Soil Moisture Forecast And Crops Irrigation Schedule Multi-objective Optimization

Hebei Plain is an important grain production base in north China.In addition to thelower rainfall,the falling underground water level these years,the local economy takes upirrigation water as the the rapid development leads to irrigation water availability cannotmeet the needs of irrigation which makes conflict between people blindly pursue high yieldof grain and the shortage of water obviously. So how to solve the agricultural water-savingproduction, use the limited agricultural irrigation water more efficiently become a hotspotin the research of the agriculture and water conservancy technical personnel problems.In this paper, the author carry out the spatial variation law of soil moisture andno-sufficient irrigation experimental study based on Wangdu irrigation experimentalstation. Statistical analysis of spatial variation regularity of different depth of soil layerwater content through measured the following soil water content of0-30cm,30-50cm and50-80cm three different depth.It is concluded that spatial distribution of soil moistureconform to the law of the approximate normal distribution. Find out different spatialcharacteristics of soil and related distance based on half variance analysis and establish thestructure model of spatial variability forsoil water which provide t basis for the selection oftest station spacing and depth of soil moisture.The measured and the related to soil moisture content wind speed, evaporation,rainfall, air temperature, relative humidity and sunshine time as the input variables and soilmoisture content as the output variable establish a BP neural network prediction model ofsoil moisture. The trained network model use the soil water balance model which the nextstep of irrigation schedule optimization model involve.The water production function models of winter wheat and summer maize wasestablished based on no-sufficient irrigation test data in Wangdu irrigation station.Sensitive index is an important parameter in the model which reflect the degree of watershortage impact on the production in a phase. SM-PSO algorithm were used to solvesensitive index of Jensen model aiming at the problems of partial estimation and low fittingprecision for traditional regression analysis algorithm.The results show that SM-PSOalgorithm had higher fitting precision than traditional regression analysis algorithm due tocombined local searching superiority of simplex and overall searching superiority of PSOalgorithm.Based on Jensen model and soil water balance model, a multi-objective optimizationmodel of irrigation scheduling for winter wheat-summer maize rotation was established which use the irrigation date and volume as decision variables and the least irrigation waterin a year and the maximum crop output value as decision goal. The optimization modelsolved with group non-dominated sorting genetic algorithmⅡ（GNSGA-Ⅱ）. Finally, theexact irrigation volume and irrigation dates were obtained under different availableirrigation amount and limited irrigation frequency in a year which can provide decisionirrigation reference for decision makers.Finally, water-saving technology and management models are discussed fromwater-saving engineering measures, water-saving agronomic measures and water-savingmanagement measures of the three aspects of Hebei plains.