Distributed Multi-scheme Optimization via A Projection Multi-agent System Approach

In this paper, a projection multi-agent system approach is developed for distributed multi-scheme constrained optimization. In the distributed multi-scheme model, each agent has more than one candidate objective function subject to bound constraints, local (in)equality constraints, and resource allocation constraints. The on/off constraints are adopted to characterize the selection of schemes, but they introduced nonconvexity to the model. To overcome divergence caused by the nonconvexity, a projection multi-agent system is designed based on the augmented Lagrange method and projection operator, and it is proven to be convergent to a local minimizer. Finally, a numerical example is elaborated to illustrate the validity of the main results.