METHOD FOR CONTROLLING OPERATING MODES OF PRODUCTION AND INJECTION WELLS OF OIL FIELD AND MULTILAYER CYCLIC NEURAL NETWORK

摘要

FIELD: oil production.;SUBSTANCE: invention relates to predicting and controlling the flow rate of fluid in oil wells. To implement the method for controlling the operation of injection and production wells of an oil field, based on a control device having an artificial neural network with cyclic communication, a forecast of fluid flow rate in time is created. To create a forecast, while switching to the next calculated time step, the calculation results obtained on the output neuron for the previous time step are fed to the input layer of neurons of the current step. After training the neural network, an optimization problem is solved to determine the optimal injectivity of injection wells and the flow rate of the producing fluid, which ensures an increase in the oil flow rate. The obtained values of fluid flow rates and injectivity are set in the wells automatically or manually. A device for controlling well operation modes based on a neural network contains a multilayer cyclic neural network, including: the first input layer, the number of neurons of which is equal to the number of input data. Several hidden layers, the total number of which and the number of neurons contained on them are selected experimentally. The third output layer, containing one neuron, is responsible for predicting the fluid flow rate at the current time step. To take into account temporal effects, a cyclic connection was additionally introduced between the output neuron, which is responsible for the fluid flow rate at the previous time step, and the input neuron at the current time step.;EFFECT: invention improves forecast accuracy, provides the ability to predict changes in fluid flow rate over time, selection of optimal operating modes for production and injection wells, an increase in oil production. ;3 cl, 4 dwg