Application of artificial neural network to the rapid feedback of potential ecological risk in flood diversion zone

As one of the ecological consequences due to intensified human activities in the upper catchments of the Yangtze River, the riverbed rising accompanied with soil erosion in the catchments has resulted in the significant increase of water level vs. the same flow discharge, which in turn enhanced the potential frequency of flood diversion and the uncertainty of flood risk in the diversion zone. In this paper, an artificial neural network (ANN) model based on radial-basis-function (RBF) was established for flood risk ranking at five safety polders. The computational speed for potential flood/ecological risk prediction as well as the feedback capability of integrated system was significantly enhanced through key techniques such as combination of the conventional hydrodynamic model and the ANN model, and integration of GIS and model. The site-specific and multi-site-specific RBF-ANN models developed herein not only provide useful tools for rapid prediction of flood routing process, but also show much promise for rapid feedback of the site-specific risk and real-time diversion process control. (C) 2003 Elsevier Ltd. All rights reserved.Automation & Control SystemsComputer Science, Artificial IntelligenceEngineering, MultidisciplinaryEngineering, Electrical & ElectronicSCI(E)EI0ARTICLE2105-1191