Water-rich tight sandstone fluid discrimination method based on cost-sensitive learning

• Main Authors: WANG LEI, YI XI, CHEN XINGTING, LEI CANRU, WU JUAN, ZHAO DAN, LUO RENZE, LIAO BO, CAO RUI, LIN HONGYU, LIU HENG
• Format: Patent
• Language: Chinese; English
• Published: 01.12.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a water-rich tight sandstone fluid discrimination method based on cost-sensitive learning, and the method enables a water-rich tight sandstone fluid recognition problem to be analogous to a class imbalance classification problem, and proposes a cost-sensitive gradient boosting decision tree (PC-SC-GBDT) constructed based on parameters according to the thought of solving the class imbalance classification problem. Firstly, logging composite parameters are constructed on a data level, and a logging curve, the composite parameters and physical property parameters serve as input characteristics of a model at the same time; secondly, in an algorithm level, a sample weight updating mode in the gradient boosting decision tree is optimized, a recall rate index is introduced into a weight updating formula, and different weight updating modes are given to different types of training sample bases; and finally, performing hyper-parameter optimization on the model through multiple optimization algo