Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Chine → 2021 → Hydrological Drought Prediction Based on Hybrid Model in the Wei River Basin

Xi’an University of Technology (2021)

Hydrological Drought Prediction Based on Hybrid Model in the Wei River Basin

梁浩

Titre : Hydrological Drought Prediction Based on Hybrid Model in the Wei River Basin

Auteur : 梁浩

Grade : Master These 2021

Université : Xi’an University of Technology

Résumé partiel
Under the dual effects of climate change and human activities,global environmental degradation,water shortages and drought,which is closely related to human survival,have attracted the attention of many countries and regions.Hydrology drought with runoff as the main research element presents non-linear,non-stationarity and complexity.Therefore,it is always a hot and difficult issue to construct a high-accuracy prediction model for hydrological drought.Effectively identification of hydrological drought laws and accurate prediction of hydrological drought will help to alleviate the adverse effects of hydrological drought on regional socio-economic development,and play an important role in rational allocation of water resources in the river basin,as well as drought prevention and drought resistance.Taking the Wei River Basin in semi-arid region of China as the research area and hydrological drought as the research object.The hydrological drought index is constructed,the rationality of the index is verified,the characteristics of drought evolution are analyzed,and the applicability of a single model is evaluated.In addition,A variety of hybrid prediction models of hydrological drought were built,and the prediction effects of different models were compared.The main purpose is to provide a high-precision prediction model for the basin drought prediction,in order to provide technical support for the construction of the basin drought early warning system.The main research results obtained are as follows :(1)In the calculation of the hydrological drought index,the fitting effect of parametric and non-parametric methods for runoff distribution is compared,and the results show that the fitting effect of non-parametric kernel density estimation method is better.Therefore,the non-parametric kernel density estimator is used to calculate hydrological drought index instead of gamma and index estimators.The hydrological drought events in the basin are identified by the theory of run,and the rationality of the hydrological drought index is analyzed by comparing with the actual typical drought years in the basin.The trend and periodicity of hydrological drought in the basin is revealed by hydrological statistics method.(2)Three single statistical models of multiple linear regression,artificial neural network,and support vector machine(SVM)are established,and the best single model SVM is optimized among the three models.Based on the SVM model,the prediction effect of direct prediction of hydrological drought and prediction of runoff followed by calculation of hydrological drought are compared.It is found that direct prediction of hydrological drought is more advantageous.(3)Four decomposition methods,including empirical mode decomposition(EMD),adaptive noise total ensemble empirical mode decomposition(CEEMDAN),variational modal decomposition(VMD)and wavelet decomposition(WD),are combined with SVM model respectively to build four hybrid prediction models.Compared with the single statistical model,the hybrid model can improve the prediction accuracy of hydrological drought at the research site,and the VMD-SVM model has the best prediction effect among the four hybrid models.

Mots clés : Hydrological drought prediction ;decomposition method ;support vector machine ;hybrid model ;

Présentation (CNKI)

Page publiée le 22 octobre 2021