教学优化算法的改进及应用

Improvement and application of teaching and learning based optimization

针对教学优化算法(TLBO)在解决复杂实际问题时易陷入局部最优的缺陷,提出一种改进教学优化算法(MTLBO)。在教师阶段引进自适应基准消除"原点偏好",在学生阶段引进分科学习和学习阈值的学习策略保证学员多样性。测试结果表明,该改进提高了教学优化算法的全局搜索能力和求解精度。将改进教学优化算法应用于BP神经网络的权值和阈值优化中,建立基于改进教学优化算法的BP神经网络预测模型(MTLBO-BP)。选用4个真实数据集进行对比实验,实验结果表明,该模型具有更高的预测精度。

Aiming at the shortcomings of teaching and learning based optimization(TLBO)in solving complex practical applications,an improved teaching and learning based optimization(MTLBO)was proposed.In the teaching phase,an adaptive datum was introduced to eliminate origin preference.Divided subjects and learning thresholds were employed in the learning phase to ensure the diversity of students.Test results show that MTLBO can achieve higher solution accuracy and it has more powerful global search capability.MTLBO was adopted to optimize the weights and thresholds of BP neural network and the prediction model based on MTLBO(MTLBO-BP)was established.Results of comparison tests in four real datasets show that MTLBO-BP can achieve higher prediction accuracy.