基于LSTM-ARIMA模型的隧道围岩变形预测方法研究

Research on the Prediction Method of Tunnel Surrounding Rock Deformation Based on LSTM-ARIMA Model

为了在隧道施工过程中对隧道围岩变形进行精准预测,以中东某隧道为研究对象,基于隧道围岩监测数据使用LSTM(long short-term memory)和ARIMA(autoregressive integrated moving average)模型进行拟合并预测,进一步通过方差倒数法建立组合模型,采用多个统计学指标对建立的模型预测结果进行对比分析。结果表明,组合模型解决了ARIMA模型对非平稳数据预测精度较差的问题,充分发挥了LSTM和ARIMA模型各自的优势,能够更准确地捕捉数据特征和趋势,提高了对隧道围岩变形预测的准确性和鲁棒性,可为隧道工程的安全施工提供可靠的支持和指导。

In order to accurately predict the deformation of tunnel surrounding rock during tunnel construction,a tunnel in the Middle East is taken as the research object,and based on the monitoring data of tunnel surrounding rock,the LSTM(long short-term memory)and ARIMA(autoregressive integrated moving average)models are fitted to make a combined prediction,and the combined model is further established by the inverse of the variance method,and the results of all the models are compared and analyzed by using a number of statistical indexes.The results demonstrate that the combined model effectively addresses the issue of low prediction accuracy of ARIMA model for non-stationary data.It fully leverages the advantages of LSTM and ARIMA models to accurately capture data features and trends,thereby improving the accuracy and robustness of predicting tunnel rock deformation.This paper can provide reliable support and guidance for the safe construction of tunnel engineering projects.