Research On Runoff Prediction Based On BP Network

In various people’s livelihood project hydrological forecasts plays a very important role, which is provide important reference information such as regulating runoff, flood disaster, defense using river water, irrigation provides. And it also the basis for the implementation of various scheduling decisions. It is meaningful for runoff forecast accuracy in hydrologic forecasting. Hydrological element has a strong non-linear characteristics. And it produces a more complex process, using traditional statistical conceptual hydrological model established in dealing with practical situations encountered such as:geographical location, climate and other factors. In this paper, the basic characteristics and theoretical (ANN) and BP neural network algorithm based on artificial neural network design of BP neural network model with forecasting capabilities, the model were used gradient descent algorithm and LM (Levenberg-Marquardta) algorithm as the network training algorithms.In this paper, the basic characteristics and theoretical (ANN) and BP neural network algorithm based on artificial neural network design of BP neural network model with forecasting capabilities. The model used gradient descent algorithm and LM (Levenberg-Marquardta) algorithm as the network training algorithms. BP neural network is easy to fall for the lack of local minima. The use of genetic algorithms (GA) has the characteristics of strong global search capability. The GA algorithm and BP algorithm combines design with function of GA-BP forecasting model. And the design of the model is applied to forecast runoff where in Tongren City, Guizhou Province Sinan hydrological station.A comprehensive information management system was developed According to the requirements of the project. For runoff forecasting system function method proposed algorithm runoff prediction model of BP network in the system based implementation. Through this system can improve the ability to integrate the measured data, reduce the chance of human intervention data to improve timeliness and accuracy of forecasting data, providing real-time data for effective flood prevention work, provide a reliable basis for the establishment of flood control decision-making.