An evolutionary approach to solve a system of multiple interrelated agent problems

Deterministic approaches to simultaneously solve different interrelated optimisation problems lead to a general class of nonlinear complementarity problem (NCP). Due to differentiability and convexity requirements of the problems, sophisticated algorithms are introduced in literature. This paper develops an evolutionary algorithm to solve the NCPs. The proposed approach is a parallel search in which multiple populations representing different agents evolve simultaneously whilst in contact with each other. In this context, each agent autonomously solves its optimisation programme while sharing its decisions with the neighbouring agents and, hence, it affects their actions. The framework is applied to an environmental and an aerospace application where the obtained results are compared with those found in literature. The convergence and scalability of the approach is tested and its search algorithm performance is analysed. Results encourage the application of such an evolutionary based algorithm for complementarity problems and future work should investigate its development as well as its performance improvements. (C) 2015 The Authors. Published by Elsevier B.V.