Abstract: To overcome these weakness of premature, low solution precision and slow convergence speed in solving higher dimension functions, an improved Teaching Learning Based Optimization（TLBO） algorithm with double populations competition is proposed. Two teachers with the best achievement individual are chosen from population and other individuals are divided into two student populations by imperialist competitive. Each teacher guides itself student population. In the iteration, teacher can attract other individual in another population to become his own population member. Opposition-based learning is introduced to improve the learning ability of teacher. Comparison with related algorithms is given on some classical benchmark functions. The results show that the proposed algorithm has better convergence rate and accuracy for numerical optimization, suitable to solve the high dimensional optimization problem.

Key words: teaching learning based optimization, premature, double populations, competition, opposition-based learning

摘要： 为了克服教与学优化算法在求解高维函数问题时，容易早熟，收敛速度慢，解精度低的弱点，提出一种引入竞争机制的双种群教与学优化算法。在该算法中设置两个教师，并基于帝国竞争优化机制将种群初始化成为两个学生种群，每一个教师带领自己的种群独立进化。在进化过程中，教师可以利用自己的影响力将外种群内的成员吸收进入自己的种群。为了提高教师个体的学习能力，引入反向学习机制。在多个Benchmark函数的测试表明，改进算法解精度较高，全局收敛能力强，适合求解较高维度的函数优化问题。

关键词: 教与学优化, 早熟, 双种群, 竞争, 反向学习