Two strategies to avoid overfitting in feedforward networks
Neural Networks
Prediction of parameters characterizing the state of a pollution removal biologic process
Engineering Applications of Artificial Intelligence
| Hi-index | 0.00 |
The present work takes part in a global development of reliable and robust tools allowing real-time controlling and supervising of the Tet catchment area, the main river of the Pyrenees-Orientales department (Southern France). The impact of the Tet on the department life is significant and the management of its water quality must be largely improved and better supervised. The main purpose of the work was to develop ''rain flow'' predictive models, using Elman recurrent neural networks and based on the identification of localized rain events. These neural models allow understanding the dynamic evolution, according to rain events, of the Tet flow at a selected point and of the Perpignan WWTP (WasteWater Treatment Plant) influent flow. Their most interesting characteristic is their capability to predict big increases in river flow and in plant influent flow. The neural models have been integrated as thematic layers in a geographical information system (GIS) allowing an efficient management and update of the records used to develop the models.