Methods for constrained optimization of expensive mixed-integer multi-objective problems, with application to an internal combustion engine design problem

Engineering design optimization problems increasingly require computationally expensive high-fidelity simulation models to evaluate candidate designs. The evaluation budget may be small, limiting the effectiveness of conventional multi-objective evolutionary algorithms. Bayesian algorithms (BOAs) are an alternative approach for but underdeveloped in terms support constraints and non-continuous variables—both which prevalent features real-world problems. This study investigates two constraint handling strategies BOAs introduces first BOA mixed-integer problems, intended use on a engine problem. new empirically compared their closest competitor this problem—the algorithm NSGA-II, itself equipped with components. Performance is also analysed benchmark have similar problem, cheaper evaluate. offer statistically significant convergence improvements between 5.9% 31.9% over NSGA-II across 500 evaluations. Of methods, constrained expected improvement offers better than penalty function approach. For identify improved feasible designs offering 36.4% reductions nitrogen oxide emissions 2.0% fuel consumption when notional baseline design. recommended engineering