Oil and gas well perforation impact load prediction method based on artificial intelligence

Abstract

The invention discloses an oil and gas well perforation impact load prediction method based on artificial intelligence. The method comprises the steps that a perforation impact load data set is established based on a finite element simulation model considering multiple parameters; performing feature analysis on the perforation impact load data set according to a Pearson's correlation coefficient to obtain a primary and secondary relationship of different parameters influencing the perforation impact load; normalizing the principal component data set by adopting a deviation standardization method, and dividing the normalized principal component data set into a training set, a verification set and a test set; constructing a perforation impact load analysis model, training the model by using the training set, optimizing model parameters by using the verification set, and testing a model effect by using the test set to obtain a completely trained perforation impact load analysis model; and inputting the oil and gas well influence parameters into the impact load analysis model to predict the perforation impact load. The perforation impact load calculation method which is more accurate, reliable and intelligent can be provided, and powerful support is provided for oil and gas well perforation intelligent prediction.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas well perforation model construction, and in particular to an oil and gas well perforation impact load prediction method based on artificial intelligence. Background Art

[0002] With the growth of global energy demand, the increasing depletion of resources and the increasingly stringent environmental regulations, traditional oil and gas exploration, mining and production methods face huge challenges. In order to meet these challenges, the oil and gas industry has gradually introduced artificial intelligence technology (AI technology) to process and analyze complex, multi-dimensional oil and gas data, explore the potential laws therein, and help the industry make more accurate decisions. For example, AI can be used to adjust parameters in real time during the drilling process to avoid equipment failure or catastrophic events, while reducing the interference of human operations; it can also be used to analyze real-time data, identify potential risk factors, and provide early warnings to provide intelligent emergency response plans in the event of emergencies. With the improvement of environmental protection requirements, AI technology can also optimize emission monitoring, waste management and other aspects to help oil and gas companies achieve sustainable development. In general, AI technology is gradually becoming the core technology of the oil and gas industry in various links such as exploration, mining, production, and management, helping the industry to achieve breakthrough progress in improving efficiency, ensuring safety, reducing costs and promoting sustainable development.

[0003] Perforation is a key step in establishing an effective oil and gas transmission channel between the wellbore and the oil and gas reservoir, and is responsible for improving the oil and gas development effect, increasing the single well production and oil and gas recovery rate. As perforation technology develops towards high hole density, large charge and deep penetration, the perforation impact load gradually increases, resulting in many string safety issues. Field practice shows that accurate calculation of perforation impact load can optimize perforation parameters and string combination design, improve wellbore stability, and contribute to safe perforation operations.

[0004] Therefore, in order to fill the shortcomings of the existing perforation impact load calculation method, the present invention proposes an oil and gas well perforation impact load prediction method based on artificial intelligence. By adopting a numerical simulation method to model the downhole perforation impact load calculation model, a perforation data set is generated, and a PSO-BP neural network algorithm is used to construct a downhole perforation impact load calculation method, so as to achieve more accurate, reliable and intelligent perforation impact load calculation, and provide strong support for intelligent prediction of oil and gas well perforation. Summary of the invention

[0005] In view of this, the present invention provides an oil and gas well perforation impact load prediction method based on artificial intelligence to solve the problem that the current perforation impact load calculation is inaccurate and requires a lot of manual intervention, resulting in the inability to provide technical support for safe perforation operations.

[0006] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides an artificial intelligence-based method for predicting oil and gas well perforation impact loads, comprising:

[0008] A perforation impact load data set is established based on a finite element simulation model that considers multiple parameters;

[0009] Performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain the principal component data set according to the primary and secondary relationships;

[0010] The principal component data set is normalized by using a deviation normalization method to eliminate scale differences between different features;

[0011] Divide the normalized principal component data set into training set, validation set and test set;

[0012] Constructing a perforation impact load analysis model based on a neural network, training the model using the training set, optimizing the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, testing the model effect using the test set, and obtaining a fully trained perforation impact load analysis model;

[0013] The influencing parameters of the oil and gas well are obtained, and the influencing parameters are input into the impact load analysis model to predict the perforation impact load.

[0014] Furthermore, a perforation impact load data set is established based on a finite element simulation model considering multiple parameters, including:

[0015] Randomly sample historical prior data to obtain multiple sets of simulation calculation parameters of different dimensions;

[0016] Establishing a finite element calculation model to calculate the perforation impact load corresponding to each group of simulation calculation parameters;

[0017] Each set of simulation calculation parameters and its corresponding perforation impact load are collected to establish a perforation impact load data set.

