Marine Predators Algorithm (MPA): A Nature-Inspired Optimization Approach

The Marine Predators Algorithm (MPA) is a nature-inspired optimization algorithm that follows the natural rules governing optimal foraging strategies and implements velocity strategies observed in predator-prey interactions within marine ecosystems. The algorithm's implementation typically involves three main phases based on different velocity ratios between predators and prey, effectively balancing exploration and exploitation throughout the optimization process. To comprehensively evaluate MPA's performance, we conducted tests using 29 benchmark functions, the CEC-BC-2017 test suite, randomly generated landscapes, three engineering benchmarks, and two practical engineering design problems related to ventilation and building energy efficiency. The algorithm's core mechanism involves simulating predator movement patterns using Brownian and Lévy walk strategies, with position updates governed by velocity ratios that change according to iteration cycles. Through comparative analysis with three categories of existing optimization methods, MPA demonstrates significant advantages across multiple dimensions. First, compared to traditional metaheuristic algorithms like GA and PSO, MPA more effectively follows natural patterns and exhibits superior performance in problem-solving scenarios, particularly in its adaptive transition between exploration and exploitation phases. Second, when measured against recently developed algorithms such as GSA, CS, and SSA, MPA shows improved performance metrics and better convergence characteristics. Finally, in comparison with high-performance optimizers including CMA-ES, SHADE, and LSHADE-cnEpSin, MPA has achieved competitive results in IEEE CEC competitions, demonstrating its robustness in handling complex optimization landscapes. In summary, the Marine Predators Algorithm represents a highly promising optimization technique with broad application potential across various problem domains. The algorithm's implementation typically features efficient memory storage for elite solutions and incorporates fitness-based position updates that enhance solution quality. With further research and refinement, MPA is expected to find applications in numerous fields, providing improved solutions for complex real-world engineering challenges.