Research On Water And Fertilizer Integrated Irrigation Strategy Of Tea Plantation Based On Random Forest Algorithm

Intelligent irrigation with integrated water and fertilizer is one of the important means of modern planting production,the introduction of intelligent decision classification technology into agricultural production and field management is conducive to improving the efficiency of resource utilization,achieving intelligent water and fertilizer control in agricultural production,and avoiding excessive waste of water and fertilizer.At present,agricultural information analysis technology based on data mining is developing rapidly.It is of great significance to select suitable solutions to realize the control of water and fertilizer irrigation and improve the economic benefits of crops in combination with the research object.The demand for water and fertilizer in tea plantation differs depending on the environmental conditions during the growing period of the tea trees and the weather forecast conditions.Therefore,differentiated irrigation and fertilization should be implemented for tea trees.This article takes Taiping Houkui Tea as the research object to carry out research on the intelligent irrigation strategy of integrated water and fertilizer for tea plantation.Construct a Random Forest algorithm model to classify the water and fertilizer needs of the tea plantations,reasonably control the water and fertilizer supply,and realize the scientific management of water and fertilizer supply in tea plantations.The main work tasks and results in this thesis are as followed:1)The data used in this thesis are the environmental data,soil data and weather forecast data of the Houkui tea plantation in Taiping area.It is necessary to preprocess the obtained source data,use Laida criteria to deal with data outliers,and at the same time fill in the deleted data and the vacancy value by the average interpolation method to improve the accuracy of the data.2)There are many factors that affect the demand for water and fertilizer of tea trees,such as light intensity,soil temperature and humidity,and weather forecast data and so on.The mutual information method is used to determine the factors affecting the water and fertilizer demand of the tea garden.Sampling processing is required due to the imbalance of the fertilization data of the tea plantation.The character data in the collected data is encoded exclusively,and the final determined series of impact factors are used as input,and the demand levels of moisture and fertilizer are used as output,to build decision classification models separately.3)This thesis compares the model of Random Forest algorithm,Logistic Regression algorithm and SVM algorithm,respectively constructs irrigation decision model and fertilization decision model,and determines which algorithm model can achieve better classification through classification decision evaluation index,confusion matrix,ROC curve.Through experimental analysis,the Random Forest algorithm can make good decisions in both irrigation decisions and fertilization decisions,which could meet the requirements of intelligent fertigation and irrigation decisions in tea plantation.4)The water and fertilizer decisions obtained by the Random Forest algorithm are used for tea plantation irrigation and fertilization.Based on the advantages of water and fertilizer integration,the obtained irrigation and fertilization levels are matched according to the prior knowledge to obtain the total irrigation of water and fertilizer solutions,and then to fertilize and irrigate the tea garden.