【摘要】 变形监测是工程安全监测必不可少的工作，高精度的变形预测预报能够为安全监测提供变形预警。随着对人工智能模型的深入研究，利用神经网络模型对变形时序数据进行预报的研究也逐渐增多。为研究神经网络模型在变形预测中的效果，将结合CEEMDAN的LSTM模型应用在滑坡变形预报中，通过实验比较其与单一LSTM模型、传统的ARMA模型以及Kalman滤波模型在变形预报中的精度。实验结果表明，结合CEEMDAN的LSTM模型的预报精度明显优于其他模型，预测值与测量值的偏差在2 mm以内，平均相对误差为0.59%,模型效果较好。更多还原

【Abstract】 Deformation monitoring is an essential part to ensure the safety of the project.High precision deformation prediction can provide great help for safety monitoring.With the in-depth study of artificial intelligence model, the research using neural network model to forecast deformation time series data is gradually increasing.In order to study the effect of neural network model in deformation prediction, LSTM model combined with CEEMDAN is applied in landslide deformation prediction.The accuracy of LSTM model is compared with single LSTM model, traditional ARMA model and Kalman filter model in deformation prediction through experiments.The experimental results show that the prediction accuracy of LSTM model combined with CEEMDAN is obviously better than that of the other models.The deviation between the predicted value and the measured value is less than 2 mm, and the relative error is 0.59%,and the prediction effect is better.更多还原