[0018] Furthermore, the perforation impact load data set is characterized by Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, including:

[0019] The Pearson correlation coefficients between different parameters and perforation impact loads were calculated respectively, and the calculation formula is expressed as:

[0020]

[0021] Among them, r is the Pearson correlation coefficient, n is the number of samples in the data set, and X i is the parameter value of the i-th sample, Y i is the perforation impact load value of the i-th sample, is the sample mean of the parameter values, is the sample average of the perforation impact load values.

[0022] Furthermore, the principal component data set is obtained by screening according to the primary and secondary relationship, including:

[0023] A heat map is drawn according to the calculation result of the Pearson correlation coefficient, in which different colors are used to distinguish the linear positive correlation and the linear negative correlation parameters, and different color depths are used to distinguish the strength of the correlation;

[0024] The linear influence of different parameters on the perforation impact load is distinguished according to the color and the depth of the color.

[0025] Furthermore, the principal component data set is normalized using a deviation normalization method, including:

[0026] The formula for the normalization process is:

[0027]

[0028] Among them, x is the current principal component value, x min is the minimum value of the parameter category to which the principal component belongs, x max is the maximum value of the parameter category to which the principal component belongs.

[0029] Furthermore, a perforation impact load analysis model based on neural network is constructed, including:

[0030] Based on BP neural network, an initial perforation impact load analysis model with input layer, hidden layer and output layer is constructed;

[0031] The weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm; wherein the fitness function of each particle is the loss function of the initial perforation impact load analysis model;

[0032] Forward propagation and back propagation of the model weights and biases output by the particle swarm optimization algorithm are performed, a loss function is calculated, and model parameters are updated iteratively through gradient descent;

[0033] When the number of iterations reaches the preset total number of iterations, a fully trained perforation impact load analysis model is obtained.

[0034] Furthermore, the weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm, including:

[0035] According to the fitness function value of the best particle in each round of iteration, a fitness curve reflecting the convergence status in the particle swarm optimization process is drawn.

[0036] In a second aspect, the present invention also provides an oil and gas well perforation impact load prediction system based on artificial intelligence, comprising:

[0037] A data set building module, used for building a perforation impact load data set based on a finite element simulation model considering multiple parameters;

[0038] An analysis module, used for performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient, obtaining the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain a principal component data set according to the primary and secondary relationships;

[0039] A normalization module, used for normalizing the principal component data set by using a deviation normalization method to eliminate scale differences between different features;

[0040] A data partitioning module is used to divide the normalized principal component data set into a training set, a validation set, and a test set;

[0041] A model building module is used to build a perforation impact load analysis model based on a neural network, train the model using the training set, optimize the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, and test the model effect using the test set to obtain a fully trained perforation impact load analysis model;

[0042] The prediction module is used to obtain the influencing parameters of the oil and gas wells and input the influencing parameters into the impact load analysis model to predict the perforation impact load.

[0043] In a third aspect, the present invention further provides an electronic device comprising a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, an artificial intelligence-based oil and gas well perforation impact load prediction method as described in any of the above technical solutions is implemented.

[0044] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, an artificial intelligence-based oil and gas well perforation impact load prediction method as described in any of the above technical solutions is implemented.

[0045] Compared with the prior art, the artificial intelligence-based oil and gas well perforation impact load prediction method proposed by the present invention, firstly, establishes a perforation impact load data set through finite element simulation, and considers the influence of multiple parameters, which can comprehensively analyze the contribution of different factors to the perforation impact load; secondly, the Pearson correlation coefficient is used to perform feature analysis on the data set, and the main component data set is screened out to avoid the interference of irrelevant features, which can improve the efficiency of the model and the interpretability of the model; thirdly, the deviation standardization method is used for normalization processing, which eliminates the influence of the difference in different feature scales, so that all parameters have the same measurement unit, which helps to improve the stability and convergence speed of model training. Finally, a neural network is used as a tool for perforation impact load analysis, which can be trained through a large amount of data to learn complex nonlinear relationships. The present invention can construct an accurate and efficient perforation impact load calculation method, lay a theoretical foundation for predicting perforation impact load using artificial intelligence methods, and provide important reference and technical support for the research of artificial intelligence large models in the oil and gas industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic flow chart of an oil and gas well perforation impact load prediction method based on artificial intelligence provided by the present invention;

[0047] Figure 2 The Pearson correlation coefficient matrix heat map provided by the present invention;

[0048] Figure 3 A network topology diagram of the initial perforation impact load analysis model provided by the present invention;

[0049] Figure 4 An algorithm flow chart of the perforation impact load analysis model provided by the present invention;

