基于WOA-XGBoost的膜下滴灌棉花蒸散量预测模型  

Evapotranspiration prediction model of cotton under film drip irrigation based on WOA-XGBoost

作　　者：曹缘 王振华[1,2,3] 张继红 刘宁宁 李文昊 张金珠[1,2,3] CAO Yuan;WANG Zhenhua;ZHANG Jihong;LIU Ningning;LI Wenhao;ZHANG Jinzhu(College of Water Conservancy&Architectural Engineering,Shihezi University,Shihezi,Xinjiang 832000,China;Key Laboratory of Modern Water-saving Irrigation of Xinjiang Production&Construction Group,Shihezi,Xinjiang 832000,China;Key Laboratory of Northwest Oasis Water-saving Agriculture,Ministry of Agriculture and Rural Affairs,PR China,Shihezi,Xinjiang 832000,China)

机构地区：[1]石河子大学水利建筑工程学院,新疆石河子832000 [2]现代节水灌溉兵团重点实验室,新疆石河子832000 [3]农业农村部西北绿洲节水农业重点实验室,新疆石河子832000

出　　处：《排灌机械工程学报》2024年第12期1280-1286,共7页Journal of Drainage and Irrigation Machinery Engineering

基　　金：兵团重大科技项目(2021AA003);国家重点研发计划项目(2022YFD1900405,2021YFD19008003);国家自然科学基金资助项目(52279040);石河子大学高层次人才项目(RCZK202319)。

摘　　要：为了科学准确地预测膜下滴灌棉花蒸散量,基于鲸鱼优化算法(whale optimization algorithm,WOA)和极端梯度提升树(XGBoost),提出了WOA-XGBoost棉花蒸散量预测模型.采用最大互信息系数(maximal information coefficient,MIC)筛选影响棉花蒸散量的关键因素,依据相关系数排序构建输入组合,代入WOA-XGBoost模型进行模拟.并与XGBoost,SVM,WOA-SVM和PSO-XGBoost预测结果进行对比验证.结果表明:太阳辐射、最低气温、最高气温、相对湿度、风速和土壤温度与棉花蒸散量相关性较大,其MIC值分别为0.722,0.546,0.496,0.475,0.379和0.219,基于上述6个因素构建的WOA-XGBoost模型综合性能最优,R^(2),MAE,RMSE和MAPE分别为0.922,0.038 mm/h,0.064 mm/h和0.221,预测精度均优于相同输入参数下的其他4种模型.因此,推荐使用WOA-XGBoost模型模拟相关因素与膜下滴灌棉花蒸散量之间的非线性关系.研究可为精确计算膜下滴灌棉花蒸散量提供科学依据,为灌溉决策优化提供参考.To accurately predict cotton evapotranspiration under mulched drip irrigation,a WOA-XGBoost cotton evapotranspiration prediction model based on the whale optimization algorithm(WOA)and the extreme gradient boosting tree(XGBoost)was proposed.The maximal mutual information coefficient(MIC)was utilized to identify the key factors impacting cotton evapotranspiration,and the input combinations were formulated based on the order of correlation coefficients and inputted into the WOA-XGBoost model for simulation.The prediction results were compared and verified with those obtained from XGBoost,SVM,WOA-SVM,and PSO-XGBoost modes.The results show that solar radiation,minimum and maximum air temperatures,relative humidity,wind speed,and soil temperature are highly correlated with cotton evapotranspiration,exhibiting MIC values of 0.722,0.546,0.496,0.475,0.379,and 0.219,respectively.The WOA-XGBoost model constructed on the basis of the above six factors has the best overall performance with R^(2),MAE,RMSE,and MAPE of 0.922,0.038 mm/h,0.064 mm/h and 0.221,respectively.The predictive accuracy surpass the other four models utilizing the same input parameters.Therefore,it is recommended to use the WOA-XGBoost model to simulate the non-linear relationship between the relevant factors and the evapotranspiration of cotton under film drip irrigation.This study offers a scientific foundation for accurately calculating cotton evapotranspiration under mulched drip irrigation and serves as a reference for optimizing irrigation decisions.

关 键 词：蒸散量 棉花 极端梯度提升树模型 鲸鱼优化算法 预测模型 

分 类 号：S274.4[农业科学—农业水土工程]