基于深度学习的大坝变形预测模型

Dam Deformation Prediction Model based on Deep Learning

鉴于高性能的混凝土坝变形动态预测模型是预测结构性态演化、评价安全服役状况和保障稳定高效运行的关键措施。以混凝土坝原型变形监测数据为基础,借助开源深度学习框架TensorFlow建立了基于深度学习的混凝土坝变形预测模型。工程实例应用结果表明,基于深度学习的混凝土坝变形预测模型各项评价指标均优于现浅层神经网络模型和传统的统计模型,实现了动态高精度预测混凝土坝运行性态,具有很强的工程实用性。

High performance dynamic prediction model of concrete dam deformation is an effective measure to simulate and predict the operation behavior,comprehensively evaluate the health status in service and guarantee the safe and effective operation of concrete dam.Based on the monitoring data of concrete dam prototype,this paper established a deep learning-based concrete dam deformation safety prediction model with the help of Tensorflow,an open source deep learning framework.The application of engineering example shows that the evaluation indexes of the concrete dam deformation prediction model based deep learning are superior to the existing shallow neural network model and statistical model.Therefore,the deep learning based concrete dam deformation prediction model realizes the high performance and dynamic prediction of concrete dam operation behavior,which has strong engineering practicability.