Physics-constrained indirect supervised learning

作     者：Yuntian Chen Dongxiao Zhang Yuntian Chen;Dongxiao Zhang

作者机构：Intelligent Energy LabFrontier Research CenterPeng Cheng LaboratoryShenzhen 518000China School of Environmental Science and EngineeringSouthern University of Science and TechnologyShenzhen 518055China 

出 版 物：《Theoretical & Applied Mechanics Letters》 (力学快报（英文版）)

年 卷 期：2020年第10卷第3期

页      面：155-160页

核心收录：

学科分类：12[管理学] 08[工学] 0831[工学-生物医学工程（可授工学、理学、医学学位）] 0830[工学-环境科学与工程（可授工学、理学、农学学位）] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 0707[理学-海洋科学] 081104[工学-模式识别与智能系统] 0815[工学-水利工程] 0805[工学-材料科学与工程（可授工学、理学学位）] 0813[工学-建筑学] 0802[工学-机械工程] 0824[工学-船舶与海洋工程] 0835[工学-软件工程] 0814[工学-土木工程] 0825[工学-航空宇航科学与技术] 0836[工学-生物工程] 0701[理学-数学] 0811[工学-控制科学与工程] 0801[工学-力学（可授工学、理学学位）] 0702[理学-物理学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：partially funded by the National Natural Science Foundation of China (Grants 51520105005 and U1663208) 

主　　题：Supervised learning Indirect label Physics constrained Physics informed Well logs 

摘      要：This study proposes a supervised learning method that does not rely on *** use variables associated with the label as indirect labels,and construct an indirect physics-constrained loss based on the physical mechanism to train the *** the training process,the model prediction is mapped to the space of value that conforms to the physical mechanism through the projection matrix,and then the model is trained based on the indirect *** final prediction result of the model conforms to the physical mechanism between indirect label and label,and also meets the constraints of the indirect *** present study also develops projection matrix normalization and prediction covariance analysis to ensure that the model can be fully ***,the effect of the physics-constrained indirect supervised learning is verified based on a well log generation problem.