摘要
深基坑监测会对周围邻近建筑物造成影响、进而导致其沉降。而对深基坑相邻建筑物实施沉降监测有助于控制基坑开挖,能够及时发现沉降危害。为提高深基坑相邻建筑物沉降预测的精度,提出了一种小波神经网络模型并以其前4期的监测数据预测后1期的累积沉降量。通过5个沉降监测点、近2 a时间的监测数据,对比了自回归模型和小波神经网络模型的预测精度,结果表明:小波神经网络模型的短期预测精度优于自回归模型,其长期预测精度与自回归模型相当。小波神经网络模型的稳定性好,且其预测精度不会随着时间的推移而衰减。对于变形特征较复杂的监测点仍能够取得较高的精度,能够为各类复杂的基坑变形预测提供有效且可靠的指导。
Deep foundation pit monitoring can cause an impact on nearby buildings,leading to settlement.The settlement monitoring of adjacent buildings helps to control the excavation of foundation pits and timely detect settlement hazards.To improve the accuracy of settlement prediction for adjacent buildings in deep foundation pits,this paper proposes a wavelet neural network model,which predicts the cumulative settlement of the following phase based on monitoring data from the previous four phases.The prediction accuracy of autoregressive model and wavelet neural network model was compared based on monitoring data from 5 settlement monitoring points over the past 2 years.The results show that the short-term prediction accuracy of the wavelet neural network model is superior to that of the autoregressive model,and its long-term prediction accuracy is comparable to that of the autoregressive model.The wavelet neural network model has good stability and prediction accuracy does not decay over time.It can still achieve high accuracy on monitoring points with complex deformation characteristics,providing effective and reliable guidance for various complex foundation pit deformation predictions.
作者
任定春
黄纪远
REN Dingchun;HUANG Jiyuan(Sinohydro Bureau 7 Co.,Ltd.,Chengdu Sichuan 610213;College of Geomatics Science and Technology,Nanjing University of Technology,Nanjing Jiangsu 211816)
出处
《四川水力发电》
2023年第S02期35-39,共5页
Sichuan Hydropower
关键词
小波基函数
神经网络
自回归模型
沉降监测
基坑沉降
预测方法
Wavelet basis function
Neural network
Autoregressive model
Settlement monitoring
Foundation Pit Settlement
Prediction Method
