Harvest prediction method, device and program

专利名称：Harvest prediction method, device and

program

发明人：宮崎 浩美,持田 宏平,井田 明,任 恩澤,小宮 悠,嶌田

洸一,金井 伸也

申请号：JP2019074473申请日：20190409公开号：JP2020171217A公开日：20201022

专利附图：

摘要：Problem to be solved: to predict the yield of crops using the model

constructed by machine learning, to effectively reduce the occurrence of error exceeding

tolerance and to make accurate prediction. In the model learning stageThe outcropprobability is calculated by the yield prediction model.The number of times that isoutside the allowable error range is the first evaluation axisThe mean square error of theprediction value and the result value of the yield is the second evaluation axis.Firstparameter, the outgoing probabilityWhen the correction factor for correcting theprediction value is a second parameterUsing at least one of the first and secondevaluation axesA combination of the first and second parameters is obtained.In theprediction operation stage, the outgoing value probability is compared with the firstparameter, and based on the result of the comparison, it is determined whether thecorrection value calculated by the yield prediction model or the second parameter isused for the predicted value of the producer.Diagram

申请人：株式会社セラク

地址：東京都新宿区西新宿７丁目５番２５号西新宿木村屋ビル６階

国籍：JP

代理人：星野 裕司

更多信息请下载全文后查看