Intelligent monitoring and judging method for perforation of oil and gas well
By building a high-precision wellhead vibration acquisition system and establishing a convolutional neural network model, the problems of perforation vibration signal attenuation and manual judgment of deep well oil and gas wells are solved, and the accuracy and automatic identification and judgment of perforation vibration signals are realized, improving the reliability and safety of monitoring.
Patent Information
- Application Number
- CN202510558387.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-29
- Publication Date
- 2025-06-17
AI Technical Summary
The perforation vibration signal of deep well oil and gas wells has severe attenuated, and existing monitoring methods are difficult to meet the needs of deep well monitoring, and manual judgments are likely to cause misjudgment, affecting the safety of the wellbore and the safety of construction personnel.
Build a high-precision wellhead vibration acquisition system to collect perforation vibration signals, obtain the time-frequency distribution map through short-time Fourier transform and grayscale processing, establish a convolutional neural network model, and conduct training and prediction to achieve automatic identification and judgment.
It realizes accurate extraction and automatic identification and judgment of the perforation vibration signals of deep wells, improves the accuracy and reliability of monitoring, reduces the risk of manual misjudgment, and ensures the safety of wellbore and construction personnel.
Smart Images
Figure CN120159394A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of perforation vibration monitoring for oil and gas wells, and particularly relates to an intelligent monitoring and judgment method for perforation of oil and gas wells. Background Technique
[0002] Perforation is a technology that uses the shaped charge effect of perforating charges to successively penetrate the casing, pollution zone, and part of the reservoir to establish a channel between the wellbore and the reservoir. More than 70% of oil and gas wells in China use perforation completion methods.
[0003] With the depletion of conventional oil and gas resources, the exploitation of deep oil and gas resources represented by shale oil and shale gas has become an effective means to ensure oil and gas supply. The increase in well depth has caused severe attenuation of perforation vibration signals, and previous perforation monitoring methods are difficult to meet the monitoring requirements of deep wells. In addition, manual judgment is prone to misjudgment, posing a threat to the safety of the wellbore and the lives of ground construction personnel.
[0004] Therefore, how to accurately extract the detonation characteristic signal and perform automatic recognition and judgment has become a major difficulty in current monitoring technologies. Summary of the Invention
[0005] The purpose of the embodiments of the present invention is to provide an intelligent monitoring and judgment method for perforation of oil and gas wells, aiming to solve the problems raised in the above background technique.
[0006] The embodiments of the present invention are implemented as follows. An intelligent monitoring and judgment method for perforation of oil and gas wells includes the following steps:
[0007] Step 1: Build a high-precision wellhead vibration acquisition system, collect the original signal containing perforation vibration characteristics, and establish a perforation vibration original signal database;
[0008] Step 2: Perform short-time Fourier transform on the collected wellhead vibration signal to obtain a windowed time-frequency distribution map;
[0009] Step 3: Perform gray-scale processing on the time-frequency distribution map, uniformly process the image size with a pixel of 512×512, and establish a convolutional neural network model according to the gray-scale markings;
[0010] Step 4: Use the trained convolutional neural network model to predict the perforation vibration signal.
[0011] A further technical solution is that the high-precision wellhead vibration acquisition system includes a computer, a high-precision acquisition card, and a high-precision vibration sensor;
[0012] The high-precision vibration sensor is installed on the four-way flange of the Christmas tree 4 for obtaining the original signal containing perforation vibration characteristics;
[0013] The high-precision vibration sensor is connected to the high-precision acquisition card through a communication cable, and the high-precision acquisition card is connected to the computer through a USB cable;
[0014] A data acquisition and data processing program is installed in the computer.
[0015] In a further technical solution, in step 3, establishing a convolutional neural network model specifically includes the following steps:
[0016] Step 3.1: Divide the convolutional neural network model into a training set and a test set, where the training set accounts for 80% and the test set accounts for 20%;
[0017] Step 3.2: The convolutional neural network contains 6 convolutional layers, continuously performing convolution, activation, and pooling. The convolutional kernel is a 3×3 matrix, the activation function is the rectified linear unit ReLU(x) = max(0, x), and the pooling layer uses max pooling, that is, the maximum value in each pooling window is selected as the output;
[0018] Step 3.3: Normalize the convolutional result so that it is in the gray scale interval of [0, 255]; perform 6 times of convolution and pooling in sequence;
[0019] Step 3.4: Flatten the result of the last convolutional layer (8×8) into a one-dimensional vector (64×1) as the fully connected layer;
[0020] Step 3.5: Establish a neural network corresponding model between the fully connected layer and the output result;
[0021] Step 3.6: Adjust the weights and bias values according to the loss calculated during training. The backpropagation of the adjusted weights includes the filter kernel weights used in the convolutional layer and the weights used in the fully connected layer;
[0022] Step 3.7: Test the trained convolutional neural network model with the test set, and calculate the error between the test value and the true value; until the error is within an acceptable range, a relatively ideal convolutional neural network training model is obtained.
