Agent-based shop floor schseduling

Abstract

Shop floor scheduling problem is known for a long time and is considered an academic exercise for solving by constraint solvers or for linear solvers. In a small production scale it can be solved easily. For a large scale production, the problem becomes too large and we need major heuristics to solve it. We introduce a way to decompose the major problem into smaller sub-problems and to quickly find a satisfying solution using human thinking inspired heuristics. The results combines scalable agent based solution and multiple constraint solver instances running in parallel.