基于递归神经网络的基坑工程变形多步预测方法研究

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名

邮箱

手机号码

标题

留言内容

验证码

基金项目:

广东省自然科学基金资助项目(编号:20010055);广东工业大学重点学科基金(编号:20216)

RESEARCH ON MULTI-STEP PREDICTION OF DEEP EXCAVATION DEFORMATION BASED ON RECURRENT NEURAL NETWORK

1.
Geotechnical Engineering Institute,Guangdong University of Technology Guangzhou 510640

摘要: 针对深基坑系统的复杂的非线性及基坑工程变形多步预测的重要性，将人工神经网络技术引入其中。分析了用BP网络进行多步预测时存在的不足，提出了基于递归神经网络的基坑工程变形多步预测模型。通过一软土深基坑工程变形多步预测实例的分析，论证了递归神经网络用于基坑工程变形多步预测的可靠性和实用性。该方法有效可行，在其他领域的多步预测中同样具有广阔的应用前景。
- 深基坑 /
- 递归神经网络 /
- 多步预测 /
- 变形
Abstract: An artificial neural network is introduced in the light of the complexity, nonlinearity of a deep excavation and the importance of multi-step prediction of its deformation.The defect of multi-step prediction by BP network is analyzed and a multi-step prediction model of an excavation deformation based on recurrent neural networks is also proposed.The reliability and practicability of the multi-step prediction of the excavation deformation by the recurrent neural networks are demonstrated through the multi-step prediction of the deformation of deep excavation in soft soil.It can be widely used for the muti-step prediction in other fields.
- deep excavation /
- recurrent neural networks /
- multi-step prediction /
- deformation