融合经验反思机制的教与学优化算法

PDF下载HTML

Title:

Teaching and learning optimization algorithm based on empirical reflection mechanism

作者:

吴迪¹, 贾鹤鸣², 刘庆鑫³, 齐琦³, 王爽²

1. 三明学院 教育与音乐学院, 福建 三明 365004;
2. 三明学院 信息工程学院, 福建 三明 365004;
3. 海南大学 计算机科学与技术学院, 海南 海口 570228

Author(s):

WU Di¹, JIA Heming², LIU Qingxin³, QI Qi³, WANG Shuang²

1. School of Education and Music, Sanming University, Sanming 365004, China;
2. School of Information Engineering, Sanming University, Sanming 365004, China;
3. School of Computer Science and Technology, Hainan University, Haikou 570228, China

关键词:

教与学优化算法; 经验反思机制; 动态自适应权重; 元启发式算法; 基准函数; 压力容器设计问题; 焊接梁设计问题; Wilcoxon秩和检验

Keywords:

teaching and learning-based optimization; empirical reflection mechanism; dynamic adaptive weight; meta-heuristic algorithm; benchmark functions; pressure vessel design problem; welded beam design problem; Wilcoxon rank-sum test

分类号:

TP301.6

DOI:

10.11992/tis.202112043

摘要:

针对传统教与学算法存在易陷入局部最优、收敛速度慢和求解精度低等问题，提出一种融合经验反思机制的教与学优化算法(empirical reflection teaching learning based optimization, ERTLBO)。首先在教学阶段引入经验反思机制，遴选精英个体引导普通个体向教师靠近，提高班级整体水平，从而提高算法全局探索能力。其次在学习阶段引入动态自适应权重，能够根据学生的适应度值对位置进行自适应扰动，进而实现个体位置的动态更新，提高算法跳出局部最优的能力。仿真实验选取23个基准测试函数对ERTLBO同其他变体和流行算法进行性能测试。实验结果表明，ERTLBO算法具有更好的寻优性能和求解稳定性。最后，通过2个工程设计问题进一步验证ERTLBO解决实际问题的有效性和优越性。

Abstract:

In this paper, an empirical reflection teaching learning based optimization (ERTLBO) algorithm is proposed to solve the problems of easy falling into local optimum, slow convergence speed and low accuracy in traditional teaching-learning-based optimization. Firstly, the empirical reflection mechanism is introduced in the teaching stage to select elite individuals, guide ordinary individuals to approach teachers, and improve the overall level of the class, so as to improve the overall exploration ability of the algorithm. Secondly, the dynamic adaptive weight is introduced in the learning stage, which can adaptively disturb the position according to the fitness value of students, so as to realize dynamic updating of individual position and improve the ability of the algorithm to jump out of local optimum. In the experiment, 23 benchmark functions are selected to test the performance of ERTLBO, variant, and popular algorithms. Experimental results show that ERTLBO has better optimization performance and solution stability. Finally, the effectiveness and superiority of ERTLBO in solving practical problems are further verified through two engineering design problems.