Study on Data Mining and Intelligent Early Warning of Water Dispatch for the South-to-North Water Diversion Middle Route Project

Abstract

The South-to-North Water Diversion Middle Route Project is mainly open channel, the line is long, there are many risk sources along the line, and the water dispatch technology is difficult. It is great significant to deeply excavate the dispatch regulations and establish the intelligent early warning models by the modern information processing technology to ensure safe and stable water diversion. Considering the advantages of information diffusion technology in dealing with incomplete sample data, the regression model of radial basis function neural network based on information diffusion is proposed and applied to the analysis and calculation of discharge in water dispatch. Before calculating by neural network, the training sample data are fuzzed by information diffusion to improve the generalization ability of the model. The information diffusion approximate reasoning model is used to clean the real-time data. On the basis of data treatment by information diffusion, the multi-directional intelligent early warning model including water level, flow and opening is established to timely find the abnormalities in the dispatch process and trigger the alarm. The results show that: the regression model suggested based on information diffusion and neural network can smooth the sample data better, and has high fitting accuracy and prediction ability. The intelligent dispatch early warning model can automatically and accurately capture the abnormalities of various early warning factors and give the corresponding alarms to ensure timely response in case of emergencies.