摘要
为提高大坝变形预测精度,针对大坝变形监测序列的非线性、非平稳性等特点,提出一种基于具有自适应噪声的完整集成经验模态分解(CEEMDAN)-相空间重构(PSR)-核极限学习机(KELM)的大坝变形预测模型。首先利用CEEMDAN算法将大坝变形监测序列分解成为若干不同频率的子序列,然后对各序列进行相空间重构,依据重构的各个子序列分别建立相应的KELM预测模型,最后对各子序列预测结果进行叠加求和得到最终预测结果。通过实例对比分析表明,该模型在大坝变形预测中预测精度较高,对于大坝变形安全监测具有一定的实用价值。
In order to improve the prediction accuracy of dam deformation and to overcome the deficiencies in monitoring and predicting non-stationary dam deformation,based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)-phase space reconstruction(PSR)and an extreme learning machine with kernels(KELM),a hybrid model of dam deformation prediction was presented.Firstly,the dam deformation monitoring sequence was decomposed into several sub-sequences with different frequencies by CEEMDAN algorithm,then the phase space reconstruction of each sequence was carried out and the corresponding KELM prediction model was established according to each sub-sequence of reconstruction.Finally,the final prediction result was obtained by superimposing the prediction results of each sub-series.The comparison and analysis of examples show that the model has a high prediction precision and has a certain practical value for dam deformation safety monitoring.
作者
周兰庭
徐长华
袁志美
卢韬
ZHOU Lanting;XU Zhanghua;YUAN Zhimei;LU Tao(College of Water Conservancy and Hydropower Engineering,Hohai University,Nanjing 210098,China)
出处
《人民黄河》
CAS
北大核心
2019年第6期138-141,145,共5页
Yellow River
基金
国家自然科学基金资助项目(51209078)
关键词
大坝变形预测
集成经验模态分解
相空间重构
核极限学习机
dam deformation prediction
ensemble empirical model decomposition
phase space reconstruction
extreme learning machine with kernels