摘要
对中老昆万(昆明—万象)铁路景寨隧道大变形进行现场实测与分析,同时考虑时间和距离两个因素的影响,采用BP神经网络算法对隧道拱顶沉降数据进行拟合和预测,并与实测值、施工现场采用的指数函数拟合值进行对比。结果表明,隧道开挖初期隧道变形受距掌子面距离和时间双重影响,支护后以受距离影响为主,距掌子面距离达到75 m后,逐渐转变为以受时间影响为主;BP神经网络算法在量测过程中可不断进行曲线拟合并给出预测值,拟合值比指数函数拟合值更贴合实测值,快捷适用,且精度较高,有利于及时采取施工措施。采用BP神经网络算法拟合及预测时对初始输入数据的依赖性强,可增加原始数据,使预测模型更贴合实际。
The large deformation of Jingzhai tunnel of Kunming-Vientiane railway between China and Laos was measured and analyzed on site. Considering the influence of time and distance,BP neural network algorithm was used to fit and predict the tunnel vault settlement data,and compared with the measured value and the fitting value of exponential function used in the construction site. The results show that the tunnel deformation at the initial stage of tunnel excavation is affected by the distance from the tunnel face and time. After support,it is mainly affected by distance. After the distance from the tunnel face reaches 75 m,it gradually changes to be mainly affected by time. BP neural network algorithm can continuously fit the curve and give the predicted value in the measurement process. The fitting value fits the measured value,which is better than the fitting value of exponential function. BP neural network algorithm is fast,applicable and has high accuracy,which is conducive to taking construction measures in time. It is highly dependent on the initial input data when fitting and predicting,and increasing the original data can make the prediction model more practical.
作者
刘博峰
LIU Bofeng(Rail Transportation Branch,China Communications Construction Company Limited,Beijing 100088,China)
出处
《铁道建筑》
北大核心
2022年第7期106-109,共4页
Railway Engineering
关键词
铁路隧道
时空影响
监控量测
BP神经网络算法
拱顶沉降
变形预测
掌子面
railway tunnel
spatiotemporal influence
monitoring measurement
BP neural network algorithm
vault settlement
deformation prediction
tunnel face
