|
Ponomarev R.Yu.
|
ryponomarev@tnnc.rosneft.ru |
“Tyumen petroleum research center” LLC Tyumen |
|
Migmanov R.R.
|
|
“Tyumen petroleum research center” LLC Tyumen |
|
Ziazev R.R.
|
|
“Tyumen petroleum research center” LLC Tyumen |
DOI: 10.24412/2076-6785-2023-5-64-68
Abstract
The application of hybrid modeling — forecasting and optimization of well parameters using a combination of hydrodynamic and neural network modeling is considered. The hydrodynamic model serves as a reference model for training an artificial neural network. The trained neural network model is used in the future to solve the optimization problem based on the stochastic method of simulated annealing. The result is optimal technological parameters of well operation that ensure maximum oil production. The approach makes it possible to reduce machine time and improve the quality of justification of well operation modes necessary for effective field development.
Materials and methods
The analysis of the observed methods of neural network modeling in cases of field development was carried out, the architecture of the neural network was built to solve the problem, and multivariate calculations were carried out for the dynamics of the hydrodynamic model. Approbation and evaluation of the use of combined modeling in the problem of estimating a mining deposit of a field was carried out.
Keywords
neural network modeling, hydrodynamic modeling, field development
