THE NUMERICAL SIMULATION OF LOW STRAIN DYNAMIC RESPONSE FOR FOUNDATION PILES AND THE APPLICATION FOR DEFECT DIAGNOSIS BY ARTIFICIAL NEURAL NETWORKS
The artificial neural network (ANN) is one of the popular approach for foundation pile diagnosis. However,a large number of lerning samples is necessary. There are masses of records by pile integrity tester (PIT) for actual engineering piles,but it is difficult to determine their exact defects by the straightforward interpretation. Even though the designed model piles embedded into the soil can provide their exact integrity parameters,they are very expensive. Thus,the threedimension finite element method (FEM) of structural dynamic mechanics is employed to simulate the low strain dynamic response for more than 200 piles,including several defect types of model piles that are free of soil and engineering piles that are surrounded with soil. The soil damping is performed by the structural damping coefficients of local elements. And the resulting numerical database is used to produce the lerning samples for ANN to diagnose the pile quality. The satisfactory prediction of the integrity for actul model pile and engineering pile indicates that the approach is effective.
Numerical simulation Finite element method Foundation pile Dynamic integrity testing Artificial neural network Quality diagnosis
Can-Hui Zhang Shao-Qiang Yang Jian-Guo Zhang Jian-Hua Xie Jin-Lin Zhang
Department of Civil Engineering,Xiamen University,Xiamen 361005,P.R. China The Construction Safety and Quality Supervision Station of Xiamen,Xiamen 361000,P.R.China
国际会议
The Eleventh International Symposium on Structural Engineering(第十一届结构工程国际研讨会)
广州
英文
960-964
2010-12-01（万方平台首次上网日期，不代表论文的发表时间）