Derivation of a correlation for drag coefficient in two-dimensional bounded supercavitating flows, using artificial neural networks

Abstract

Artificial neural networks (ANNs) are used as a new approach for the determination of the relations between drag coefficient and Cavitation Number with cavity geometry in supercavitating flows which have been most widely used in the hydrodynamics researches. Also the result of the ANNs as a cost function potentially will be used in an optimization algorithm. Instead of complex differential equations and limited experimental data, faster and simpler solutions were obtained using equations derived from the ANN model. For training of the ANN the numerical results are used that are obtained from a boundary element method (BEM). At this problem, a two-dimensional supercavitation potential inviscid flow pasts a symmetric two-dimensional cavitator, which is placed perpendicular to the flow in a channel of infinite width and immediately a cavity is formed behind the cavitator. It was found that the coefficient of multiple determination (R2-value) between the actual and ANN predicted data is equal to about 0.9998 for the drag coefficient and Cavitation number. As seen from the obtained results, the calculated cavity geometry for all drag coefficients and Cavitation Numbers are obviously within acceptable limits.