Stochastic project scheduling problems with flexible resource profiles: Combining Operations research methods with machine learning techniques

• Applicant: Professorin Dr. Julia Rieck
• Subject Area: Management and Marketing
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 520722132

In many industries, work in projects is used as the predominant form of work. Typically, projects are carried out according to the four phases of the project life cycle. In particular, a detailed project planning must be done before the realization of a project. Project planning starts before all necessary information is known, for example the durations of the project activities are still uncertain. To deal with the uncertainty, the activity durations can be defined as random variables. In addition, it is often the case that in the planning phase only the total resource requirements (e.g., in terms of person-hours) needed in total to perform an activity can be determined. Thus, the activity durations and the associated resource requirements in the individual time periods of the execution must be determined. Considering these two aspects, stochastic project scheduling problems with flexible resource profiles arise, which will be addressed in the research project. The aim of the project is to develop powerful mathematical models and algorithms of operations research and machine learning to solve the present, new problem. Due to the uncertainty, chance-constrained conditions will be integrated in the model formulations and a solution with sample-average-approximation approaches will be provided. Metaheuristic methods will be implemented based on policies that prescribe the start of activities at specific decision points in a dynamic manner. Meta-learning will be used to dimension model parameters for underling learning and optimization processes. In addition, reinforcement learning will be used to learn an optimal policy when searching for a solution. The developed problem is of high practical relevance, especially for consulting companies that carry out many engineering projects on a daily basis, or for relief organizations that organize response actions to disasters, emergencies, or major technical accidents as projects.

• DFG Programme: Research Grants