Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Multi-Objective Resource-Constrained Project Scheduling: A Comparative Study of Metaheuristic Approaches: Balancing Cost, Time & Resources in Telecom Project Scheduling
Blekinge Institute of Technology, Faculty of Engineering, Department of Industrial Economics.
Blekinge Institute of Technology, Faculty of Engineering, Department of Industrial Economics.
2025 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesisAlternative title
Flerobjektiv Resursberänsad Projektschemaläggning: En Jämförelsestudie av Metaheuristiska Metoder : Att Balansera Kostnad, Tid och Resurser i Schemaläggning av Telekomprojekt (Swedish)
Abstract [en]

Background. Real-world projects often need to be optimized in several objectives simultaneously while adhering to resource constraints. The multi-objective resource-constrained project scheduling problem (MO-RCPSP) models these trade-offs but is computationally difficult to solve exactly as project size increases (in theoretical computer science it is defined as a NP-hard-problem). This motivates using metaheuristic approaches that generate diverse schedules which explore the different optimal objective-trade-offs (Pareto-optimal solutions). 

Objective. A comparative evaluation of three metaheuristic algorithms: Multi-Objective Harris Hawks Optimization (MOHHO), Particle Swarm Optimization (MOPSO) and Ant Colony Optimization (MOACO), alongside the control algorithms; NSGAII and SPEA2. The algorithms will optimize schedules across small, medium and large scheduling instances with regard to cost, duration and resource utilization.

Method. A unified tri-objective formulation was implemented together with three metaheuristic algorithms. Hyperparameter tuning was performed via grid search on a medium-sized instance under a 30-second time budget. The optimal configurations found were then used for the main experiment. In the main experiment each algorithm executed ten independent 30-second trials per instance, maintaining an archive of non-dominated solutions. Performance was assessed with the metaheuristic performance metrics: normalized hypervolume, generational distance, spread and coverage, supported by three-dimensional Pareto visualizations.

Result. On small instances (10 tasks), MOPSO achieved the highest normalized hypervolume (0.737) and best diversity. On medium instances (50 tasks), NSGAII led in convergence and coverage (NHV = 0.719, coverage = 0.734). On a large real-world instance (147 tasks), NSGAII remained dominant (NHV = 0.670) while MOHHO achieved the best generational distance, indicating strong convergence. MOACO underperformed at all scales.

Conclusion. Metaheuristic performance depends on problem scale. Swarm-based strategies excel on small problems, elitist evolutionary methods perform best at medium scale and adaptive exploration–exploitation mechanisms become advantageous in large, high-dimensional scheduling. Algorithm selection should align with project complexity and computational constraints.

Abstract [sv]

Bakgrund. I praktiska projekt behöver ofta flera objektiv optimeras samtidigt under givna resursbegränsningar. Det resursbegränsade flerobjektivs-schemaläggningsproblemet (RCPSP) modellerar dessa konflikter mellan objektiv men är beräkningsmässigt komplext att fullständigt lösa när projektstorleken ökar (inom teoretisk datavetenskap är problemet definierat som NP-svårt). Det motiverar användningen av metaheuristiska metoder som genererar varierande scheman som utforskar olika objektiv-optimala lösningar (Pareto-optimal lösningar).

Syfte. Att jämföra tre metaheuristiska algoritmer: Multi-Objective Harris Hawks Optimization (MOHHO), Particle Swarm Optimization (MOPSO) och Ant Colony Optimization (MOACO), samt de etablerade kontroll-algoritmerna NSGAII och SPEA2, på schemaläggningsproblem i tre storlekar: små, medelstora och stora instanser med hänsyn till tid, kostnad och resursutnyttjande.

Metod. En enhetlig tri-objektiv formulering implementerades tillsammans med de tre metaheuristiska algoritmerna. Hyperparameter-optimering genomfördes med “grid search” på en medelstor instans under en 30-sekunders tidsbudget. De optimala konfigurationerna som identifierades användes sedan i huvudexperimentet. I huvudexperimentet genomförde varje algoritm tio oberoende 30-sekundersförsök per instans, samtidigt som ett arkiv över icke-dominerade lösningar sparades. Prestandan utvärderades med metaheuristiska utvärderingsmått: normaliserad hypervolym, generationsavstånd, spridning och täckning, kompletterat med tredimensionella Pareto-visualiseringar.

Resultat. På små instanser (10 aktiviteter) uppnådde MOPSO högst normaliserad hypervolym (0,737) och bäst spridning. På medelstora instanser (50 aktiviteter) presterade NSGAII bäst i konvergens och täckning (NHV = 0,719; coverage = 0,734). På en stor verklighetsinstans (147 aktiviteter) förblev NSGAII dominerande (NHV = 0,670), medan MOHHO uppnådde lägst generational distance, vilket indikerar stark konvergens. MOACO låg i botten i alla skalor.

Slutsatser. Val av metaheuristisk metod bör styras av problemdimension. Svärmbaserade strategier är effektiva för små problem, elitistiska evolutionära algoritmer presterar bäst på medelstora instanser och adaptiva exploration–exploitation-mekanismer ger fördelar i stora, högdimensionella schemaläggningar.

Place, publisher, year, edition, pages
2025. , p. 77
Keywords [en]
Multi-objective resource constrained project scheduling problem, Metaheuristics, Hyperparameter tuning, Normalized hypervolume, Algorithm comparison, Harris Hawks Optimization, Ant Colony Optimization, Particle Swarm Optimization, HHO, ACO, PSO, MOHHO, MOACO, MOPSO, Pareto Front, MO-RCPSP
National Category
Algorithms Computer Sciences
Identifiers
URN: urn:nbn:se:bth-27947OAI: oai:DiVA.org:bth-27947DiVA, id: diva2:1962540
External cooperation
Ericsson
Subject / course
Degree Project in Master of Science in Engineering 30,0 hp
Educational program
IEACI Master of Science in Industrial Management and Engineering
Presentation
2025-05-22, J1620, Valhallavägen 10, Karlskrona, 10:00 (Swedish)
Supervisors
Examiners
Available from: 2025-06-16 Created: 2025-05-31 Last updated: 2025-09-30Bibliographically approved

Open Access in DiVA

fulltext(3126 kB)226 downloads
File information
File name FULLTEXT01.pdfFile size 3126 kBChecksum SHA-512
26f914be73b803f1298bda59832fb046d517f04075f5c49b6e4d0261c5b0a4a88c3ab464011ccbb604ccffb4bd6e6b9ffda1416ee40a0843e68ab7e24ee21b46
Type fulltextMimetype application/pdf

By organisation
Department of Industrial Economics
AlgorithmsComputer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 227 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 276 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf