Accurate and reliable simulation models are crucial for the operation and management of systems. Developing a simulation model to forecast future states of a system is generally followed by errors in prediction. Frequently, data-based models such as support vector machines (SVM) are used as forecasting techniques. This paper introduces a modular method which couples the machine learning technique of support vector regression (SVR) as a prediction model and a modified data assimilation (MDA) technique to partially correct the predicted values based on the observation data. To improve the performance and accuracy of the system output, the ensemble Kalman filter (EnKF) as a data assimilation procedure is implemented with an optimization procedure. As a case study, inflow quantities to Zayandehroud reservoir is considered as the state vector in the assimilation process to enhance the system output. Evaluation criteria such as root mean square error (RMSE) and R-squared criteria are implemented to evaluate the performance of the proposed model. The adjusted values of a hybrid model compared to the SVR model and standard DA indicate improved performance of the proposed model.

You do not currently have access to this content.