CN-121998147-A - Behavior pattern prediction method, apparatus, device, storage medium, and computer program product

CN121998147ACN 121998147 ACN121998147 ACN 121998147ACN-121998147-A

Abstract

The application relates to the technical field of geothermal resource utilization and discloses a behavior mode prediction method, a device, equipment, a storage medium and a computer program product, wherein the method comprises the steps of obtaining historical behavior data of a geothermal well, normalizing the historical behavior data and obtaining target input data; and guiding the target input data into a preset deep learning neural network model to obtain prediction data, and predicting the behavior mode of the geothermal well according to the prediction data. The method is suitable for geothermal well behavior pattern recognition, combines existing production data of a geothermal well as priori information, utilizes a feedforward neural network algorithm to perform attribute fusion, predicts similar production environments or future production conditions of the same well, and provides scientific guidance for geothermal efficient development.

Inventors

LI YUEXIN
FANG CHAOHE
WANG SHEJIAO
DU GUANGLIN
MO SHAOYUAN
TIAN YANNING
ZHANG WENYAO

Assignees

中石油深圳新能源研究院有限公司
中国石油天然气股份有限公司

Dates

Publication Date: 20260508
Application Date: 20241106

Claims (10)

1. A behavior pattern prediction method, characterized in that the behavior pattern prediction method comprises: Acquiring historical behavior data of a geothermal well, and normalizing the historical behavior data to obtain target input data; Importing the target input data into a preset deep learning neural network model to obtain prediction data; and predicting the behavior mode of the geothermal well according to the prediction data.
2. The behavior pattern prediction method according to claim 1, wherein before the step of inputting the target input data into a preset deep learning neural network model to obtain predicted data, the method further comprises: constructing an initial deep learning neural network model based on a feedforward neural network algorithm; Training the initial deep learning neural network model by applying preset training data to obtain a training output value; And updating parameters of the initial deep learning neural network model according to the training output value to obtain a target deep learning neural network model.
3. The behavior pattern prediction method of claim 2, wherein the step of constructing an initial deep learning neural network model based on a feedforward neural network algorithm comprises: Setting a multi-layer sensor based on a feedforward neural network algorithm, wherein the multi-layer sensor comprises an input layer, a hidden layer and an output layer; and selecting a target loss function and an optimizer according to the target input data, and establishing an initial deep learning neural network model through the multi-layer perceptron, the loss function and the optimizer.
4. The behavior pattern prediction method according to claim 3, wherein the step of setting the multi-layer perceptron based on a feedforward neural network algorithm includes: Setting the dimension of an input layer of the multi-layer sensor based on the feature quantity of the target input data; selecting a proper activation function to design a hidden layer, and configuring a preset number of neurons for the hidden layer; Defining an output layer through a linear activation function according to the predicted task demand; and connecting the input layer, the hidden layer and the output layer based on a feedforward neural network algorithm to generate a multi-layer perceptron.
5. The behavior pattern prediction method of claim 2, wherein the step of updating parameters of the initial deep learning neural network model according to the training output value to obtain a target deep learning neural network model comprises: Calculating a first loss function according to the training output value; Determining a gradient of the first loss function relative to model parameters based on a back propagation algorithm; selecting a small batch of training data by using a small batch gradient descent method, adjusting parameters of the initial deep learning neural network model according to the gradient direction, and calculating a second loss function; and when the value of the second loss function is reduced to a preset loss threshold value, acquiring a target deep learning neural network model based on the adjusted parameters.
6. The behavior pattern prediction method of claim 5, further comprising, prior to the step of calculating a first loss function from the training output values: and adding the initial loss function into a square penalty term of the weight to conduct regularization, and obtaining a first loss function.
7. A behavior pattern prediction apparatus, the apparatus comprising: The data arrangement module is used for acquiring historical behavior data of the geothermal well, preprocessing the historical behavior data and acquiring target input data; the data prediction module is used for importing the target input data into a preset deep learning neural network model to obtain predicted data; and the behavior prediction module is used for predicting the behavior mode of the geothermal well according to the prediction data.
8. A behavior pattern prediction device, characterized in that the device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the behavior pattern prediction method according to any one of claims 1to 6.
9. A storage medium, characterized in that the storage medium is a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the steps of the behavior pattern prediction method according to any one of claims 1 to 6.
10. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the steps of the behavior pattern prediction method of any one of claims 1 to 6.

