Simulation of Flood Water Level Early Warning System Using Combination Forecasting Model

摘要

This research explores the use of BPNN and SVM techniques as a combined model using the Minimum Variance (MV) method to predict the upcoming flood water level events in Calinog River, Iloilo, Philippines. Rainfall and water level values are utilized as predictive variables to evaluate the performances of the individual models and the proposed combined-model as applied in the datasets. Root Mean Squared Error (RMSE) is used as a performance indicator of the trained models. Various simulation experiments are conducted to investigate the performance of the proposed model and the results show that the proposed combined-model of BPNN and SVM with their identified best control parameter values, produced a good predictive result as compared to the individual performances of SVM and the BPNN model. The proposed model yields better results that will surely help improve the effectiveness of the implementation of plans and policies of the disaster risk management of the local government unit and Iloilo Province as a whole.