HyDDRa

Scheduling is a process in w hich a certain set of tasks (activities) need to be assigned to a limited set of machines (resources) to optimize user-defined criteria and requirements. Because of its complexity, most of the scheduling problems cannot be solved exactly, but different heuristic and metaheuristic procedures are used for their solution. A particularly big challenge is dynamic scheduling, w here information about the problem (job release dates, processing time, machine failures) is not know n in advance but becomes know n during system operation. Instead of manually defining customized scheduling algorithms for different combinations of constraints and criteria, hyperheuristic procedures can automatically form the scheduling algorithm for a given problem. Hyperheuristic procedures are optimization methods that do not find the solution to the problem, but optimize the problem-solving algorithm; the most common examples of hyperheuristics are genetic programming and variants of evolutionary algorithms. This project focuses on the application of various hyperheuristic methods w ith the aim of developing dispatching rules for dynamic scheduling problems that have not been thoroughly studied in the literature. In this project proposal, the focus is placed on the unrelated machines scheduling environment and its variants. The project proposal encompasses several research directions aimed at improving the quality of hyperheuristics in automatic developing of the dispatching rules. Additionally, one of the project's goals is to extend this paradigm to related optimization problems, such as scheduling w ith resource constraints and vehicle routing. The project team possesses the necessary experience in applying hyperheuristics and optimization methods and has published preliminary results indicating the viability of further research in this area.