Optimizing Swarm Intelligence: A Comprehensive Analysis of Mutation-Based Enhancements

Swarm Intelligence (SI) represents an optimization approach inspired by the collective behavior observed in swarms during the search for food. Well-established SI methods, such as Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and Artificial Bee Colony (ABC), are complemented by newer methodologies like Cat Swarm Optimization (CSO) and Grasshopper Optimization Algorithm (GOA). Typically, exploration techniques in SI are more effective than exploitation techniques. To enhance exploration capabilities, this research employs a modification technique based on mutation, chosen for its strong exploratory attributes and low complexity. This study introduces 16 modifications by combining four frameworks with four operators. Each modification is paired with the fundamental methods for comprehensive testing. The experimental phase encompasses five benchmark functions of varying dimensions, resulting in 8,000 experiments. Three analytical assessments were conducted based on these results. The initial analysis reveals that the mutation modification has the most substantial impact on the basic ACO method. The second analysis indicates that mutation modification significantly influences the objective function in scenarios with large dimensions. The concluding analysis highlights the paramount influence of the modification incorporating the random parameter mutation framework, whereas the mutation operator modification shows comparatively less significant results. A detailed impact assessment shows that Modification 2B achieved the highest number of positive results, succeeding in 69 out of 100 tests, while 2D modifications yielded the smallest sum and average values. The influence of different frameworks and operators was further analyzed, revealing that frameworks have a more pronounced impact on performance than operators. Framework number 2, in particular, demonstrated the most significant effect on improving average impact values.