Artificial Protozoa Optimizer

Abstract

The increasing complexity of real-world challenges has necessitated the development of robust optimization methods. As technological landscapes evolve and societal demands grow more sophisticated, optimization problems have become progressively diverse and intricate, underscoring the theoretical and practical urgency of efficient optimization strategies. Metaheuristic algorithms (MAs), inspired by natural phenomena, have emerged as indispensable tools for addressing these challenges over decades of evolution. However, the “No Free Lunch” (NFL) theorem states that no single metaheuristic can outperform all others across every possible objective function. This theoretical foundation suggests that specific algorithms may exhibit superior performance when tailored to certain problem classes or real-world applications. This motivates researchers to develop new metaheuristics for specific domains. This dissertation proposes a suite of innovative metaheuristic algorithms to solve complex optimization tasks. The core research contributions are as follows: 1. Develops a novel Artificial Protozoa Optimizer (APO) inspired by the biological mechanisms of protozoa for single-objective optimization problems. The algorithm uniquely formalizes foraging, dormancy, and reproduction behaviors into a computational framework. Different from previously proposed metaheuristics, APO introduces four specialized search operators: autotrophic foraging, heterotrophic foraging, dormancy, and reproduction. The efficacy of APO is rigorously validated against an extensive benchmark suite, with experimental results demonstrating superior performance compared to state-of-the-art algorithms. 2. Proposes an innovative Membrane Algorithm based on Artificial Protozoa Optimizer (MAAPO), which enhances the global search capabilities of the original APO. First, MAAPO integrates the parallel and distributed computational paradigm of Membrane Computing (MC) to facilitate diverse information exchange. Second, the autotrophic foraging model is refined using a Roulette Fitness-Distance Balance (RFDB) selection mechanism, which identifies promising reference points to guide the search process more efficiently. 3. Formulates a Multi-Objective Artificial Protozoa Optimizer (MOAPO) tailored for complex multi-objective optimization problems. MOAPO employs non-dominated sorting and crowding distance mechanisms for population ranking to balance exploration and exploitation. Crucially, the weight factors for autotrophic and heterotrophic foraging are dynamically updated based on population ranking rather than original fitness values. Finally, an environmental selection strategy is integrated to curate the population for subsequent iterations, facilitating the discovery of a high-quality and well-distributed set of Pareto optimal solutions. 4. Introduces a Surrogate-Assisted Artificial Protozoa Optimizer (SAAPO) designed to address computationally expensive optimization problems where fitness evaluations are cost-prohibitive. This dissertation develops four innovative metaheuristics tailored for varied optimization landscapes. APO and MAAPO are designed for single-objective optimization, MOAPO addresses multi-objective optimization, and SAAPO targets expensive optimization scenarios. Furthermore, these algorithms are successfully applied to complex real-world domains, including engineering design, image segmentation, feature selection, and robotic arm trajectory optimization.