Premium
A predictive model for pressure fluctuations on sloping channels using support vector machine
Author(s) -
Guven Aytac
Publication year - 2011
Publication title -
international journal for numerical methods in fluids
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.938
H-Index - 112
eISSN - 1097-0363
pISSN - 0271-2091
DOI - 10.1002/fld.2318
Subject(s) - support vector machine , mean squared error , mathematics , correlation coefficient , range (aeronautics) , nonlinear system , jump , linear regression , regression analysis , root mean square , statistics , algorithm , computer science , artificial intelligence , engineering , physics , electrical engineering , quantum mechanics , aerospace engineering
Abstract In this study, support vector machine (SVM) is proposed as a new predictive model for pressure fluctuations beneath free jump occurring on sloping channels. The proposed model reproduces the pressure fluctuation intensity C ′ p in terms of normalized flow and channel section characteristics. The model variables were derived from dimensional analysis. The proposed model is calibrated and validated by using a wide range of experimental data. The SVM predicted C ′ p with a correlation of coefficient (CC) of 0.989 and a root mean square error (RMSE) of 0.004. Also, linear and nonlinear regression analyses are applied on the same experimental data set, and the SVM model is compared to the equations obtained from these regression analyses. CC, RMSE and average absolute deviation ( D i ) are used in the evaluation of performance of each model. The SVM model predicted the measured pressure fluctuations better than conventional regression equations. The results of this study reveal that the proposed SVM model can be effectively used in predicting the pressure fluctuation beneath free jump. Copyright © 2010 John Wiley & Sons, Ltd.