[0023] In a further technical solution, step 4 includes the following steps:
[0024] Step 4.1: Acquisition of the wellhead perforation vibration signal;
[0025] Step 4.2: Perform a short-time Fourier transform on the perforation signal, and perform gray scale processing on the obtained spectrogram to make it a three-dimensional matrix with a size of 512×512 and a gray scale value of [0, 255];
[0026] Step 4.3: Put the gray scale map matrix into the trained convolutional neural network for prediction.
[0027] An intelligent monitoring and judgment method for perforating oil and gas wells provided by an embodiment of the present invention. This method obtains the original sample data of perforating signals under different working conditions through a wellhead vibration data acquisition system; performs short-time Fourier transform on the sample data to obtain a windowed time-frequency distribution map; performs gray-scale processing on the time-frequency distribution map; constructs a convolutional neural network model, determines the parameters of the convolutional neural network through a training set, and obtains a convolutional neural network model capable of predicting perforation; then uses the trained convolutional neural network model to predict the perforation vibration signal. This method is intelligent and reliable for perforation judgment, getting rid of the disadvantage of relying on manual judgment in the traditional wellhead perforation monitoring method. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 It is a schematic structural diagram of an intelligent monitoring and judgment method for perforating oil and gas wells provided by an embodiment of the present invention;
[0029] Figure 2 It is a schematic structural diagram of a high-precision wellhead vibration acquisition system constructed in an intelligent monitoring and judgment method for perforating oil and gas wells provided by an embodiment of the present invention;
[0030] Figure 3 It is a schematic diagram of a convolutional neural network model in an intelligent monitoring and judgment method for perforating oil and gas wells provided by an embodiment of the present invention.
[0031] In the drawings: computer 1; high-precision acquisition card 2; high-precision vibration sensor 3; Christmas tree 4; four-way flange 5. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0032] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0033] The following describes the specific implementation of the present invention in detail with reference to specific embodiments.
[0034] As Figure 1 shown, an intelligent monitoring and judgment method for perforating oil and gas wells provided by an embodiment of the present invention includes the following steps:
[0035] Step 1: Build a high-precision wellhead vibration acquisition system, collect the original signal containing perforating vibration characteristics, and establish a database of original perforating vibration signals;
[0036] Step 2: Perform short-time Fourier transform on the collected wellhead vibration signal to obtain a windowed time-frequency distribution map;
[0037] Step 3: Perform grayscale processing on the time-frequency distribution map, unify the image size to 512×512 pixels, and establish a convolutional neural network model according to the grayscale markings;
[0038] Step 4: Use the trained convolutional neural network model to predict the perforation vibration signal.
[0039] As Figure 2 shown, as a preferred embodiment of the present invention, the high-precision wellhead vibration acquisition system includes a computer 1, a high-precision acquisition card 2, and a high-precision vibration sensor 3;
[0040] The high-precision vibration sensor 3 is installed on the four-way flange 5 of the Christmas tree 4 for obtaining the original signal containing perforation vibration characteristics;
[0041] The high-precision vibration sensor 3 is connected to the high-precision acquisition card 2 through a communication cable, and the high-precision acquisition card 2 is connected to the computer 1 through a USB cable;
[0042] The computer 1 is installed with a data acquisition and data processing program.
