| Return water of irrigation in domestic and foreign large and medium-sized irrigation areas has been a widespread problem,water laoss happened as a result of the poor quality of channels and serious seepage.amount of return water raised steadily with the increased areas of irrigation and quantity of non-agricultural water.A lot return water not only make the water waste burt also will make the high underground water level in irrigation area and exacerbate the non-point source pollution level which infliuence the agricultural production and the ecological system.Because of many components of drainage water volume in the irrigation district,the reduced water production,flow and lateral infiltration process is very complicate,for study of drainage water volume in the irrigation district is very complex.As of now,the research on the drainage of irrigation areas is not widely available,and the research on the prediction of the drainage area is relatively low.However,the study of the water withdrawal process,the composition of the water withdrawal,the regression pattern and the prediction model of the precipitation forecast for the typical high-yangcheng irrigation area in gansu are seldom reported.This report reveled the regularity of returned water baied on time scale and established medols to forecast the process of returned water.The main findings are as follows:(1)The typical high-lift irrigation water circulation system consists of three conversion processes: the interregional drawing water;Irrigation infiltration and channel drain;Return water.Rainfall and irrigation are infiltrated into the rift valley gravel and rock fissures and becomeparts of the groundwater.The water in the bedrock flows between the fissures,some of which flow into the ravine in the form of latent,and the rest of the water overflow into the surface runoff.In terms of groundwater in the mountain valleys,some of its directly by intercepting lead springs way for irrigation or drinking water needs,the rest is by the undercurrent form back into the river.(2)In the Water irrigation farmland the overall water is balance,and has a key position in the Yellow River irrigation water occupies,atmospheric precipitation is the secondary status,conversion on soil water,surface water and groundwater transforming frequent,large amount and hard to quantitative analysis,etc.In terms of surface water irrigation region,mainly includes: channel of water,irrigation area of abandoned water,precipitation,life and industrial sewage water and mountain torrent,mainly channel of water and irrigation field surface water.(3)Significant year periodic trends of drainage water volume in the irrigation district,showing a very typical irrigation dynamic equilibrium properties: in spring,the level of drainage water volume in the irrigation district has been low,up and down in the 400000 m3;After the spring irrigation,the area of the irrigation area was increased at a very high speed,up to 800,000 m3.Since late September,the area has been retreating from its annual peak.As the irrigation is finished,the total amount of water withdrawal has fallen back to the annual low of February.During the year,most of the retreats were concentrated in June to October,which accounted for 63.7 percent of the total.The annual changes in the irrigation area are gradually increasing,but the increase has slowed.(4)The factors that affect the amount of water withdrawal are mainly: the amount of water,the amount of evaporation and the rainfall.The influence factors by using grey correlation analysis,are defined the influence degree of the water back,showing that: one of the highest impact for diverted,followed by rainfall,influence the weakest for evaporation,this conclusion conclusion is consistent with the aid of related analysis.In addition,the main influencing factors is different in the annual and monthly withdrawal.The main control factors and function variables of the water discharge in the irrigation area are studied,and the establishment of input variables in the prediction model of the withdrawal is established.(5)In this paper,the applicable objects and conditions of each method were determined by the analysis and study of the commonly used water forecast method.Based on the characteristics of the high lift irrigation area having the typical and data collection and forecasting method of characteristics and applicability,finally select multiple stepwise regression model as a typical high lift irrigation years back the water volume forecast model,and support vector machine(SVM)model as the typical high lift irrigation month return water prediction model.(6)Using Matlab software to build irrigation area in multiple stepwise regression model for water,after the elimination of no significant impact on the model variables found that there was no significant relationship between annual precipitation and the annual evaporation.The model is tested and verified.The test verifies that the model of the annual regression model with multiple stepwise regression is characterized by simple structure and accurate prediction.The maximum relative error of model prediction was 8.23% and the mean relative error was 4.47%,and the minimum relative error was 0.70%.(7)According to the return water irrigation month forecast greatly influenced by all kinds of random factors such as climate,more random characteristics,and its data,use the development of the SVM toolbox based on Matlab to build suitable for small sample modeling of support vector machine(SVM)model to forecast month refund amount,and the test model,validate.inspection validation show that support vector institutions built on refund amount prediction model of the typical high lift irrigation month back water fitting effect is good,and the prediction accuracy.The maximum relative error of model prediction was 14.96%,the average relative error was 8.28%,and the minimum relative error was 0.81%. |