| Shaanxi Province is an excellent region for grape production in China.In recent years,the grape planting industry in Shaanxi Province has developed vigorously,and the planting area and yield have leaped to the third place in the country.However,with the continuous expansion of grape planting area,the grape industry is also affected by natural disasters,the main natural disaster is drought.For a long time,most managers only pay attention to the remedial measures after the disaster.It is of great practical significance to assess the risk of drought before the drought.Soil moisture is the most important monitoring index of agricultural drought.Objective and timely soil moisture information is an important prerequisite for understanding the degree and distribution of drought in Grape Garden and taking countermeasures,which is conducive to increasing the yield of grapes.How to select an appropriate model to predict soil moisture has become a problem to be solved in drought monitoring and early warning technology.In addition,China is short of water resources,large irrigation water consumption,and all rely on Farmers’ experience,at the same time,the management level is backward.In view of these problems,this paper establishes three different regression models of soil moisture,evaluates the advantages and disadvantages of the regression model by fitting effect,variable interpretability and other indicators,constructs the regression model of soil moisture and other factors,designs and realizes a set of drought monitoring and early warning system,and provides irrigation decisionmaking for crop planting.The main contents of this paper are as follows:(1)Modeling of soil moisture based on stepwise regressionSoil moisture is an important index to evaluate drought,which is affected by many environmental factors.Stepwise regression analysis studies the dependence of multiple variables to establish the optimal or appropriate regression model.Through stepwise regression modeling of soil moisture,analysis of fitting effect,multiple collinearity and variable coefficient can not explain the influence of dependent variables,lead to the other two regression models.(2)Soil moisture modeling based on ridge regressionThe purpose of ridge regression is to solve the problems of collinearity of the variables in the linear regression model,and the instability of the model parameters,which may lead to the violation of the symbol of the parameter estimator and the meaning of the variables.Ridge regression was used to model soil moisture,and the fitting effect was equal to that of stepwise regression.At the same time,the problem of variable symbol was solved,but the multicollinearity was still not solved,so the third regression model was used.(3)Soil moisture modeling based on partial least squares regression and comparison of three regression methodsPartial least squares regression mainly studies the regression modeling of multiple dependent variables to multiple independent variables.By comparing the three regression methods,it is found that partial least squares regression solves the multicollinearity problem,and the fitting effect of stepwise regression and ridge regression is the best.At the same time,ridge regression also solves the problem of variable symbol The reliability of the model is the best.(4)Design and development of drought monitoring and early warning systemIn this study,a drought monitoring and early warning system is designed and developed.The data collected by the two systems are transmitted to the PC software system through Zig Bee network.Various kinds of data and curves are visualized by programming,and historical data and curve query and data export are supported.The system can give early warning to the drought situation in the vineyard,and calculate the irrigation amount according to the regression model,and it can provide decision support for irrigation. |
