Research And Prediction Of Hydrological Data Based On The Hybrid Model Method

With the economic development and social progress,our life is getting better and better,but it comes with the environmental pollution,ecological destruction and waste of resources and other issues.Among those issues,the problem of water resources is the number one,and more and more people are concerned about the topic.In recent years,the hydrological geological disaster occurs frequently,which have brought great losses to people's lives and economy.The hydrological research is of great importance for flood control,drought resistance and the reasonable use of water resources.Thus,the forecasting of hydrological sequence is of great significance,and putting forward a kind of effective prediction method for hydrology data can play a guiding role for disaster prevention.However,as the complexity and instability of the hydrological data,it is difficult to receive good result for the prediction with the single model or simple hybrid model.Therefore,in recent years,the application of hybrid model is more and more.In view of this,this paper puts forward two kinds of new hybrid forecasting model,which has achieved good results.The first hybrid model is based on ensemble empirical mode decomposition(EEMD),radial basis function neural network(RBFN)and support vector machine(SVM),namely EEMD-RBFN-SVM method,which made an effective prediction for the hydrological data of the Yushu Tibetan Autonomous Region of Qinghai Province.The second hybrid model is the ‘cluster-forecast-integrated' integrated method,which is based on the K-Means algorithm,the hybrid SAM-ESM-RBFN model,and the hybrid EEMD-RBFN-SVM model.This method is used to forecast the monthly precipitation of Wu Daoliang in Qinghai Province in the next five years.In order to verify the validity of the method,the two models are compared with other simple and hybrid model,such as the RBFN,EEMD-RBFN,and the SAM-ESM-RBFN model,etc.The result shows that the proposed hybrid model has better generalization ability.