Efficient Solution of Distributed MILP in Control of Networked Systems. Issue 2 (2020)

• Record Type: Journal Article
• Authors: Liu, Zonglin; Stursberg, Olaf
• Is Part Of: IFAC-PapersOnLine. Volume 53:Issue 2(2020)
• Page Start: 6723
• Page End: 6729
• Publication Date: 2020
• Subjects: Optimization -- distributed control -- computational methods -- optimal control -- networked systems -- hybrid systems -- complexity -- integer programming; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2020.12.102
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: This paper considers the distributed solution of Mixed-Integer Linear Programming (MILP) problems, a class of problems which is of interest, e.g., in optimization-based control of networked systems involving hybrid dynamics. For a larger number of subsystems, the high combinatorial complexity arising from the integer variables usually prohibits the use of centralized solution schemes, and thus requires distributed computational approaches. The proposed approach is inspired by results based on the Shapley-Folkman-Starr theorem, but it relaxes some conservative assumptions in order to enhance the computational efficiency. Numerical experiments for different MILP problems confirm the advantage of the proposed method with respect to computation times.