基于自主学习行为的教与学优化算法

Improved teaching-learning-based optimization algorithm based on self-learning mechanism

针对教与学优化(TLBO)算法收敛精度较低、易于早熟收敛等问题,提出一种基于自主学习行为的教与学优化算法(SLTLBO)。SLTLBO算法为学生构建了更加完善的学习框架,学生在完成常规"教"阶段与"学"阶段的学习行为之外,将进一步对比自己与教师、最差学生的差异,自主完成多样化的学习操作,以提高自己的知识水平,提高算法的收敛精度;同时学生通过高斯搜索的自主学习反思行为跳出局部区域,实现更好的全局搜索。利用10个基准测试函数对SLTLBO算法进行了性能测试,并将SLTLBO算法与粒子群优化(PSO)算法、智能蜂群(ABC)算法以及TLBO算法进行结果比对,实验结果验证了SLTLBO算法的有效性。

Aiming at the problems of low convergence precision and premature convergence in Teaching-Learning-Based Optimization( TLBO) algorithms, an improved Self-Learning mechanism-based TLBO( SLTLBO) algorithm was proposed. A more complete learning framework was constructed for students in SLTLBO algorithm. Besides, after completing nomal learning in "teaching" and "learning" stage, students would further compare their differences from the teachers and the worst students,then various learning operations were implemented independently, so as to enhance their knowledge level and improve the convergence accuracy of the algorithm. Meanwhile, the students carried out self-examination through Gaussian searching to jump out of the local area and achieved better global search. The performance of SLTLBO was tested on 10 benchmark functions and compared with the algorithms including Particle Swarm Optimization( PSO), Artificial Bee Colony( ABC) and TLBO. The experimental results verify the effectiveness of the proposed SLTLBO algorithm.