Abstract: In order to improve the development accuracy and global search ability of chimpanzee optimization algorithms,this paper establishes a mathematical model by observing and summarizing the behavioral characteristics of children aged 4~6 in kindergarten. Furthermore, a new metaheuristic algorithm is proposed: kindergarten children optimization algorithm (KCOA). In terms of algorithm design, kindergarten children have three social behaviors: toy attraction, finding partners, and little red flowers. The toy attraction strategy reflects the leading role of the optimal individual. The strategy of finding partners increases mutual communication among ordinary children and enhances their ability to independently explore population space, avoiding causing the entire population to fall into local extreme points and search stagnation due to incorrect judgment of the optimal individual. The little red flower strategy evaluates the current location status of children in real time, updates inferior solutions in a timely manner, improves algorithm convergence speed and optimization accuracy.Comparative analysis, Wilcoxon rank sum statistical test, Friedman ranking, and partial CEC2014 testing are conducted on the optimization of 23 benchmark test functions. The KCOA algorithm has significant advantages in development accuracy and optimization stability compared to optimization algorithms such as chimpanzees, two improved chimpanzee algorithms, and particle swarm optimization. Finally, the effectiveness of the algorithm is confirmed through two engineering problems. The kindergarten children optimization algorithm reduces the optimal cost of spring and reducer design problems by 0.85% and 2.13%, respectively, compared to the chimpanzee optimization algorithm.

Key words: kindergarten children optimization algorithm, toy attraction strategy, partner finding strategy, little red flower strategy

摘要： 为了提升黑猩猩优化算法开发精度和全局搜索能力，通过观察、总结4~6岁幼儿园小朋友行为特点，建立数学模型，提出一种新的元启发式算法：幼儿园小朋友优化算法（kindergarten children optimization algorithm,KCOA）。在算法设计上，幼儿园小朋友拥有玩具吸引、找伙伴、小红花三种社会行为。玩具吸引策略体现最优个体的引领作用。找伙伴策略增加普通小朋友间相互交流、提升普通小朋友个体自主探索种群空间能力，避免因最优个体错误判断让整个种群陷入局部极值点、搜索停滞。小红花策略实时评估当前小朋友位置状态，及时更新劣解、提升算法收敛速度和寻优精度。通过对23个基准测试函数的寻优对比分析、Wilcoxon秩和统计检验、Friedman排名以及部分CEC2014测试函数寻优结果对比，KCOA算法相比黑猩猩、两种改进黑猩猩、粒子群等优化算法在开发精度和寻优稳定性上都具有显著优势。最后，通过两个工程问题证实了该算法的有效性。KCOA算法求解弹簧、减速器设计问题对比黑猩猩优化算法最优开销分别减少0.85%、2.13%。

关键词: 幼儿园小朋友优化算法, 玩具吸引策略, 找伙伴策略, 小红花策略