Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises
Schroth, T. F. 2025. Development of a Methodology for Artificial Intelligence-based Clustering of Stow and Pick Operations in order to Improve Logistics Performance in Small and Medium-sized Enterprises. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://s...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Stellenbosch : Stellenbosch University
2025
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613854160125952 |
|---|---|
| access_status_str | Open Access |
| author | Schroth, Timo Friedrich |
| author2 | Von Leipzig, Konrad |
| author_browse | Schroth, Timo Friedrich Von Leipzig, Konrad |
| author_facet | Von Leipzig, Konrad Schroth, Timo Friedrich |
| author_sort | Schroth, Timo Friedrich |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Schroth, T. F. 2025. Development of a Methodology for Artificial Intelligence-based Clustering of Stow and Pick Operations in order to Improve Logistics Performance in Small and Medium-sized Enterprises. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/45918fc0-abbc-4876-89ea-e653d2cad76a |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/132441 |
| institution | Stellenbosch University (South Africa) |
| language | English |
| last_indexed | 2026-06-10T12:42:44.343Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/132441 Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises Schroth, Timo Friedrich Von Leipzig, Konrad Hummel, Vera Zincume, Philani Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. Artificial intelligence -- Industrial applications Business logistics -- Data processing Algorithms Order picking systems Stowage -- Data processing UCTD Schroth, T. F. 2025. Development of a Methodology for Artificial Intelligence-based Clustering of Stow and Pick Operations in order to Improve Logistics Performance in Small and Medium-sized Enterprises. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/45918fc0-abbc-4876-89ea-e653d2cad76a Thesis (MEng)--Stellenbosch University, 2025. ENGLISH ABSTRACT: This research focuses on developing an AI-driven methodology to optimize stow and pick operations within intralogistics by reducing travel distances between operations. The proposed methodology uses a simulation model and AI based clustering techniques to improve logistics performance, particularly in warehouse environments. Initially, the solution development consists of two key phases. First, a simulation model of the Werk150 logistics system was modeled using AnyLogic software for simulating stow and pick operations in a three-dimensional warehouse. To compensate for the lack of real data, synthetic data was generated to train the AI model. The goods involved in the stow and pick operations were defined as components of a scooter, and the simulation took into account real-world variables such as motion as well as master simulation parameters and MTM standard times for manual operations. Thus, a digital twin of the logistics system was created, enabling controlled experiments and scenario analysis. During the second phase, the synthetic data was prepared for AI-based cluster analysis. The covered travel distance was selected as the primary metric and the data was restructured for processing by an unsupervised machine learning model using TensorFlow. The AI-driven cluster analysis, based on the k-means algorithm, aimed to group stow and pick operations into clusters that minimize the covered travel distance. Two approaches were tested: combined clustering, in which stow and pick operations can be mixed within clusters, and separate clustering, in which stow and pick operations are clustered independently. The aim of the study was to determine which approach offered a greater reduction in travel distances and hence better logistics performance. Moreover, the development of the unsupervised machine learning model to conduct the AI cluster analysis began with data preprocessing to ensure high input quality, followed by mapping the logistics system in three-dimensional coordinates. The k-means algorithm was selected for the cluster analysis because of its characteristics in grouping operations to reduce travel distances by ensuring spatial proximity. Iterative clustering cycles were performed, processing data in subsamples to optimize cluster composition and minimize travel distances. After clustering, the Dijkstra shortest path algorithm was applied to calculate travel distances within clusters between the operations to determine the cumulated travel distance across all clusters. The results of the combined and separate clustering approaches were compared. In addition, the key findings showed that both clustering approaches reduced travel distances, with the combined clustering method offering a potentially greater reduction. The results of the AI model were stored in an output file for further analysis and evaluation. To conclude, this research leveraged the potential of AI-driven clustering to improve logistics performance in terms of effectiveness and efficiency by reducing travel distance between operations in the clusters. The integration of simulation and AI models within