Improved Slime Mould Algorithm Using Logical Chaos Perturbation and Reference Point Non-Dominated Sorting for Multi-Objective Optimization

The Slime Mould Algorithm (SMA) has gained significant attention from researchers due to its powerful multi-point search capability and its simple and practical structure. Several advanced versions of SMA have been proposed. However, most existing methods primarily focus on the single-objective research domain, leaving the research on multi-objective SMA relatively limited. Additionally, the basic SMA lacks robust global search capability, and extending it to the multi-objective domain often results in a loss of solution diversity. To address these limitations, this paper introduces a general multi-objective SMA framework. It incorporates a logical chaotic single-dimensional perturbation mechanism to enhance individuals’ search traversal in the decision space. Furthermore, a non-dominated sorting mechanism based on the reference point is employed to select a more diverse set of solutions for the subsequent evolution of the next generation. Through experiments conducted on 28 basis functions using seven advanced multi-objective algorithms (CMOPSO, NSGA-II, NSGA-III, MOEAD, PSEA-II, SPEA-II, and NSLS), the results demonstrate that the multi-objective SMA outperforms other algorithms in terms of convergence, accuracy, and diversity.