Solving MINLP containing noisy variables and black-box functions using Branch-and-Bound

Abstract Since existing MINLP solvers such as DICOPT require the existence of explicit deterministic equations, they are unable to address systems containing noise and unknown model equations. In this paper we propose a new algorithm based on a Branch-and-Bound main structure to solve MINLP problems involving noise and black-box models, where a response surface method developed in our earlier work is applied to solve a noisy NLP at each node of the branch and bound tree. The algorithm is applied to an example problem to further clarify the steps of the proposed approach.