Algebraic surrogate‐based process optimization using Bayesian symbolic learning

Abstract Here, we propose a strategy for the global optimization of process flowsheets, fundamental problem in systems engineering, based on algebraic surrogates that are built from rigorous simulations via Bayesian symbolic regression. The applied method provides closed‐form expression can be optimized to optimality using state‐of‐the‐art solvers, where BARON or ANTIGONE were solvers choice. When predicting unseen test data, models show similar accuracy level compared traditional Gaussian processes. However, they more easily due their analytical structure, which allows user apply well‐established deterministic solvers. We capabilities our approach several case studies, ranging units full flowsheets. performance is assessed by comparing CPU time model building, prediction identified model, and subsequent with proven benchmark.