Description

Behavior pattern prediction method, apparatus, device, storage medium, and computer program product Technical Field The present application relates to the field of geothermal resource utilization technologies, and in particular, to a behavior pattern prediction method, apparatus, device, storage medium, and computer program product. Background At present, geothermal resource evaluation and exploration and development mainly establish a physical model according to a stratum heat and mass transfer mode, gridding numerical simulation is carried out, physical parameter support is required for simulation accuracy, but geothermal well production process influence factors are complex, and if the geothermal well production boundary and accurate gridding division are required to be met, huge calculation cost is required, and geothermal production simulation is difficult to popularize in a large scale. Disclosure of Invention The application mainly aims to provide a behavior mode prediction method, a device, equipment, a storage medium and a computer program product, and aims to solve the technical problems that the prior geothermal resource evaluation and exploration and development mainly establish a physical model according to a stratum heat and mass transfer mode, gridding numerical simulation is carried out, the simulation accuracy needs physical parameter support, but the geothermal well production process has complex influence factors, and the geothermal well production boundary and accurate gridding division are required to be met if the geothermal well production boundary and the accurate gridding division are required, huge calculation cost is required, and the geothermal production simulation is difficult to popularize in a large scale. To achieve the above object, the present application proposes a behavior pattern prediction comprising: Acquiring historical behavior data of a geothermal well, and normalizing the historical behavior data to obtain target input data; Importing the target input data into a preset deep learning neural network model to obtain prediction data; and predicting the behavior mode of the geothermal well according to the prediction data. Optionally, before the step of inputting the target input data into a preset deep learning neural network model to obtain the predicted data, the method further includes: constructing an initial deep learning neural network model based on a feedforward neural network algorithm; Training the initial deep learning neural network model by applying preset training data to obtain a training output value; And updating parameters of the initial deep learning neural network model according to the training output value to obtain a target deep learning neural network model. Optionally, the step of constructing an initial deep learning neural network model based on the feedforward neural network algorithm includes: Setting a multi-layer sensor based on a feedforward neural network algorithm, wherein the multi-layer sensor comprises an input layer, a hidden layer and an output layer; and selecting a target loss function and an optimizer according to the target input data, and establishing an initial deep learning neural network model through the multi-layer perceptron, the loss function and the optimizer. Optionally, the step of setting the multi-layer sensor based on the feedforward neural network algorithm includes: Setting the dimension of an input layer of the multi-layer sensor based on the feature quantity of the target input data; selecting a proper activation function to design a hidden layer, and configuring a preset number of neurons for the hidden layer; Defining an output layer through a linear activation function according to the predicted task demand; and connecting the input layer, the hidden layer and the output layer based on a feedforward neural network algorithm to generate a multi-layer perceptron. Optionally, the step of updating parameters of the initial deep learning neural network model according to the training output value to obtain a target deep learning neural network model includes: Calculating a first loss function according to the training output value; Determining a gradient of the first loss function relative to model parameters based on a back propagation algorithm; selecting a small batch of training data by using a small batch gradient descent method, adjusting parameters of the initial deep learning neural network model according to the gradient direction, and calculating a second loss function; and when the value of the second loss function is reduced to a preset loss threshold value, acquiring a target deep learning neural network model based on the adjusted parameters. Optionally, before the step of calculating the first loss function according to the training output value, the method further includes: and adding the initial loss function into a square penalty term of the weight to conduct regularization, and obtaining a