Abstract: The teaching and learning based optimization in higher optimization problems has some weakness, such as premature maturity, and poor solution accuracy. To overcome these shortcomings, an improved TLBO with adaptive competitive learning (ITLBOAC) was proposed. Firstly, a nolinear dynamic variation coefficient is introduced into the teaching process to adjust the state to individual learning in the iterative optimization process. As a result, the current individual can learn more from the teacher in the early stage to improve its state quickly, and keep self-status more in the later stage to slow down the influence of the teacher. Secondly, an adaptive competition "learning" operator between nearest neighbor individuals based on ecological co-competition mechanisms is introduced. Choose its nearby neighbors as they eventually shift from cooperative evolution to competitive learning. The experiments are conducted on 12 classic testing functions, and the experimental results compared with other representative variants show that the proposed algorithm is much better than compared TLBO at not only the accuracy of solutions but also for the convergence speed and robustness, fit for solving multimode and high dimension function optimization problems. Select compression springs and three-rod truss design problem to test, the optimal values obtained by ITLBOAC decreased by about 3.03% and 0.34%, respectively, compared with TLBO. ITLBOAC is a trustworthy algorithm in solving constraint engineering problems.

摘要： 针对求解较高维度问题时，教与学优化（TLBO）算法容易早熟、解精度降低等问题，提出一种自适应竞争学习教与学优化（ITLBOAC）算法。首先，在“教”算子中引入非线性变化的权重参数，分别用于决定当前个体自身状态的保持能力，以及调整当前个体向教师学习的态度，使当前个体在早期更多向教师进行学习，迅速提升自身状态，而后期更多保持自身状态，减缓教师对它的影响。其次，以生态学协同竞争机制为基础，引入基于近邻个体间的自适应竞争“学”算子。当前个体选择它的近邻个体由协作演化逐渐过渡到竞争性学习。在12个Benchmark函数上的测试结果表明，ITLBOAC具有更好的解精度、稳定性和收敛速度，较TLBO算法有较大幅度提升，适合于求解较高维度的连续型优化问题。选择压缩弹簧和三杆桁架设计问题测试，ITLBOAC获得的最优值分别比TLBO下降了约3.03%和0.34%。在求解约束工程问题时，ITLBOAC同样值得信任。