Self-Adaptive White Shark Multi-Objective Optimization Approach with Mutual Information Based Knowledge Discovery for Reconfigurable Manufacturing Systems

Manufacturing companies frequently experience changes in production volumes and shorter product lifecycles in the competitive and unstable market of today. RMS is essential for overcoming these obstacles. Because of the many trade-offs involved, optimizing efficiency in Reconfigurable Manufacturing Systems (RMS) is a difficult task for stakeholders and decision-makers, despite its advantages. A unique Self-Assisted White Shark Optimization (SAWSO) approach with Mutual Information based knowledge discovery is developed for Reconfigurable Manufacturing Systems (RMS) to address this problem. The optimization of task assignments, resource distribution across workstations, and RMS buffer capacities are the main topics of this study. This study’s main goal is to investigate how work tasks and resources are distributed among workstations within an RMS. To achieve optimal resource allocation and buffer capacity allocation in the RMS, the SAWSO algorithm is employed in this work. This algorithm optimally assigns tasks to workstations while considering constraints such as maximizing throughput, minimizing total buffer capacity, migration cost, execution time, and makespan. To enhance the resource allocation efficiency, an improved knowledge discovery based on MI is introduced. This involves the use of an improved Mutual Information (IMI) approach. Mutual Information (MI) is then computed for each objective function (migration cost, buffer capacity, and throughput) with corresponding fitness values. The Harris Hawks Optimization (HHO) is 6.575, Ant Colony Optimization (ACO) is 7.794, Bald Eagle Search (BES) is 2.716, Butterfly Optimization (BOA) is 2.201, Hunger Game Search (HGS) is 3.222, Arithmetic Optimization Algorithm (AOA) is 2.795, simulation-based multi-objective optimization nondominated sorting genetic algorithm (SMO-NSGA) is 2.899, and Multi-Objective Particle Swarm Optimization (MOPSO) is 3.367, respectively. The fitness value is calculated for all constraints, and their thresholds are determined by taking the mean of their fitness values. Based on the specific conditions, resources are allocated to workstations efficiently, categorized as excellent, good, medium, and worst, depending on the defined metrics.