Application of Machine Learning for Predicting the Trapping Efficiency of Vortex Tube Silt Ejectors

A vortex tube silt ejector is employed to extract sediments from the canal. It consists of a duct laid across the whole bed of the canal with a slit along its top edge, and compared to the other alternative sediment-extraction devices, it is very efficient and economical. A Vortex Tube silt ejector is a device that is used to remove unwanted sediment from the irrigation and power canal. The vortex tube basically removes the sediment particle with the rotational action of the entered flow and receives it to the escaped channel. In this study, M5P, M5Rules, Random Forest (RF), and Gradient Boosting Method (GBM) approaches were employed to predict the trapping efficiency of the vortex tube ejector. Data was obtained by conducting experiments on the vortex tube silt ejector. The input data set consists of sediment size (mm), the concentration of sediment (ppm), the ratio of slit thickness and diameter of the tube (t/d), and extraction ratio (%), whereas trapping efficiency (%) was considered as output. The comparative analyses with conventional models reveal that the GBM outperforms the other ML models, achieving a Coefficient of Correlation (CC) of 0.9985, Root Mean Square Error (RMSE) of 0.769, and a Mean Absolute Error (MAE) of 0.531, indicating superior accuracy with lessor errors.