Scheduling in shop floor manufacturing using systematic solution space navigation

• Applicant: Professor Dr.-Ing. Norbert Gronau
• Subject Area: Production Systems, Operations Management, Quality Management and Factory Planning
• Term: from 2015 to 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 274091201

For scheduling problems many approaches for solutions exist, while the navigation through the solution space was never investigated. Existing approaches aim for an optimisation of the algorithm itself or for the better representation or visualisation of the solution space. The fine tuning of an optimisation strategy using dynamically generated knowledge of the solution space is missing till today. A severe research deficit exists in the realm of looking for god solutions based on intended and dynamic adapted navigation strategies through the solution space. Also investigations are missing that point out how important the spanning, rearranging and indexing of the solution space is. The approach to systematically index the solution space and search in the so coded space has not yet been undertaken. The main goal of this research project is to enlarge the classical operators of genetic algorithms (selection, mutation, recombination) with the targeted navigation in the solution space as a new and additional operator. Specially interesting is the possibility to rummage through the solution space without recurring patterns. Beside prim numbers also Fibonacci-sets additional number rows are to be investigated. To reach this goal the research project addresses the following research questions: How the solution space of a multi-dimensional scheduling problem can be structured to be able to be addressed and browsed following certain criteria? Which performance benefit a conventional scheduling approach can get when during the opening solutions from the solution space can be selected following certain criteria? Which additional benefit is possible with manipulating operations in Genetic algorithms as a supplement for selection, mutation and recombination? For which order scenarios the solution space navigation is suitable? When it is possible to find good solutions in very huge solution spaces new possibilities for real time planning are offered directly after a disturbance occurs. This leads to tremendous productivity increase and therefore also to an increase in competitive advantages. The ambitious goal of the project is to cut the time needed to solve a scheduling problem by at least one magnitude. Instead 60 sec one 6 sec are necessary to find a good solution, instead one hour only 6 min.

• DFG Programme: Research Grants