A Prediction Method of Remaining Oil Distribution Based on Autoregressive Network Model

Abstract

The invention discloses a method for predicting the distribution of remaining oil based on an autoregressive network model, which belongs to the technical field of oil reservoir development. The simulator builds a sample library; builds an autoregressive network model of a convolutional neural network and a convolutional long short-term memory kernel to capture the complex nonlinear mapping relationship between input data and output data; trains the constructed neural network model in the training set; In the test sample set, the minimum absolute value error L1 and relative L1 error are used to evaluate the performance of the trained surrogate model; the autoregressive network model that has been trained and has good evaluation performance is output, the reservoir data is collected in real time, the model is input, and the remaining oil distribution is predicted in real time. The invention can greatly shorten the remaining oil distribution prediction time, thereby shortening the time of the automatic history matching process of the oil reservoir that needs to perform multiple oil reservoir production predictions.

Description

technical field [0001] The invention belongs to the technical field of oil reservoir development, and in particular relates to a method for predicting the distribution of remaining oil based on an autoregressive network model. Background technique [0002] When the numerical simulation method is used to calculate the reservoir performance, due to the limitation of people's understanding of the reservoir geological conditions, the physical property parameters of the oil layer used in the simulation calculation may not accurately reflect the actual situation of the oil reservoir. Therefore, There is still a certain difference between the simulation calculation results and the actual observed reservoir dynamics, sometimes even a big difference. The dynamic prediction made on this basis is bound to be not completely accurate, and even lead to wrong conclusions. To reduce this discrepancy and make the dynamic predictions as close to reality as possible, a history-matching approa...

AI Technical Summary

Problems solved by technology

[0006] The existing reservoir remaining oil distribution prediction proxy model can only consider the reservoir geological static parameters, and the reservoir remaining oil distribution prediction proxy model cannot be used for the dynamic parameters, and the traditional reservoir numerical simulation calculation involves many grids and calculation Due to the shortcomings of large amount and long time consumption, the present invention proposes a method for predicting remaining oil distribution based on autoregressive network model, which can improve the performance of existing proxy models and is effective in the task of predicting remaining oil distribution and history matching in reservoirs. Improve calculation speed and save calculation time

Images