随机交叉-自学策略改进的教与学优化算法

PDF下载HTML

Title:

Teaching-learning-based optimization algorithm based on random crossover-self-learning strategy

作者:

黎延海¹, 雍龙泉¹, 拓守恒³

1. 陕西理工大学 数学与计算机科学学院，陕西 汉中 723001;
2. 陕西省工业自动化重点实验室，陕西 汉中 723001;
3. 西安邮电大学 计算机学院，陕西 西安 710121

Author(s):

LI Yanhai¹, YONG Longquan¹, TUO Shouheng³

1. School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong 723001, China;
2. Shaanxi Key Laboratory of Industrial Automation, Hanzhong 723001, China;
3. School of Computer Science and Technology, Xi’an University of Posts and Telecommunications, Xi’an 710121, China

关键词:

群体智能; 教与学优化; 随机交叉; “自学”策略; Benchmark函数; 非原点最优; 多样性分析

Keywords:

swarm Intelligence; teaching-learning-based optimization; random crossover; “self-learning” strategy; Benchmark function; non-origin-optimal; diversity analysis

分类号:

TP391

DOI:

10.11992/tis.201910045

摘要:

针对非原点最优的复杂优化问题(最优解不在坐标原点)，提出了一种基于随机交叉-自学策略的教与学优化算法(teaching and learning optimization algorithm based on random crossover-self-study strategy, CSTLBO)。对标准教与学优化算法的“教阶段”和“学阶段”的空间扰动进行了几何解释，改进了原有的“教阶段”和“学阶段”，并引入随机交叉策略和“自学”策略来提高算法的全局寻优能力。通过使用20个Benchmark函数进行仿真，并与6种改进的教与学优化算法进行结果比较及Wilcoxon秩和检验分析，结果表明CSTLBO算法能有效避免陷入局部最优，具有良好的全局搜索能力，求解精度高，稳定性好。

Abstract:

This paper presents a teaching-learning-based optimization algorithm based on a random crossover–self-learning strategy (CSTLBO) for solving non-origin-optimal complex optimization problems in circumstances in which the optimum solution is outside the origin of the coordinates. First, the spatial perturbation in the “teaching” and “learning” stages of the standard TLBO algorithm is interpreted geometrically, and the original “teaching” and “learning” stages are improved. Then, random crossover and “self-learning” strategies are introduced to improve the global optimization ability of the algorithm. To assess the performance of the CSTLBO algorithm, 20 Benchmark functions were tested, and Wilcoxon rank sum tests were applied to the experimental results of six improved TLBO algorithms. The experimental results indicate that the CSTLBO algorithm is able to avoid premature convergence, and has the advantages in global exploration, solution quality, and stability.