analysis problem. PLS is used to project the original multidimensional space of input factors onto a lower dimensional latent space. We then apply a radial basis function (RBF) adaptive sampling feasibility analysis on this lower dimensional space to identify the feasible region of the process. A simple low-dimensional representation of the feasible region is thus obtained with this combined PLS-RBF approach. The performance of the methodology is tested on a mathematical example involving six inputs. We show the ability of this PLS-RBF approach to reduce the computational burden of the feasibility analysis while maintaining an accurate and robust identification of the feasible region.