Comparative Study of Coupling Models of Feature Selection Methods and Machine Learning Techniques for Predicting Monthly Reservoir Inflow

Effective reservoir operation under the effects of climate change is immensely challenging. The accuracy inflow forecasting one essential factors supporting operations. This study aimed to investigate coupling models feature selection (FS) and machine learning (ML) algorithms predict monthly inflow. was carried out using data from Huai Nam Sai in southern Thailand. Eighteen years recorded (i.e., inflow, storage, rainfall, regional indices) with up a 12-month time lag were utilized. Three ML techniques, i.e., multiple linear regression (MLR), support vector (SVR), artificial neural network (ANN)were compared their capabilities. In addition, two FS genetic algorithm (GA) backward elimination (BE) methods, studied four predictable intervals, consisting 3, 6, 9, 12 months advance. Ten-fold cross-validation used for model evaluation. Study results revealed that methods GA BE) Could improve performance SVR ANN predicting forecasting, but they have no on MLR. Different developed suitable different time-step-ahead. BE-ANN provided best three-time-ahead (T + 3) nine-time-ahead 9) by giving an OI 0.9885 0.8818, NSE 0.9546 0.9815, RMSE 1.3155 1.2172 MCM/month, MAE 0.9568 0.9644 r 0.9796 0.9804, respectively. GA-ANN showed highest prediction six-time-ahead 6), 0.8997, 0.9407, 2.1699 1.7549 0.9759. twelve-time-ahead 12), 0.9515, 0.9835, 1.1613 0.9273 0.9835.