基于EEMD-AEFA-LSTM的混凝土坝变形预测模型

Deformation Prediction Model of Concrete Dam Based on EEMD-AEFA-LSTM

对混凝土坝变形监测资料进行合理的数据挖掘和准确的预测分析是确保大坝安全长效运行的关键手段,大坝变形时间序列受到温度、水位等环境量的影响,具有周期性、非线性等变化特征,现有的智能算法无法很好地捕捉序列的非线性关系。对此,提出了基于EEMD-AEFA-LSTM模型的混凝土坝变形预测模型,采用集合经验模态分解EEMD有效分解变形时间序列,通过人工电场算法AEFA优化的长短期记忆网络LSTM模型对各分解分量进行预测并重构预测结果。选取某混凝土坝EX16、EX24测点的变形监测资料开展预测研究。结果表明,所建EEMD-AEFA-LSTM模型的预测精度明显高于AEFA-LSTM、PSO-LSTM、GA-LSTM模型,预测结果的平均绝对误差、均方误差、均方根误差均为最小值,为混凝土坝变形的精确预测提供了新途径。

Reasonable data mining and accurate prediction and analysis of concrete dam deformation monitoring data are the key means to ensure the safe and long-term operation of the dam.Due to the impact of environmental variables such as temperature and water level,the dam deformation time series has periodic,nonlinear and other change characteristics.Existing intelligent algorithms can not capture the nonlinear relationship of sequences well,A concrete dam deformation prediction model based on EEMD-AEFA-LSTM model was proposed.Ensemble empirical mode decomposition was used to effectively decompose the deformation time series.The long short-term memory network model optimized by the artificial electric field algorithm was used to predict the decomposition components and reconstruct the prediction results.The deformation monitoring data of EX16 and EX24 measuring points of a concrete dam were selected for prediction research.The results show that the prediction accuracy of the EEMD-AEFA-LSTM model is significantly higher than that of the AEFA-LSTM model,PSO-LSTM model,and GA-LSTM model.The average absolute error,mean square error,and root mean square error of the prediction results are the minimum values,providing a new way for accurate prediction of concrete dam deformation.