the methodology made it feasible to optimize processes even in environments with limited real data and demonstrated a scalable and adaptable approach to improving intralogistics. AFRIKAANSE OPSOMMING: Hierdie navorsing fokus op die ontwikkeling van ’n KI-gedrewe metodologie om stoor- en verkrygingsoperasies binne die intralogistieke omgewing te optimeer deur die vervoertyd tussen operasies te verminder. Die voorgestelde metodologie gebruik ’n simulasie-model en KI-gebaseerde groeperingstegnieke om logistieke prestasie te verbeter, veral in pakhuisomgewings. Aanvanklik bestaan die ontwikkeling van die oplossing uit twee belangrike fases. Eerstens is ’n simulasie-model van die Werk150-logistieke stelsel gemodelleer met behulp van die “AnyLogic”-sagteware om stoor- en afhaaloperasies in ’n driedimensionele pakhuis te simuleer. Om vir die gebrek aan werklike data voorsiening te maak, is sintetiese data gegenereer om die KI-model op te lei. Die goedere wat by die stoor- en afhaaloperasies betrokke is, is as komponente van ’n bromponie gedefinieer, en die simulasie het werklike veranderlikes soos beweging, asook hoofsimulasieparameters en MTM-standaardtye vir handmatige operasies, in ag geneem. Sodoende is ’n digitale tweeling van die logistieke stelsel geskep, wat beheerde eksperimente en scenario-analise moontlik maak. Gedurende die tweede fase is die sintetiese data voorberei vir KI-gebaseerde groepe-analise. Die afgelêde vervoertyd is as die primêre maatstaf gekies, en die data is herstruktureer vir verwerking deur ’n onbewaakte masjienleer-model met behulp van die sagteware TensorFlow. Die KI-gedrewe groepe-analise, gebaseer op die “k-means” algoritme, is daarop gemik om stoor- en afhaaloperasies in groepe te plaas wat die afgelêde reistyd verminder. Twee benaderings is getoets: ‘n gekombineerde groepering, waarin stoor- en afhaaloperasies binne groepe gemeng kan word; en afsonderlike groepering, waarin stoor- en afhaaloperasies onafhanklik gegroepeer word. Die doel van die studie was om te bepaal watter benadering ’n groter vermindering in reistyd en dus beter logistieke prestasie bied. Verder het die ontwikkeling van die onbewaakte masjienleer-model om die KI-groepe-analise uit te over begin met data voor-verwerking om hoë insetkwaliteit te verseker, gevolg deur die kaartvorming van die logistieke stelsel in driedimensionele koördinate. Die “k-means” algoritme is vir die groepe-analise gekies vanweë sy kenmerke om operasies te groepeer om reistyd te verminder deur ruimtelike nabyheid te verseker. Iteratiewe groepe-siklusse is uitgevoer en data in deelmonsters verwerk om groepe-samestelling te optimiseer en reistyd te verminder. Na groepering is die “Dijkstra kortste-pad" algoritme toegepas om reistye binne groepe tussen die operasies te bereken om die totale reistyd oor alle groepe te bepaal. Die resultate van die gekombineerde en afsonderlike groepe-benaderings is vergelyk. Die sleutelbevindinge het gewys dat beide groeperingsbenaderings reistye verminder, met die gekombineerde groepe-metode wat moontlik ’n groter vermindering bied. Die resultate van die KI-model is in ’n uitvoerlêer gestoor vir verdere analise en evaluering. Ten slotte het hierdie navorsing die potensiaal van KI-gedrewe groepering gebruik om logistieke prestasie in terme van doeltreffendheid en effektiwiteit te verbeter deur reistyd tussen operasies in die groepe te verminder. Die integrasie van simulasie en KI-modelle binne die metodologie het dit moontlik gemaak om prosesse te optimiseer selfs in omgewings met beperkte werklike data, en het ’n skaalbare en aanpasbare benadering tot die verbetering van intralogistiek gedemonstreer. Masters 2025-06-06T14:06:30Z 2025-06-06T14:06:30Z 2025-03 Thesis https://scholar.sun.ac.za/handle/10019.1/132441 en Stellenbosch University xviii, 257 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Artificial intelligence -- Industrial applications Business logistics -- Data processing Algorithms Order picking systems Stowage -- Data processing UCTD Schroth, Timo Friedrich Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title | Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title_full | Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title_fullStr | Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title_full_unstemmed | Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title_short | Development of a methodology for artificial intelligence-based clustering of stow and pick operations in order to improve logistics performance in small and medium-sized enterprises |
| title_sort | development of a methodology for artificial intelligence based clustering of stow and pick operations in order to improve logistics performance in small and medium sized enterprises |
| topic | Artificial intelligence -- Industrial applications Business logistics -- Data processing Algorithms Order picking systems Stowage -- Data processing UCTD |
| url | https://scholar.sun.ac.za/handle/10019.1/132441 |
| work_keys_str_mv | AT schrothtimofriedrich developmentofamethodologyforartificialintelligencebasedclusteringofstowandpickoperationsinordertoimprovelogisticsperformanceinsmallandmediumsizedenterprises |