International Joint Laboratory of Intelligent Network Theory and Key Technology, Henan University, Kaifeng, China.
College of Software, Henan University, Kaifeng, China.
Sci Rep. 2023 Jul 4;13(1):10768. doi: 10.1038/s41598-023-37958-5.
Tree-seed algorithm is a stochastic search algorithm with superior performance suitable for solving continuous optimization problems. However, it is also prone to fall into local optimum and slow in convergence. Therefore, this paper proposes an improved tree-seed algorithm based on pattern search, dimension permutation, and elimination update mechanism (PDSTSA). Firstly, a global optimization strategy based on pattern search is used to promote detection ability. Secondly, in order to maintain the diversity of the population, a random mutation strategy of individual dimension replacement is introduced. Finally, the elimination and update mechanism based on inferior trees is introduced in the middle and later stages of the iteration. Subsequently, PDSTSA is compared with seven representative algorithms on the IEEE CEC2015 test function for simulation experiments and convergence curve analysis. The experimental results indicate that PDSTSA has better optimization accuracy and convergence speed than other comparison algorithms. Then, the Wilcoxon rank sum test demonstrates that there is a significant difference between the optimization results of PDSTSA and each comparison algorithm. In addition, the results of eight algorithms for solving engineering constrained optimization problems further prove the feasibility, practicability, and superiority of PDSTSA.
树种子算法是一种具有优越性能的随机搜索算法，适用于求解连续优化问题。然而，它也容易陷入局部最优解，收敛速度较慢。因此，本文提出了一种基于模式搜索、维度置换和淘汰更新机制的改进树种子算法（PDSTSA）。首先，采用基于模式搜索的全局优化策略来提高检测能力。其次，为了保持种群的多样性，引入了个体维度替换的随机突变策略。最后，在迭代的中后阶段引入了基于劣树的淘汰和更新机制。随后，在 IEEE CEC2015 测试函数上，将 PDSTSA 与七种代表性算法进行了仿真实验和收敛曲线分析。实验结果表明，PDSTSA 具有比其他比较算法更好的优化精度和收敛速度。然后，Wilcoxon 秩和检验表明，PDSTSA 与每个比较算法的优化结果之间存在显著差异。此外，八种算法求解工程约束优化问题的结果进一步证明了 PDSTSA 的可行性、实用性和优越性。
