Forschungsthemen
[PM-FPG] Research and evaluation of strategies to solve extended resource constrained project scheduling problems
In the project HybridPPS, a special Job shop scheduling problem needs to be solved.
The standard case involves projects, each with a number of jobs.
Those jobs depend on other jobs, have a fixed duration and requesting different resources.
Resources have a fixed capacity and are either globally available or only for a certain project.
Such resource-constrained project scheduling problems (RCPSP) are usually NP-hard problems.
Many solution strategies exists for solving these problems.
Small problems can be solved exact (optimally) with mathematical approaches.
Heuristic approaches are suitable to solve larger problems near optimal in an acceptable runtime.
However, to get an initial schedule, all constraints have to be satisfied.
There exists a lot of strategies to satisfy constraints (mathematical programing, discrete event simulation).
In this thesis, a stochastic RCPSP (sRCPSP) should be considered.
Here, durations of jobs and capacities of resources are not fixed, but follow a given stochastic distribution.
This implies, that standard solutions are likely to be infeasible.
The main question is which solution strategy is suitable.
Objectives of work:
- Describing and understanding the considered problem (sRCPSP)
- Describing requirements for solving the concrete problem in HybridPPS
- Literature review for solution strategies, especially in the field of discrete event simulation
- Concept of a solution strategy for the problem formulated in HybridPPS
- If applicable, prototypical implementation of this concept using the existing infrastructure
Betreuer: René Schöne