Facies-Controlled Modeling for Permeability of Tight Gas Reservoir Based on Hydrodynamic and Geophysics Characteristics

Table 1

Important parameters of the BP network learning process.

Weights			Learning rate	Allowable error	Maximum number of training	Number of hidden layer neurons
Train set (%)	Validation set (%)	Test set (%)
70	10	20	1.00 × 10^−1	1.00 × 10^−3	1000	15