| 摘要: |
| 针对传统烟花算法及其混合变体中存在的早熟收敛和种群多样性不足等问题，本文提出了一种基于引导向量的动态混合烟花算法。该算法在爆炸操作中引入了一种基于随机位移的动态搜索策略，结合动态调整的爆炸幅度，改进火花生成机制，从而有效避免过早收敛至局部最优，同时增强种群的多样性。在变异操作中，通过引导向量引导最优火花向有前景的方向偏移，生成变异火花，提升整体搜索效率并增强跳出局部最优的能力。此外，针对非最优火花的局部开发能力不足，采用差分变异、交叉和选择操作优化失败者锦标赛机制中的失败烟花，以进一步提高寻优能力。本文还对算法的收敛性进行了理论证明。通过在CEC2017测试集函数上的仿真实验、消融实验及解决置换流水车间调度问题的应用，验证了算法在精度、多样性保持及跳出局部最优方面的优越性及各项改进措施的有效性。 |
| 关键词: 烟花算法 差分进化 引导变异火花 CEC2017 |
| DOI： |
| 投稿时间：2025-02-22修订日期：2025-06-20 |
| 基金项目:国家自然科学基金项目（面上项目，重点项目，重大项目） |
|
| GHFWA：A Dynamic Hybrid Fireworks Algorithm with Guided Vector |
|
shang jian ping, yv sen wu, nie cong, song zhi yong, ruan xiang wei, zhuang zi xun
|
| (Hubei Radio Monitoring Center) |
| Abstract: |
| To address the issues of premature convergence and insufficient population diversity in traditional fireworks algorithms and general hybrid fireworks algorithms, this paper proposes a dynamic hybrid fireworks algorithm with guided vectors. The algorithm introduces a dynamic search strategy with random displacement in the explosion operation, combined with dynamically adjusted explosion amplitudes, to improve the spark generation mechanism. This effectively avoids premature convergence to local optima while enhancing population diversity. In the mutation operation, guiding vectors are used to direct optimal fireworks toward promising directions, generating mutated sparks to improve overall search efficiency and strengthen the ability to escape local optima. Furthermore, to address the insufficient local exploitation capability of non-optimal fireworks, differential mutation, crossover, and selection operations are applied to optimize the loser fireworks in the tournament mechanism, further enhancing optimization performance. The paper also provides a theoretical proof of the algorithm's convergence. The convergence of the algorithm is then theoretically proven. Finally, through simulation experiments, ablation experiments on the CEC2017 test set functions, and applications in solving the permutation flowshop scheduling problem, the superiority and effectiveness of the proposed algorithm in terms of precision, diversity maintenance, and escaping local optima are validated. |
| Key words: Fireworks algorithm Differential evolution Guided mutation spark CEC2017 |
