Prediction of local scour depth around bridge pier under live bed and
Prediction of local scour depth around bridge pier under live bed and
amir sheikhzadeh1 gholam abbas barani2
IOI Code
:
XYBG-AZAGF
Publication :
International Conference of Civil, Architecture and Urban Planning Elites(cau2016.com)
Abstract :
Abstract
In this study, a numerical self-organized model named Group Method of Data handling (GMDH) based artificial intelligence is applied to estimatethe scour depth around bridge piers for live-bed and clear water conditions. Bridge piers used in this research include roud, sharp, cylindrical, and square piers. The GMDH network based on a feed forward model is improved by Back Propagation (BP) algorithm. Hence, for learning process of the GMDH algorithm, quadratic polynomial including six weighting coefficients is applied in structure of the GMDH network. Effective parameters on the maximum scour depth are classified into median diameter of bed sediments, length of pier, diameter of pier, average velocity upstream of pier, and approaching flow depth. The use of Buckingham theorem, five non-dimensional parameters as input variables were choosen to predict the scour depth. Through the training and testing stages, performances of proposed GMDH models would be evaluated. Result of GMDH networks are compared with those obtained using other empirical equations based regression technique
Keywords : local scour bridge pier clear water condition live bed condition group method of data handling back propagation algorithm