Method for predicting tight oil fracturing range based on DL model of physical constraint

Abstract

The invention provides a method for predicting a tight oil fracturing range based on a DL model of physical constraints, and belongs to the technical field of tight oil fracturing. The method comprises the following steps: firstly, providing a reasonable physical hypothesis for a studied problem and setting related parameters, then establishing a two-dimensional axisymmetric numerical calculation model, dividing a network and setting time steps, establishing a deep learning model, setting a structure and other parameters of a deep neural network model, and obtaining a prediction result by using a test set; and evaluating a prediction result by using an L2 norm and a decision coefficient R2 as evaluation indexes, comparing the prediction precision of the neural network under the condition of considering the physical constraint and the prediction precision of the neural network under the condition of not considering the physical constraint, and finally changing the training data size on the basis of the neural network added with the physical constraint and measuring the prediction capability of the neural network. The method can be applied to rapid prediction of flow field distribution of two-drive model partitions, is high in prediction accuracy and adaptability, is high in calculation speed, and can well solve the problem of unknown tight oil flow field distribution and the problem of small sample prediction.

Description

technical field [0001] The invention relates to the technical field of tight oil fracturing, in particular to a method for predicting the range of tight oil fracturing based on a physically constrained DL model. Background technique [0002] In most cases, the deep learning algorithm is considered as a black box without considering any prior knowledge, including physical equations, empirical formulas, etc. The learning process leads to the correct solution and enables it to learn all the rules between input and output, and can effectively remove the false solutions of the system. And models trained from training data alone cannot guarantee that the fundamental laws of physics relevant to engineering problems are satisfied. In addition, in most engineering applications, data acquisition is a time-consuming task with high cost and high cost. Therefore, how to effectively learn the relationship between data from small samples has become a key issue. Contents of the invention...

AI Technical Summary

Problems solved by technology

However, models trained from training data alone cannot guarantee that the fundamental laws of physics relevant to engineering problems are satisfied.

In addition, in most engineering applications, data acquisition is a time-consuming task with high cost and high cost. Therefore, how to effectively learn the relationship between data from small samples has become a key issue

Images