Application of the Nearest Neighbor Algorithm to Predict Cavitation Damage on Spillways

In this paper, the Nearest Neighbor Algorithm has been applied to predict cavitation damage on dam spillways. In this research, based on flow velocity and cavitation index, five different damage levels from ‘no cavitation damage’ to ‘major cavitation damage’ have been determined. The hydraulic characteristics of flow over the Shahid Abbaspour dam spillway were calculated for different flow rates. Then, the Nearest Neighbor Algorithm has been applied to predict cavitation damage levels and locations on this dam spillway for different flow rates. Comparison of the model results with the observed damages occurred during past floods on this spillway structure, shows that this algorithm predict damage levels and locations appropriately. Finally, the efficiency and precision of the model results have been evaluated by some statistical coefficients. Appropriate values of the correlation coefficient (r=0.896), the Mean Absolute Error (MAE=0.101), the Root Mean Square Error (RMSE=0.108) and Efficiency of model (EF=0.813) confirm that the present model can be suitable and efficient.