Surrogate-based distributed optimisation for expensive black-box functions

This paper considers distributed optimisation problems with black-box functions using surrogate-assisted methods. Since the cost and their derivatives are usually impossible to be expressed by explicit due complexity of modern systems, function calls have performed obtain those values. Moreover, often expensive evaluate, therefore designers prefer reduce number evaluations. In this paper, surrogate-based methods utilised approximate true functions, conditions for constructing smooth convex surrogates established, which requirements eliminated. To improve quality surrogate models, a distance-based infill strategy is proposed balance exploitation exploration, guarantees density decision sequence in compact set. Then, algorithm developed solve reformulated auxiliary sub-problems, convergence established via Lyapunov theory. Simulation examples provided validate effectiveness theoretical development demonstrate potential significance framework.