Integrated Scheduling Framework for Flexible Job Shops: Hybrid Metaheuristics for Static and Dynamic Conditions in Real-World Environments

다국어 초록 (Multilingual Abstract)

In modern manufacturing environments, effective production planning plays an important role in optimizing various manufacturing process components. Scheduling is the critical part of this planning, since it determines the optimal solution that allocates jobs and operations to available resources subject to constraints. To represent real-world manufacturing flexibility, the flexible job shop scheduling problem (FJSP) extends the classic job shop scheduling problem (JSSP) by allowing multiple machine choices for each operation, each with a different processing time, thereby increasing the computational complexity and realism, making it an NP-hard problem. Traditional scheduling methods are insufficient to address this complexity, as they fail to adapt dynamically to real-time uncertainties such as machine breakdowns or urgent job arrivals. This gap motivates the need for an intelligent and hybrid optimization framework that can deliver flexible, efficient, and sustainable scheduling solutions under realistic industrial conditions. Considering real-world scenarios, manufacturing systems frequently face dynamic conditions such as unexpected job arrivals and machine breakdowns, which lead to dynamic flexible job shop scheduling (DFJSP). This dissertation presents two integrated research streams focused on optimizing Flexible Job Shop Scheduling Problems (FJSP) under both static and dynamic environments, for single and multiple criteria, using hybrid metaheuristic algorithms. The initial research presents a hybrid metaheuristic algorithm that integrates Genetic Algorithm (GA), Simulated Annealing (SA), and Variable Neighborhood Search (VNS) to efficiently solve both static and dynamic FJSP. The GA has the capability for global exploration, and it is integrated with the local search algorithms SA and VNS, which ensure accelerated convergence and the avoidance of local optima. Experiments on 40 benchmark datasets revealed that the proposed GA–SA–VNS framework achieved superior performance, outperforming traditional approaches in 38 cases. Additionally, the algorithm was extended to dynamic scheduling by adapting a rescheduling strategy for the single-objective DFJSP to events such as machine breakdowns and new job arrivals, resulting in a significant reduction in makespan and improved optimization under dynamic conditions. The other research topic extends the framework to address multi-criteria optimization by including real-world factors such as due dates, Sequence-Dependent Setup Time, and variability in processing times, which is referred to as Multi-Criteria Flexible Job Shop Scheduling MC- FJSP. To assign the due date to the standard FJSP dataset, the Total Work Content (TWK) was adapted for the extension of the dataset. A hybrid GA-VNS algorithm was developed, integrating a multi-population strategy based on dispatching rules to diversify the population initialization. The objectives are to minimize the makespan, total tardiness, and total setup time, demonstrating improvements in static scheduling across two different environments compared to baseline GA results. The dynamic events, such as machine breakdowns and job arrivals, are demonstrated across multiple criteria. Overall, the study established an integrated scheduling framework with the real-world manufacturing environment. At last, the web-based FJSP platform is designed to demonstrate how manufacturing industries utilize the FJSP principle for an order-driven production process. The developed platform bridges the gap between theoretical and industrial implementation by transforming the optimization algorithm. The system provides comprehensive machine utilization analysis with interactive Gantt chart visualizations and real-time order completion tracking to support dynamic resource allocation based on customer demands. It enhances the visibility of scheduling decisions, improves production transparency, and supports smart manufacturing initiatives.

목차 (Table of Contents)

• Chapter 1. Introduction 1
• 1.1. Motivation 2
• 1.2. Contribution 3
• 1.3. Structure of Thesis 5
• Chapter 2. Literature Review 7
• 2.1. Flexible Job Shop Scheduling Problem 7
• 2.2. Solution Approaches for FJSP 10
• 2.3. Initial Population Aproach 14
• 2.4. Dynamic FJSP 15
• 2.5. Multi-Criteria FJSP 17
• 2.6. Application Cases of FJSP 18
• 2.7. Analysis of Research Gaps 19
• Chapter 3. Single-Objective Flexible Job Shop Scheduling Under Static Conditions 21
• 3.1. Problem Formulation 21
• 3.2. Assumptions 24
• 3.3. Constraints 25
• 3.3.1. Job Precedence Constraint 25
• 3.3.2. Machine Assignment Constraint 25
• 3.3.3. Non-Overlap Constraint 26
• 3.3.4. Non-Negativity Constraint 26
• 3.3.5. Processing Time Constraint 26
• 3.4. Dataset 27
• 3.5. Proposed Hybrid Metaheuristic Algorithm 29
• 3.5.1. Encoding 30
• 3.5.2. Initial Population 30
• 3.5.3. Selection 31
• 3.5.4. Cross-over Operator 32
• 3.5.5. Local Search 33
• 3.5.6. Stopping Criteria 35
• 3.5.7. Connectivity to DFJSP 35
• 3.6. Results and Discussion 36
• Chapter 4. Multi-Criteria Flexible Job Shop Scheduling Under Static Conditions 39
• 4.1. Problem Formulation 39
• 4.1.1. Multiple Criteria 40
• 4.1.2. Multiple Objectives 42
• 4.2. Constraints for MCFJSP 44
• 4.2.1. Precedence Constraint 44
• 4.2.2. Machine Capacity Constraint 44
• 4.2.3. Machine Assignment Constraint 44
• 4.2.4. Setup Time Constraint 45
• 4.2.5. Non-Negative Constraint. 45
• 4.3. Dataset for MC-FJSP 46
• 4.3.1. MPM Environment 46
• 4.3.2. SPM Environment 49
• 4.4. Proposed Algorithm for MCFJSP 52
• 4.4.1. Encoding 53
• 4.4.2. Multi-Strategy Population 54
• 4.4.3. GA Operator: Selection 56
• 4.4.4. GA Operator: Crossover 56
• 4.4.5. Scheduling Operations 56
• 4.4.6. Fitness Function 59
• 4.4.7. Local Search 60
• 4.4.8. Termination Criteria for HA 62
• 4.4.9. Connectivity to DFJSP 62
• 4.5. Results and Discussion 63
• 4.5.1. MPM Environment 63
• 4.5.2. SPM Environment 67
• Chapter 5. Dynamic Flexible Job Shop Scheduling 70
• 5.1. Job Arrival 70
• 5.1.1. Proposed Re-scheduling Method 70
• 5.2. Machine Breakdown 74
• 5.2.1. Proposed Re-scheduling Method 74
• 5.3. Results and discussion 79
• 5.3.1. Job Arrival 79
• 5.3.2. Machine Breakdown 84
• Chapter 6. Order Driven FJSP-based platform 89
• 6.1. User Module 89
• 6.1.1. User Authentication and Access 89
• 6.1.2. User Dashboard 90
• 6.2. Producer Module 92
• 6.2.1. All Order Analysis 92
• 6.2.2. Factory Environment 94
• 6.2.3. Optimization Result with Scheduling Analysis 95
• 6.3. Discussion 97
• Chapter 7. Conclusion and Discussion 98
• References: 100
• Acknowledgements 111
• List of Publications 112