Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives
Thesis (PhD)--Stellenbosch University, 2026.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Stellenbosch : Stellenbosch University
2026
|
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613903798665216 |
|---|---|
| access_status_str | Open Access |
| author | Prinsloo, Trudie |
| author2 | Bekker, James |
| author_browse | Bekker, James Prinsloo, Trudie |
| author_facet | Bekker, James Prinsloo, Trudie |
| author_sort | Prinsloo, Trudie |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (PhD)--Stellenbosch University, 2026. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/135923 |
| institution | Stellenbosch University (South Africa) |
| language | English |
| last_indexed | 2026-06-10T12:43:33.016Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2026 |
| publishDateRange | 2026 |
| publishDateSort | 2026 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/135923 Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives Prinsloo, Trudie Bekker, James Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. Thesis (PhD)--Stellenbosch University, 2026. Prinsloo, T. 2026. Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives. Unpublished doctoral dissertation. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/647d1829-a481-4058-840e-17379cbb4397 The research presented in this dissertation examines the integration of multi-objective ranking and selection (MORS) techniques with metaheuristic algorithms for solving large-scale, stochastic, multi-objective optimisation (MOO) problems. Many real-world decision problems involve multiple competing objectives, stochastic simulation outputs, and large, complex design spaces. Although metaheuristics are effective in exploring these spaces and approximating Pareto-optimal solutions, they lack statistical guarantees, making the reliability of their outputs uncertain. In contrast, ranking and selection (R&S) procedures provide statistical guarantees, specifically for the probability of correct selection (P(CS)), but they are not suitable for navigating large solution spaces. This trade-off between exploration efficiency and statistical confidence motivates the hybrid approach developed in this research. The study focuses on Procedure MMY, a recently developed MORS procedure that extends the indifference-zone (IZ) approach to multi-objective settings, providing exact P(CS) guarantees under simulation noise. An initial literature review compares IZ and optimal computing budget allocation (OCBA) methods within the broader context of multi-objective simulation optimisation (MOSO). Although OCBA-based MORS approaches have been integrated with metaheuristics, the literature reveals a gap in similar work involving IZ-based methods like Procedure MMY. This research addresses this gap by proposing a hybrid approach in which population-based metaheuristics generate candidate solutions and Procedure MMY is applied to statistically validate the resulting Pareto sets. A methodological contribution lies in addressing the computational burden of Procedure MMY, which requires calculating Rinott's parameter h via computationally intensive double integrals. To overcome this, the study introduces a Neural Network-based approach to approximate three parameters - h1, h2 and h3 - significantly reducing computational effort while maintaining accuracy. Additionally, the research demonstrates that h values derived for high-dimensional objective spaces can be applied to lower-dimensional problems without compromising P(CS) guarantees, enhancing the flexibility of Procedure MMY in practical applications. The hybrid approach is tested on three simulation optimisation problems: an inventory problem (rQ), a buffer allocation problem (BAP), and an extended buffer allocation problem with inventory and energy (BAPrQE). Each problem is tackled using three metaheuristics - the multi-objective optimisation algorithm using the cross-entropy method (MOOCEM), the non-dominated sorting genetic algorithm (NSGA-II), and the competitive mechanism based multi-objective particle swarm optimiser (CMOPSO) - each with and without the integration of Procedure MMY. The quality of the resulting Pareto sets is assessed using hypervolume metrics and statistically compared using t-tests. Results shows that MMY-enhanced solutions are often statistically different and have lower hypervolumes due to the pruning of unsupported solutions. Despite the increased computational time, Procedure MMY provides the significant advantage of guaranteeing P(CS), thereby improving the reliability of outcomes. In summary, this work advances the field by demonstrating the practical integration of Procedure MMY with metaheuristics, introducing an efficient machine learning method for estimating Rinott's h, and quantifying the trade-offs between statistical reliability and computational cost in hybrid MORS-based optimisation procedures. Doctoral 2026-04-15T09:32:21Z 2026-04-15T09:32:21Z 2026-03 Thesis https://scholar.sun.ac.za/handle/10019.1/135923 en Stellenbosch University 153 pages application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Prinsloo, Trudie Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title | Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title_full | Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title_fullStr | Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title_full_unstemmed | Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title_short | Using metaheuristics with the ranking and selection indifference-zone procedure MMY in problems with many objectives |
| title_sort | using metaheuristics with the ranking and selection indifference zone procedure mmy in problems with many objectives |
| url | https://scholar.sun.ac.za/handle/10019.1/135923 |
| work_keys_str_mv | AT prinslootrudie usingmetaheuristicswiththerankingandselectionindifferencezoneproceduremmyinproblemswithmanyobjectives |