基于多元结构应力特征的沉井基础下沉速度预测

Sinking speed prediction of an open caisson foundation based on the characteristics of multivariate structural stress data

沉井基础下沉速度预测对于确保沉井安全平稳下沉和预防潜在施工风险具有重要实际意义。基于沉井底部结构应力监测数据,分别采用二维卷积神经网络和三维卷积神经网络,构建沉井下沉快慢类别与沉井下沉速度值预测模型,通过提取结构应力监测数据的空间特征和时空特征,实现沉井下沉速度预测。依托常泰长江大桥主塔沉井基础下沉过程,分别验证2个预测模型的准确性与实用性,之后进一步完成取土下沉过程中沉井基础下沉快慢的实时预测,并分析预测步长和应力时空特征对下沉速度预测精度的影响。结果表明:提出的模型可以成功预测沉井下沉速度类别与大小,取得很好的预测效果;模型在实际工程中的可靠性和实用性较高;结构应力的时空特征对模型超前预测效果具有重要影响。研究成果实现了沉井下沉速度的实时预测分析,对沉井下沉智能化决策具有重要的参考价值。

The sinking speed prediction of open caisson foundations has important practical significance for ensuring the sinking safety and sinking steady,and to prevent potential construction risks.Based on the structural stress monitoring data at the bottom of open caisson foundations,a two-dimensional convolutional neural network(CNN)and a three-dimensional CNN are applied for proposing a sinking speed category prediction model and a sinking speed value prediction model.The spatial and spatial-temporal characteristics of the structural stress monitoring data are extracted to predict the sinking speed.The accuracy and practicability of the two prediction models were verified by applying them to an open caisson foundation of the main tower in the Changtai Yangtze River Bridge Project.Then,the real-time prediction of the sinking speed category in the sinking process was simulated,and the influence of the prediction step and the spatial-temporal characteristics of the structural stress on prediction accuracy were analyzed.The results show that the proposed models can successfully predict the category and value of the sinking speed.The reliability and practicability of the models were verified in practical engineering that the proposed sinking speed prediction models have good prediction performance.Moreover,the spatial-temporal characteristics of structural stress have an important influence on prediction accuracy.The real-time sinking speed prediction of open caisson foundations was achieved,which has important reference value for intelligent decision-making in the monitoring during sinking process.