基于修正的BP神经网络算法对桩基承载力预测方法研究
|
摘要
在岩土工程中如何准确预测桩基竖向承载力是一件非常重要的事情。针对现有研究存在的不足,基于标准BP神经网络算法,加入一动量因子,建立了修正的BP神经网络模型,对单桩的竖向承载力进行了预测。以镇江市勘察测绘研究院所完成的地质勘查报告为工程背景,以地震波静力触探测试(SCPTU)测得的4个指标(锥尖阻力、锥侧摩阻力、剪切波速和孔隙水压力)为输入参数,桩基承载力为输出参数。通过与现场静载试验进行比对,得到了相关系数较高的桩基荷载响应曲线。经过与传统预测方法进行比较发现,用修正的BP神经网络算法可以有效预测桩基竖向承载力,精度较高。
How to exactly predict the pile bearing capacity is of great significance in geotechnical engineering. Aiming at the shortage of the present research,a modified BP neural network predicting model based on normal artificial neural network algorithm is built to predict the pile bearing capacity. Putting four indicators( cone resistance,cone lateral friction resistance,shear wave velocity and pore water pressure) obtained from SCPTU as inputs and the vertical bearing capacity as output. Compared with the static test,the load response curve of pile with high correlation is presented. Compared with traditional predicting methods,it is obvious that the modified BP neural network model can validly predict pile bearing capacity.
How to exactly predict the pile bearing capacity is of great significance in geotechnical engineering. Aiming at the shortage of the present research,a modified BP neural network predicting model based on normal artificial neural network algorithm is built to predict the pile bearing capacity. Putting four indicators( cone resistance,cone lateral friction resistance,shear wave velocity and pore water pressure) obtained from SCPTU as inputs and the vertical bearing capacity as output. Compared with the static test,the load response curve of pile with high correlation is presented. Compared with traditional predicting methods,it is obvious that the modified BP neural network model can validly predict pile bearing capacity.
引文
[1]Paul W.Mayne.Evaluating Axial Drilled Shaft Response By Seismic Cone[J].Foundations&Ground Improvement,GSP113,ASCE,Reston/VA,pp.655~669.
[2]De Ruiter J,Beringen F L.Pile foundations for large North Sea structures[J].Mar Geotechnique,1979,39(3):267~314.
[3]Bustamante M,Gianeselli L.Pile bearing capacity prediction by means of staticpenetrometer CPT[C].Proceedings of the second European symposium onpenetration testing,Amsterdam:Balkema,1982:493~500.
[4]Meyerhof G G.Bearing capacity and settlement of pile foundations[J].Journal of Geotechnical Engineering,ASCE,1976,102(3):197~228.
[5]Tumay M T,Boggess R L,Acar Y.Subsurface investigation with piezocone penetrometer[C].Proc Cone Penetration Testing and Experience,New York:ASCE,1982:325~342.
[6]Schmertmann J H.Guidelines for cone penetration test[M].Washington,D.C:performance and design,1978.
[7]Gwizdaa K.Polish design methods for single axially loaded piles[C].Design of Axially Loaded Piles,European Practice,Brussels:1997:291~306.
[8]Goh,A.T.C.Nonlinear modelling in geotechnical engineering using neural networks[J].Australian Civil Engineering Transactions,1994,36(4):293~297.
[9]Goh,A.T.C.Empirical design in geotechnics using neural networks[J].Geotechnique,1995,45(4):709~714.
[10]Goh,A.T.C.Back-propagation neural networks for modeling complex systems[J].Artificial Intelligence in Engineering,1995,39(3):143~151.
[11]Goh,A.T.C.Pile driving records reanalyzed using neural networks[J].Geotech.Engrg.,ASCE.1996,122(6):492~495.
[12]Abu-Kiefa,M.A.General regression neural networks for driven piles in cohesionless soils[J].Geotech.&Geoenv.Engrg,ASCE,1998,124(12):1177~1185.
[13]Shahin,M.A.,Jaksa,M.B,and Maier,H.R.Predicting the settlement of shallow foundations on cohesionless soils using back-propagation neural networks[R].Research Report No.R 167,2000.
[14]F.Pooya Nejad a.Prediction of pile settlement using artificial neural networks based on standard penetration test data[J].Computers and Geotechnics,2009,36(9):1125~1133.
[15]蒋建平,阎长虹,高广运.地基土压缩系数的RBF神经网络预测[J].江苏大学学报·自然科学版,2011,32(2):232~235.
[16]张丽萍.基于小波概率神经网络的单桩竖向承载力预测模型及应用研究[J].工业建筑,2012,42(9):107~110.
[17]Mohamed A.Load-settlement modeling of axially loaded steel driven piles using CPT-based recurrent neural networks[J].Soils and Foundations,2012,117(4):1969~2015.
[2]De Ruiter J,Beringen F L.Pile foundations for large North Sea structures[J].Mar Geotechnique,1979,39(3):267~314.
[3]Bustamante M,Gianeselli L.Pile bearing capacity prediction by means of staticpenetrometer CPT[C].Proceedings of the second European symposium onpenetration testing,Amsterdam:Balkema,1982:493~500.
[4]Meyerhof G G.Bearing capacity and settlement of pile foundations[J].Journal of Geotechnical Engineering,ASCE,1976,102(3):197~228.
[5]Tumay M T,Boggess R L,Acar Y.Subsurface investigation with piezocone penetrometer[C].Proc Cone Penetration Testing and Experience,New York:ASCE,1982:325~342.
[6]Schmertmann J H.Guidelines for cone penetration test[M].Washington,D.C:performance and design,1978.
[7]Gwizdaa K.Polish design methods for single axially loaded piles[C].Design of Axially Loaded Piles,European Practice,Brussels:1997:291~306.
[8]Goh,A.T.C.Nonlinear modelling in geotechnical engineering using neural networks[J].Australian Civil Engineering Transactions,1994,36(4):293~297.
[9]Goh,A.T.C.Empirical design in geotechnics using neural networks[J].Geotechnique,1995,45(4):709~714.
[10]Goh,A.T.C.Back-propagation neural networks for modeling complex systems[J].Artificial Intelligence in Engineering,1995,39(3):143~151.
[11]Goh,A.T.C.Pile driving records reanalyzed using neural networks[J].Geotech.Engrg.,ASCE.1996,122(6):492~495.
[12]Abu-Kiefa,M.A.General regression neural networks for driven piles in cohesionless soils[J].Geotech.&Geoenv.Engrg,ASCE,1998,124(12):1177~1185.
[13]Shahin,M.A.,Jaksa,M.B,and Maier,H.R.Predicting the settlement of shallow foundations on cohesionless soils using back-propagation neural networks[R].Research Report No.R 167,2000.
[14]F.Pooya Nejad a.Prediction of pile settlement using artificial neural networks based on standard penetration test data[J].Computers and Geotechnics,2009,36(9):1125~1133.
[15]蒋建平,阎长虹,高广运.地基土压缩系数的RBF神经网络预测[J].江苏大学学报·自然科学版,2011,32(2):232~235.
[16]张丽萍.基于小波概率神经网络的单桩竖向承载力预测模型及应用研究[J].工业建筑,2012,42(9):107~110.
[17]Mohamed A.Load-settlement modeling of axially loaded steel driven piles using CPT-based recurrent neural networks[J].Soils and Foundations,2012,117(4):1969~2015.