Reservoir Inflow Prediction Using Hybrid Data Assimilation and Support Vector Machine Model

Accurate and reliable reservoir inflow forecast is crucial for real-time reservoir operation and management. Developing a simulation model to forecast future states of a system generally follows with errors in prediction. Frequently, data-based models such as Support Vector Machines (SVM) are used as simulation and forecasting techniques in hydrologicalmodeling. This paper introduces a hybrid model which couples the machine learning technique of Support VectorRegression for prediction (SVR) and Ensemble Kalman Filter (EnKF) as data assimilation procedure in order to improvethe performance and accuracy of inflow to reservoir. Inflow quantities to Zayandehroud reservoir is considered as state vector in assimilation process to develop better estimation for inflow. Evaluation criteria such as Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) are implemented to evaluate the performance of SVR-EnKF model. As a result, the adjusted values of SVR-EnKF compared to SVR model indicate improved performance of proposed model

کلیدواژه ها:

Reservoir Inflow Prediction ، Support Vector Regression ، Ensemble Kalman Filter ، Data Assimilation