基于XGBoost的水禽养殖粉尘预测研究  

XGBoost based dust prediction for waterfowl breeding

作　　者：李祥铜 曹亮 李湘丽 刘双印 徐龙琴 黄运茂 LI Xiangtong;CAO Liang;LI Xiangli;LIU Shuangyin;XU Longqin;HUANG Yunmao(College of Information Science and Technology, Zhongkai University of Agriculture and Engineering, Guangzhou 510225, China;Intelligent Agriculture Engineering Research Center of Guangdong Higher Education Institutes, Guangzhou 510225, China;Guangzhou Key Laboratory of Agricultural Products Quality & Safety Traceability Information Technology, Guangzhou 510225, China;Guangdong Province Key Laboratory of Waterfowl Healthy Breeding, Guangzhou 510225, China;College of Animal Science, Zhongkai University of Agriculture and Engineering, Guangzhou 510225, China;Library, Zhongkai University of Agriculture and Engineering, Guangzhou 510225, China;College of Mechanical and Electric Engineerings, Shihezi University, Shihezi 832000, China)

机构地区：[1]仲恺农业工程学院信息科学与技术学院,广东广州510225 [2]广东省高校智慧农业工程技术研究中心,广东广州510225 [3]广州市农产品质量安全溯源信息技术重点实验室,广东广州510225 [4]广东省水禽健康养殖重点实验室,广东广州510225 [5]仲恺农业工程学院动物科技学院,广东广州510225 [6]仲恺农业工程学院图书馆,广东广州510225 [7]石河子大学机械电气工程学院,新疆石河子832000

出　　处：《仲恺农业工程学院学报》2021年第2期11-16,共6页Journal of Zhongkai University of Agriculture and Engineering

基　　金：国家自然科学基金(61871475、61471133);广东省科技计划(2015A040405014、2016A070712020、2017B010126001、2017A070712019、2019B020215003);广东省教育科学“十三五”规划项目(2018GXJK072);广东省学位与研究生教育改革研究项目重点项目(2019JGXM64);广东省教育厅科研项目(2016KQNCX075、2017GCZX001、2017KTSCX094、2017KTSCX095、2017KQNCX098、ZHNY1903);广州市科技计划(201903010043、201905010006);教育部产学合作协同育人项目(201802235009、201802153181、201802153182、201802153191)资助.

摘　　要：为了提高水禽养殖中粉尘预测精度,提出基于XGBoost的水禽养殖粉尘预测模型.通过对粉尘相关参数进行相关性分析,提取出更重要的参数进行预测,简化了模型,降低计算难度,然后将归一化后的数据输入模型进行训练优化,最后通过与其他传统模型进行对比分析,提出的预测模型评价指标平均绝对百分比误差、平均绝对误差和均方根误差分别为0.0104、0.1902、0.2406,均低于对比模型,验证了提出的XGBoost模型对于水禽养殖粉尘预测具有很好的预测精度与鲁棒性能,为水禽养殖智能化提供一种新的有效方法.In order to improve the accuracy of dust prediction in waterfowl breeding,a dust prediction model for waterfowl breeding based on XGBoost was proposed.More important parameters were extracted for prediction,which simplified the model to reduce the calculation difficulty.Then,the normalized data,which was inputted into the model for training and optimization,were compared with that of other traditional models finally.The average absolute percentage error,the average absolute error,and the root mean square error of the prediction model evaluation indicators were 0.0104,0.1902,and 0.2406,respectively,which were lower than that of the comparative model.The results verified that the XGBoost model proposed had better prediction accuracy and robust performance,which could provide a novel effective method for the intelligentization of waterfowl breeding.

关 键 词：粉尘预测 XGBoost 水禽养殖 特征相关性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]