Abstract
This master’s thesis focuses on the investigation and implementation of multi-objectiveoptimization algorithms, particularly NSGA-II, NSGA-III, and MOEA/D. The main
focus is on the application of these algorithms to a complex multi-objective optimization problem provided by the company KREISEL Electric.
The optimization problem is a complex physical system model of the battery-integrated
charging station Chimero from KREISEL Electric. This model is realized in MatlabSimulink and serves as a cost function, thus evaluating the solutions found for the
optimization algorithms. Various parameters are to be optimized so that performance,
efficiency, and lifespan are as high as possible. The parameters to be optimized are finite
discrete values.
The challenges in solving this problem are diverse, including the long computation
time per evaluation and the necessity of adapted mutation and crossover operators for
discrete optimization.
The goal of this work is to describe, implement, compare, and test the functions of
these algorithms. Furthermore, the usability of these algorithms for multi-objective optimization of the optimization problem of KREISEL Electric will be examined. The
results of this work could contribute to finding more efficient solutions for complex discrete problems and improving the application of multi-objective optimization algorithms
in practice.
The test results of the algorithms show that the NSGA-III algorithm delivers the best
results for most test problems. However, the MOEA/D and the NSGA-II are not far
behind the results of the NSGA-III and also show strong performance. The practical
applications demonstrate that all three algorithms can be used successfully and find solutions that, according to the simulation, represent an improvement on the parameters
that already have been used in practice.
Date of Award | 2024 |
---|---|
Original language | German (Austria) |
Supervisor | Stephan Dreiseitl (Supervisor) & Norbert Heublein (Supervisor) |