Oil Reservoirs and Exploitation of Oil Reservoirs

Well testing entered petroleum engineering in ۱۹۳۷ as a tool to understand the actual behavior of the reservoir in the face of changes in the well. Artificial neural networks with a hidden layer have the ability to solve most nonlinear problems. In this study, an artificial neural network with a hidden layer was used to determine the reservoir model from pressure-derived diagrams. The number of neurons in the output layer is equal to the number of reservoir models considered, while the number of hidden layer neurons is an optimization problem and the problem is complexity, the complexity of the relationship between input and output, the amount of data available for network training, and the amount of noise. Educational data depends. A small number of them may not be able to converge the network to the desired error, while a large number may lead to the network not becoming popular. The minimum data required for network training based on an exploratory method should be ten times the number of links in the network. In leading networks, if the mean relative error and the square error of the test data are plotted against the number of hidden layer neurons, a structure that provides the minimum measurement error value and the appropriate value of the regression coefficient is selected as the optimal structure. The appropriate training algorithm is determined by identifying the algorithm that requires the least time for training. In other words, an algorithm with the minimum required training.

Keywords:

well testing , Neural Network , Regression coefficient , Algorithm , Oil