Applied Soft Computing ( IF 8.7 ) Pub Date : 2021-09-04 , DOI: 10.1016/j.asoc.2021.107866 Sanjoy Chakraborty 1, 2 , Sushmita Sharma 3 , Apu Kumar Saha 3 , Sandip Chakraborty 4
Differential evolution and its variants have already proven their worth in the field of evolutionary optimization techniques. This study further enhances the success history-based adaptive differential evolution (SHADE) by hybridizing it with a modified Whale optimization algorithm (WOA). In the new algorithm, the two algorithms, SHADE and modified WOA, carry out the search process independently and share information like the global best solution and whole population and thus guides both the algorithms to explore and exploit new promising areas in the search space. It also reduces the chance of being trapped in local optima and stagnation. The proposed algorithm (SHADE–WOA) is tested, evaluating CEC 2017 functions using dimensions 30, 50, and 100. The results are compared with modified DE algorithms, namely SaDE, SHADE, LSHADE, LSHADE-SPACMA, and LSHADE-cnEpSin, also with modified WOA algorithms, namely ACWOA, AWOA, IWOA, HIWOA, and MCSWOA. The new algorithm’s efficiency in solving real-world problems is examined by solving two unconstrained and four constrained engineering design problems. The performance is verified statistically using non-parametric statistical tests like Friedman’s test and Wilcoxon’s test. Analysis of numerical results, convergence analysis, diversity analysis, and statistical analysis ensures the enhanced performance of the proposed SHADE–WOA.
中文翻译:
SHADE-WOA:一种用于全局优化的元启发式算法
差分进化及其变体已经在进化优化技术领域证明了它们的价值。这项研究通过将其与改进的 Whale 优化算法 (WOA) 混合,进一步增强了基于成功历史的自适应差分进化 (SHADE)。在新算法中,SHADE和改进的WOA两种算法独立进行搜索过程,共享全局最佳解和整体种群等信息,从而引导两种算法在搜索空间中探索和开发新的有前景的领域。它还减少了陷入局部最优和停滞的机会。测试提出的算法 (SHADE-WOA),使用维度 30、50 和 100 评估 CEC 2017 函数。 结果与修改后的 DE 算法进行比较,即 SaDE、SHADE、LSHADE、LSHADE-SPACMA,和 LSHADE-cnEpSin,也有改进的 WOA 算法,即 ACWOA、AWOA、IWOA、HIWOA 和 MCSWOA。通过解决两个无约束和四个约束的工程设计问题来检验新算法在解决实际问题中的效率。使用非参数统计检验(如 Friedman 检验和 Wilcoxon 检验)对性能进行统计验证。数值结果分析、收敛分析、多样性分析和统计分析确保了所提出的 SHADE-WOA 的增强性能。使用非参数统计检验(如 Friedman 检验和 Wilcoxon 检验)对性能进行统计验证。数值结果分析、收敛分析、多样性分析和统计分析确保了所提出的 SHADE-WOA 的增强性能。使用非参数统计检验(如 Friedman 检验和 Wilcoxon 检验)对性能进行统计验证。数值结果分析、收敛分析、多样性分析和统计分析确保了所提出的 SHADE-WOA 的增强性能。