基于小波变换与差分变异BSO-BP算法的大坝变形预测
作者:
作者单位:

河海大学物联网工程学院

作者简介:

通讯作者:

中图分类号:

TP183

基金项目:

国家重点研发计划资助(2018YFC0407101), 国家基金青年项目(61403121), 中央高校基本科研业务费专项资金资助(2019B22314).


Dam deformation prediction based on wavelet transform and differential mutation BSO-BP algorithm
Author:
Affiliation:

College of IoT Engineering, Hohai University

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对现有大坝变形预测模型的预测精度不高; BP神经网络的参数和结构很难确定且容易陷入局部极值等问题, 本文引入小波变换理论把原始的大坝变形序列分解成几个子序列, 然后对每个子序列使用头脑风暴算法(Brainstorming algorithm, BSO)优化BP神经网络的参数和结构. 另外, 考虑到基本的BSO算法中使用高斯变异的生成策略, 其固定的传递函数不利于探索解的分布且计算复杂度较高, 因此, 本文把差分进化算法中的差分变异思想用到基本的BSO算法中, 建立了一种基于小波变换和差分变异头脑风暴算法优化BP神经网络的大坝变形预测模型. 实验证明, 与其他预测模型相比, 本文提出的预测模型具有更高的预测精度.

    Abstract:

    The prediction accuracy of the existing dam deformation prediction model is not high; the parameters and structure of the BP neural network are difficult to determine and it is easy to fall into the local extremum. This paper introduces the wavelet transform theory to decompose the original dam deformation sequence into several subsequences, then use the Brainstorming algorithm (BSO) to optimize the parameters and structure of the BP neural network for each subsequence. In addition, considering the Gaussian mutation in the basic BSO algorithm, its fixed transfer function is not conducive to exploring the distribution of solutions and its computational complexity is high. Therefore, this paper applies the differential mutation idea in differential evolution algorithm to the basic BSO algorithm, and establishes a prediction model based on wavelet transform and differential mutation brainstorming algorithm for optimizing BP neural network. Experiments show that compared with other prediction models, the prediction model proposed in this paper has higher prediction accuracy.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2019-10-12
  • 最后修改日期:2021-03-09
  • 录用日期:2020-03-18
  • 在线发布日期:
  • 出版日期: