OPTIMIZATION OF SCHEDULING TECHNIQUES IN FMS IN THE CONTEXT OF INDIAN INDUSTRY

Abstract

In today’s rapid and highly demanding production environment, manufacturing enterprises are increasingly adopting Flexible Manufacturing Systems (FMS) to enhance productivity, responsiveness, and operational efficiency. An FMS represents an integrated and automated manufacturing environment that combines programmable machine tools, automated material handling systems, automated storage and retrieval units, and centralized computer control to efficiently manufacture a wide variety of components. As a cornerstone of Industry 4.0, FMS plays a critical role in enabling intelligent, data-driven, and adaptive manufacturing systems capable of responding to dynamic market demands. A fundamental challenge in FMS operation is scheduling optimization, which involves determining the optimal sequence of jobs while allocating limited and interdependent resources such as machines, tools, and Automated Guided Vehicles (AGVs). Effective scheduling significantly enhances system performance by reducing make span, minimizing tardiness, improving resource utilization, lowering operational costs, and increasing throughput. In the context of Industry 4.0, scheduling optimization becomes even more crucial, as it directly supports the realization of smart, autonomous, and energy-efficient manufacturing environments. Traditional optimization approaches such as linear programming, dynamic programming, and exact mathematical models have been widely used for manufacturing scheduling. However, due to the combinatorial complexity, non-linearity, multi-objective nature, and large search spaces inherent in FMS, these methods often become computationally infeasible for real-world applications. In contrast, metaheuristic algorithms, inspired by natural and evolutionary processes, provide robust and scalable alternatives. Their ability to balance exploration and exploitation, handle multiple conflicting objectives, and adapt to complex constraints makes them particularly suitable for FMS scheduling problems. With recent advances in computational power and intelligent optimization strategies, metaheuristics have emerged as powerful tools for addressing large-scale industrial scheduling challenges. In this research, three novel hybrid metaheuristic methodologies are proposed to address simultaneous scheduling problems in FMS under Industry 4.0 paradigms: 1. Dynamic-Particle Multi-Swarm Optimization (Dy-PSO) 2. Novel Variant of Particle Swarm Optimization (NvPSO) and Walrus Optimization Algorithm (WaOA) 3. Novel Genetic and Adaptive Artificial Bee Colony Algorithm (NG-AABCA) vii The first study introduces Dynamic-Particle Multi-Swarm Optimization (Dy-PSO), a novel scheduling framework primarily focused on makespan minimization in FMS. Dy-PSO employs multiple interacting swarms with adaptive parameter control to prevent premature convergence and stagnation. A spatial exclusion strategy is incorporated to avoid redundant exploration of previously visited regions of the solution space. A significant methodological contribution of this study is the integration of machine learning, where a Random Forest Regressor, enhanced through Genetic Algorithm–based learning, is used to predict and guide scheduling decisions. This hybridization of evolutionary optimization and predictive learning establishes a data-driven scheduling framework that enhances solution quality and convergence speed. Dy-PSO is evaluated under three realistic scheduling scenarios: (i) simultaneous scheduling of jobs, tools, and AGVs, (ii) scheduling without tool constraints, and (iii) scheduling integrated with machine learning assistance. Comparative analysis with conventional PSO demonstrates substantial reductions in makespan and superior robustness across all scenarios. The second study focuses on energy-aware scheduling through a Novel Variant of PSO (NvPSO) and the Walrus Optimization Algorithm (WaOA). This study addresses the growing need for sustainable manufacturing by simultaneously minimizing makespan, tardiness, and total energy consumption. NvPSO incorporates logistic map–based parameter tuning, enhancing diversity and search efficiency. Experimental results across 13 diverse job sets show that NvPSO achieves up to 9% reduction in energy consumption while completely eliminating tardiness penalties, outperforming conventional algorithms such as the Artificial Immune System (AIS) and Modified Genetic Tabu Algorithm (MGTA). While WaOA demonstrates faster convergence and lower computational complexity, NvPSO consistently delivers superior solution quality for larger and more complex scheduling instances, highlighting its suitability for energy-efficient Industry 4.0 manufacturing systems. The third study proposes the Novel Genetic and Adaptive Artificial Bee Colony Algorithm (NG- AABCA) for minimizing makespan, penalty costs, and total tardiness. NG-AABCA introduces cognitive (ε₁) and social (ε₂) learning mechanisms, which are generally underutilized in classical ABC algorithms, to exploit global knowledge and enhance convergence behavior. The algorithm further integrates Genetic Algorithm elitism and Random-Restart Hill-Climbing, effectively balancing solution diversity and intensification. Computational results demonstrate that NG- AABCA achieves a 5.3% reduction in makespan and an 8.7% reduction in tardiness compared to conventional metaheuristics, resulting in improved productivity and more efficient utilization of manufacturing resources. viii Extensive computational experiments were conducted using MATLAB R2019a on an Intel Core™ i7 platform, and the proposed algorithms were validated across benchmark datasets as well as realistic FMS configurations. The results confirm that the suggested hybrid metaheuristic approaches consistently outperform traditional optimization methods in terms of solution quality, convergence speed, robustness, and scalability. In several cases, the algorithms identified new best-known makespan values, demonstrating their effectiveness in exploring high-quality solution spaces. Overall, this research makes significant contributions by developing adaptive, hybrid, and energy-aware scheduling frameworks that address the complexities of simultaneous resource scheduling in FMS. The integration of metaheuristics, machine learning, and sustainability objectives positions the proposed approaches as powerful decision-support tools for Industry 4.0–enabled smart manufacturing, offering both theoretical advancements and strong potential for industrial applicability.