Study On Annual Runoff Forecast Model Of Changshui Hydrological Station In Luohe River Basin

With the development of science and technology and the progress of mankind,water resources security issues have received much attention.The implementation of medium and long-term runoff forecast can not only play an important role in reservoir dispatching,drought resistance,water supply,power generation and irrigation,but also effectively alleviate water use conflicts between various departments,so that limited water resources can maximize the comprehensive benefits.Therefore,this paper analyzes the hydrological characteristics and mid-and long-term runoff forecasting of the Luohe River Basin.Based on the runoff data from 1960-2016 in the Longshui Hydrological Station of the Luohe River Basin.Through the MannKendall trend test method,the interannual variation of runoff,precipitation and average temperature in Changshui Hydrological Station was analyzed.The results showed that the runoff and precipitation showed a significant downward trend,and the average temperature showed a significant upward trend;Through the Mann-Kendall mutation test analysis,it was found that the runoff and precipitation in the Changshui hydrological station all had mutations in 1978;The periodic analysis by wavelet analysis shows that there is a certain degree of similarity in the period of controlling the runoff and precipitation changes in the Luohe River Basin.Combined with the characteristics of runoff data in Changshui Hydrological Station,19 impact factors were selected,minimum pressure,precipitation,average pressure,average 2 minute wind speed,average temperature,average maximum temperature,daily precipitation > 0.1 mm days,maximum wind speed,percentage of sunshine,Great wind speed,maximum air pressure,etc.The correlation coefficient method and principal component analysis method were used to screen the influencing factors,and 10 correlation predictors with large correlations were obtained,which were used as scheme 1 and scheme 2.On the basis of the analysis of hydrological characteristics,a multivariate linear regression model,BP neural network model and ELM model based on scheme 1 and scheme 2 were established respectively,and the annual runoff total forecast was carried out.According to the forecast results,the two schemes and the three forecasting models are graded.The overall scheme 1 is better than the scheme 2.The qualification rates of the multiple linear regression model,BP neural network model and ELM model are 63.64%,72.73% and 90.91% respectively.Among them,BP neural network model and ELM model prediction results are relatively more suitable for the runoff characteristics of the basin.At the same time,the ELM runoff prediction model improves the BP neural network training time and the problem of easy to fall into local minimum.Therefore,the ELM model is superior to the BP neural network in terms of training speed and prediction accuracy.The research results of this paper have certain reference value for the macro-control of water resources in the Luohe River Basin and flood control and disaster mitigation.