摘要: |
利用人工神经网络理论预测带有表面缺陷的连续油管的低周疲劳寿命。基于自组织特征映射神经网络(SOFM)与径向基函数神经网络(RBF),考虑连续油管表面缺陷的影响,建立连续油管寿命预测的混合网络模型。该模型利用SOFM神经网络的自组织聚类能力对样本进行分类,并将其分类中心及对应的权值向量传递给RBF神经网络,作为RBF神经网络径向基函数的中心,再利用RBF神经网络非线性逼近能力预测连续油管寿命。结果表明,SOFM和RBF混合神经网络的预测结果在精度与稳定性上优于BP神经网络。 |
关键词: 连续油管 疲劳寿命 表面缺陷 人工神经网络 |
DOI:10.3969/j.issn.1673-5005.2018.03.016 |
分类号:TE931.2 |
文献标识码:A |
基金项目:国家科技重大专项(2015ZX05072004) |
|
Fatigue life prediction of coiled tubings based on artificial neural network |
YU Guijie, ZHAO Chong, CHI Jianwei, ZHANG Jiaxing
|
(College of Pipeline and Civil Engineering in China University of Petroleum(East China), Qingdao 266580, China)
|
Abstract: |
In the study, we use the artificial neural network theory to predict the low cycle fatigue life of coiled tubings with surface defects. Based on the self-organizing feature map (SOFM) and radial basis function (RBF) neural networks, and considering the effect of coiled tubing surface imperfections, a hybrid network model is built for predicting the life of coiled tubings. Using the self-organizing clustering ability of the SOFM neural network, in the model we classify the sample data, and the classification centers and corresponding weight vectors are transmitted to the RBF neural network, as the centers of RBF activation function, and then we can predict the working lives of coiled tubings by the nonlinear approximation ability of RBF neural network. The result shows that the hybrid network model is superior to BP neural network in accuracy and stability. |
Key words: coiled tubing fatigue life surface defect artificial neural network |