The research topic of the thesis is the extension of evolutionary multi-objective optimization for real-world scheduling problems. Several novel algorithms are proposed: the diversity indicator... Show moreThe research topic of the thesis is the extension of evolutionary multi-objective optimization for real-world scheduling problems. Several novel algorithms are proposed: the diversity indicator-based multi-objective evolutionary algorithm (DI-MOEA) can achieve a uniformly distributed solution set; the preference-based MOEA can obtain preferred solutions; the edge-rotated cone can improve the performance of MOEAs for many-objective optimization; and dynamic MOEA takes the stability as an extra objective. Besides the classical flexible job shop scheduling, the thesis proposes solutions for the novel problem domain of vehicle fleet maintenance scheduling optimization (VFMSO). The problem originated from the CIMPLO (Cross-Industry Predictive Maintenance Optimization Platform) project and the project partners Honda and KLM. The VFMSO problem is to determine the maintenance schedule for the vehicle fleet, meaning to find the best maintenance order, location and time for each component in the vehicle fleet based on the predicted remaining useful lifetimes of components and conditions of available workshops. The maintenance schedule is optimized to bring business advantages to industries, i.e., to reduce maintenance time, increase safety and save repair expenses. After formulating the problem as a scalable benchmark in an industrially relevant setting, the proposed algorithms have been successfully used to solve VFMSO problem instances. Show less
A common method to solve expensive function evaluation problem is using Bayesian Global Optimization, instead of Evolutionary Algorithms. However, the execution time of multi-objective... Show moreA common method to solve expensive function evaluation problem is using Bayesian Global Optimization, instead of Evolutionary Algorithms. However, the execution time of multi-objective Bayesian Global Optimization (MOBGO) itself is still too long, even though it only requires a few function evaluations. The reason for the high cost of MOBGO is two-fold: on the one hand, MOBGO requires an infill criterion to be calculated many times, but the computational complexity of an infill criterion has so far been very high. Another reason is that the optimizer, which aims at searching for an optimal solution according to the surrogate models, is not sufficiently efficient. The main contributions of this thesis consist of 1. Decreased the computational complexity of a well-known infill criteria, Expected Hypervolume Improvement, into $n log (n)$ both in 2-D and 3-D cases; 2. Proposed a new criterion, Truncated Expected Hypervolume Improvement, to make full use of a-priori knowledge of objective functions, whenever it is available; 3. Proposed another infill criterion, Expected Hypervolume Improvement Gradient, to improve the convergence of the optimizer in MOBGO. Show less
