BP神经网络在海洋平台桩基轴向承载力中的应用研究
Application Research of BP Neural Network in Offshore Platform Pile Foundation Axial Bearing Capacity
  
DOI:
中文关键词:  神经网络  静力触探    极限承载力
英文关键词:neural network  cone penetration test  pile  ultimate bearing capacity
基金项目:
作者单位
陈 磊 中海油田服务股份有限公司天津 300450 
摘要点击次数: 48
全文下载次数: 51
中文摘要:
      目前海洋石油导管架平台桩基础的轴向极限承载力常用的设计方法为API RP2A(美国石油协会)和静力触探(CPT)的方法,在这两种方法的基础上,提出了用BP神经网络模型对桩的轴向极限承载力进行计算的思路,能够有效的预测桩的轴向极限承载力。根据BP神经网络算法具有较强的非线性映射能力和学习功能的特点,通过对影响单桩极限承载力因素的分析,依据静力触探资料建立了基于BP神经网络的单桩轴向极限承载力预测模型。通过利用API RP2A 方法分析成果对该模型进行学习训练和预测检验,证明了预测模型性能良好、具有较高的精度和收敛速度快等特点,验证了神经网络方法的可行性,预测结果能够指导桩基础设计,缩短周期。因而具有较大的工程实用价值。
英文摘要:
      At present, the design methods of pile axial ultimate bearing capacity for offshore oil jacket platform commonly used include API RP2A (American petroleum institute) and Cone Penetration Test (CPT), based on these two methods, the idea of using BP neural network model to calculate the pile axial ultimate bearing capacity is proposed,which could effectively predict the pile axial ultimate bearing capacity.According to the characteristics of BP neural network algorithm with strong nonlinear mapping ability and learning function, through the analysis of factors affecting the single pile ultimate bearing capacity, the prediction model of axial ultimate bearing capacity based on BP neural network is established based on the CPT data. By using the analysis results of API RP2A method to study, train and test the prediction of the model, it is proved that the prediction model has good performance, high accuracy and fast convergence speed, and the feasibility of the neural network method is verified. The prediction results can guide the pile foundation design and shorten the period. Therefore, it has great practical value in engineering.
查看全文  查看/发表评论  下载PDF阅读器