[0043] As Figure 3 shown, as a preferred embodiment of the present invention, in the said Step 3, establishing the convolutional neural network model specifically includes the following steps:
[0044] Step 3.1: Divide the convolutional neural network model into a training set and a test set, where the training set accounts for 80% and the test set accounts for 20%;
[0045] Step 3.2: The convolutional neural network contains 6 convolutional layers, continuously performing convolution, activation, and pooling. The convolutional kernel is a 3×3 matrix, the activation function is the rectified linear unit ReLU(x) = max(0, x), and the pooling layer selects max pooling, that is, selects the maximum value in each pooling window as the output;
[0046] Step 3.3: Normalize the convolutional result to make it in the grayscale range of [0, 255]; perform convolution and pooling 6 times in sequence;
[0047] Step 3.4: Flatten the result of the last convolutional layer (8×8) into a one-dimensional vector (64×1) as the fully connected layer;
[0048] Step 3.5: Establish a neural network corresponding model between the fully connected layer and the output result;
[0049] Step 3.6: Adjust the weights and bias values according to the loss calculated during training. The backpropagation of the adjusted weights includes the filter kernel weights used in the convolutional layer and the weights used in the fully connected layer;
[0050] Step 3.7: Test the trained convolutional neural network model using the test set, and calculate the error between the test value and the true value; until the error is within an acceptable range, a relatively ideal trained convolutional neural network model is obtained.
[0051] As a preferred embodiment of the present invention, step 4 includes the following steps:
[0052] Step 4.1: Acquisition of the wellhead perforation vibration signal;
[0053] Step 4.2: Perform short-time Fourier transform on the perforation signal, and perform gray-scale processing on the obtained spectrogram to make it a three-dimensional matrix with a size of 512×512 and gray-scale values in the range of [0, 255].
[0054] Step 4.3: Put the gray-scale image matrix into the trained convolutional neural network for prediction.
[0055] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.
Claims
1. An intelligent monitoring and judgment method for oil and gas well perforation, characterized in that: The following steps are involved: Step 1: Build a high-precision wellhead vibration acquisition system to collect raw signals containing perforating vibration characteristics and establish a perforating vibration raw signal database; Step 2: Perform short-time Fourier transform on the collected wellhead vibration signal to obtain a windowed time-frequency distribution diagram; Step 3: grayscale the time-frequency distribution map, unify the image size to 512×512 pixels, and establish a convolutional neural network model based on the grayscale mark; Step 4: Use the trained convolutional neural network model to predict the perforating vibration signal.
2. The intelligent monitoring and judgment method for oil and gas well perforation according to claim 1 is characterized in that: The high-precision wellhead vibration acquisition system includes a computer, a high-precision acquisition card and a high-precision vibration sensor; The high-precision vibration sensor is installed on the four-way flange of the Christmas tree 4 to obtain the original signal containing the perforation vibration characteristics; The high-precision vibration sensor is connected to the high-precision acquisition card via a communication cable, and the high-precision acquisition card is connected to the computer via a USB cable; The computer is installed with data acquisition and data processing programs.
3. The intelligent monitoring and judgment method for oil and gas well perforation according to claim 2 is characterized in that: In step 3, establishing a convolutional neural network model specifically includes the following steps: Step 3.1: Divide the convolutional neural network model into a training set and a test set, with the training set accounting for 80% and the test set accounting for 20%; Step 3.2: The convolutional neural network contains 6 convolutional layers, which continuously perform convolution, activation and pooling. The convolution kernel is a 3×3 matrix, the activation function is the linear rectification function ReLU(x)=max(0,x), and the pooling layer uses maximum pooling, that is, the maximum value in each pooling window is selected as the output; Step 3.3: Normalize the convolution result to make it in the grayscale range of [0,255]; perform 6 convolutions and pooling in sequence; Step 3.4: Flatten the last convolution result into a one-dimensional vector as a fully connected layer; Step 3.5: Establish a neural network corresponding model between the fully connected layer and the output result; Step 3.6: Adjust the weights and bias values based on the loss calculated during training. The backpropagation of the adjusted weights includes the filter kernel weights used in the convolutional layers as well as the weights used in the fully connected layers. Step 3.7: Perform a test set test on the trained convolutional neural network model and calculate the error between the test value and the true value; until the error is within the specified range, the ideal convolutional neural network training model is obtained.
4. The intelligent monitoring and judgment method for oil and gas well perforation according to claim 3 is characterized in that: The step 4 comprises the following steps: Step 4.1: Collection of wellhead perforation vibration signals; Step 4.2: Perform short-time Fourier transform on the perforation signal, and grayscale the spectrum obtained by the transform to make it a three-dimensional matrix with a size of 512×512 and a grayscale value of [0,255]; Step 4.3: Put the grayscale image matrix into the trained convolutional neural network for prediction.