Artificial neural network modelling and flood water level prediction using extended Kalman filter

• Published in: 2012 IEEE International Conference on Control System, Computing and Engineering pp. 535 - 538
• Main Authors: Adnan, Ramli, Ruslan, Fazlina Ahmat, Samad, Abd Manan, Zain, Zainazlan Md
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2012
• Subjects: Adaptation models; Artificial neural networks; Back Propagation Neural Network (BPN); Data models; Extended Kalman Filter (EKF); Flood Modelling and Prediction; Floods; Forecasting; Kalman filters; Neural networks; Predictive models; Rain; Rivers
• ISBN: 9781467331425; 1467331422
• DOI: 10.1109/ICCSCE.2012.6487204

Accurate flood water level prediction are essential for reliable flood forecasting modelling. Although back propagation neural network (BPN) offer advantages for flood water level prediction, nonlinearity due to input parameters are the major issue to this modelling. A novel Extended Kalman Filter (EKF) optimization algorithm was employed in this study to overcome the nonlinearity problem and come out with an optimal ANN for the prediction of flood water level 3 hours in advance. The inputs used in the algorithm were current values of rainfall at the flood location and three upstream locations of river water levels. The BPN model was trained and tested successfully with Root Mean Square Error (RMSE) and loss function (V) close to zero.