RESERVOIR OPERATION USING MACHINE LEARNING ALGORITHM

Abstract

Generating hydropower, supporting water supply, and blocking over lasting droughts are few crucial tasks of water stored in reservoir. During floods, the water delivery from the reservoir must be acceptable, to confirm that the gross volume of water is at a safe level and release from the reservoir will not trigger flooding downstream. This study aims to develop the well-versed assessments for management of reservoir and pre release water outflow using machine learning, a new and exciting area of artificial intelligence and the most valuable, time, supervised, and cost-effective approach. In this study two data-driven forecasting models, Regression Tree (RT) and Support Vector Machine (SVM) are employed using approximately 30 years of hydrological records to simulate reservoir outflow. Obtaining accurate monthly river flow discharge prediction has always been a challenging task in water resources management for that different models of SVM and RT are applied to the data accurately to predict the fluctuations in the water outflow of a Bhakra reservoir. Different input combinations were used to find the most effective release such as reservoir level (M), monthly reservoir storage (BCM), the previous inflow of reservoir (MCM), the current inflow of reservoir (MCM), evaporation of reservoir (MCM), the previous outflow of the reservoir (MCM) and time (months) and release of the reservoir. . Findings indicate that SVM (medium guassian) combination having seven different parameters gives minimum RSME (720.2), maximum R2 (0.8), minimum MAPE( 14.0197), minimum scatter index(.4239) and minimum MAE (360.69) and therefore, can be considered as the best model for the dataset with these techniques. The ability to accurately estimate changes in reservoir outflow can aid in the planning and management of reservoir water usage in the long run.