A Detailed Interpretation of An Improved Artificial Lemming Algorithm and Its Application in Engineering Problems

This paper proposes an improved artificial lemming algorithm (IMALA), a metaheuristic optimization technique based on biological principles, and explores its application in solving engineering optimization problems. To overcome the limitations of the ALA, IMALA introduces three key improvements: population initialization based on cubic chaos mapping to improve diversity, a dynamic differential mutation strategy to achieve adaptive exploration and exploitation balance, and a stochastic adversarial learning mechanism using beta distribution to accelerate convergence and improve robustness. Under the reported iteration-based setting, IMALA achieved favorable average ranks of 1.69 and 1.58 in the nine-algorithm comparisons on CEC2017 and CEC2022, respectively, although it did not obtain the best result on every benchmark function. For practical verification, IMALA is applied to three engineering problems: a step-cone pulley with minimization weight, power loss reduction in a hydrostatic thrust bearing, and piston lever design with minimization oil volume optimization. Across the three engineering optimization problems, IMALA demonstrates competitive solution quality and relatively stable performance under the adopted experimental settings. Under the same iteration limit, the fast convergence and stability of this algorithm make it a reliable tool for solving complex, high-dimensional and multimodal engineering problems, and it has significant improvements over existing methods both in theory and practical applications.