面向复杂优化问题的自适应混合进化算法
An Adaptive Hybrid Evolutionary Algorithm for Complex Optimization Problems

本文针对复杂优化问题中单一群智能算法普遍存在的“探索”与“开发”难以平衡的问题，提出了一种基于周期性评估与自适应切换的自适应混合进化算法(AHEA)。该算法融合了遗传算法(GA)的全局探索能力和鹦鹉优化算法(PO)的局部开发优势，通过周期性评估算法性能改进率，并依据设定阈值在GA与PO之间自主切换，实现了算法间层面的智能调度。实验部分在经典基准函数、高维多坑洞Shekel系列函数以及现代IEEE CEC 2022复杂复合函数共10个测试问题上，将AHEA与纯GA、纯PO进行系统性对比。结果表明，AHEA在收敛精度、优化速度、稳定性与鲁棒性方面均表现出显著优势，验证了其在动态平衡探索与开发、有效应对各类复杂优化问题上的先进性与实用性。
This paper addresses the common challenge of balancing “exploration” and “exploitation” in single swarm intelligence algorithms for complex optimization problems by proposing an adaptive hybrid evolutionary algorithm (AHEA) based on periodic evaluation and adaptive switching. The algorithm integrates the global exploration capability of the Genetic Algorithm (GA) and the local exploitation strength of the Parrot Optimization Algorithm (PO). By periodically evaluating the performance improvement rate and autonomously switching between GA and PO based on a predefined threshold, it achieves intelligent scheduling at the inter-algorithm level. In the experimental phase, AHEA is systematically compared with pure GA and pure PO across ten test problems, including classical benchmark functions, high-dimensional multimodal Shekel-series functions, and modern IEEE CEC 2022 complex composite functions. The results demonstrate that AHEA exhibits significant advantages in convergence accuracy, optimization speed, stability, and robustness, confirming its effectiveness in dynamically balancing exploration and exploitation and its practicality in addressing various complex optimization problems.