[0050] Figure 5 A schematic diagram of the fitness curve of the PSO-BP algorithm of the perforation impact load analysis model provided by the present invention;

[0051] Figure 6 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0052] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0053] See also Figure 1 This embodiment provides an artificial intelligence-based oil and gas well perforation impact load prediction method, including:

[0054] Step S101: establishing a perforation impact load data set based on a finite element simulation model taking multiple parameters into consideration;

[0055] Step S102: performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain a principal component data set according to the primary and secondary relationships;

[0056] Step S103: normalizing the principal component data set using a deviation normalization method to eliminate scale differences between different features;

[0057] Step S104: Divide the normalized principal component data set into a training set, a validation set, and a test set;

[0058] Step S105: constructing a perforation impact load analysis model based on a neural network, training the model using the training set, optimizing the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, testing the model effect using the test set, and obtaining a fully trained perforation impact load analysis model;

[0059] Step S106: using the impact load analysis model to predict the perforation impact load and obtain a fitness curve of the PSO optimization process.

[0060] The method of this embodiment, first, establishes a perforation impact load data set through finite element simulation, and considers the influence of multiple parameters, which can comprehensively analyze the contribution of different factors to the perforation impact load; secondly, the Pearson correlation coefficient is used to perform feature analysis on the data set, and the main component data set is screened out to avoid the interference of irrelevant features, which can improve the efficiency of the model and the interpretability of the model; thirdly, the deviation standardization method is used for normalization processing, which eliminates the influence of the difference in different feature scales, so that all parameters have the same measurement unit, which helps to improve the stability and convergence speed of model training. Finally, a neural network is used as a tool for perforation impact load analysis, which can be trained through a large amount of data to learn complex nonlinear relationships. The method of this embodiment combines artificial intelligence and machine learning technology to realize the automation and intelligence of the perforation process of oil and gas wells, reduce manual intervention, and improve work efficiency and accuracy. The model can quickly adapt and predict under new perforation conditions, greatly improving work efficiency.

[0061] As a preferred embodiment, in step S101, a perforation impact load data set is established based on a finite element simulation model considering multiple parameters, including:

[0062] Randomly sample historical prior data to obtain multiple sets of simulation calculation parameters of different dimensions;

[0063] Establishing a finite element calculation model to calculate the perforation impact load corresponding to each group of simulation calculation parameters;

[0064] Each set of simulation calculation parameters and its corresponding perforation impact load are collected to establish a perforation impact load data set.

[0065] As a specific embodiment, the value ranges of the input parameters of five dimensions, such as the total perforating charge, the initial wellbore pressure, the wellbore explosion space, the detonation interval time and the formation pressure, can be obtained according to the literature or field experimental data and empirical data as shown in Table 1.

[0066] Table 1 Numerical simulation parameters and their ranges

[0067] Influencing factors unit Minimum Maximum Total charge for perforation kg 2 20 Initial wellbore pressure MPa 10 100 Shaft explosion space m3 0.5 5 Detonation interval μs 10 50 Formation pressure MPa 10 100

[0068] We used random sampling to select 300 sets of numerical simulation calculation parameters, established a finite element calculation model to solve the corresponding perforation impact load, and collected perforation impact load data corresponding to different total perforation charges, initial wellbore pressure, wellbore explosion space, detonation interval time and formation pressure, thus forming a perforation impact load data set.

[0069] As a preferred embodiment, in step S102, the perforation impact load data set is subjected to feature analysis according to the Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, including:

[0070] The Pearson correlation coefficients between different parameters and perforation impact loads were calculated respectively, and the calculation formula is expressed as:

[0071]

[0072] Among them, r is the Pearson correlation coefficient, n is the number of samples in the data set, and X i is the parameter value of the i-th sample, Y i is the perforation impact load value of the i-th sample, is the sample mean of the parameter values, is the sample average of the perforation impact load values.

[0073] As a preferred embodiment, the principal component data set is obtained by screening according to the primary and secondary relationship, including:

[0074] A heat map is drawn according to the calculation result of the Pearson correlation coefficient, in which different colors are used to distinguish the linear positive correlation and the linear negative correlation parameters, and different color depths are used to distinguish the strength of the correlation;

[0075] The linear influence of different parameters on the perforation impact load is distinguished according to the color and the depth of the color.

[0076] As a specific example, Figure 2 As shown, Figure 2 A heat map is shown by plotting the results of the calculation of the Pearson correlation coefficient. Figure 2 Medium blue indicates linear positive correlation, red indicates linear negative correlation, and the depth of color indicates the strength of variable correlation. Figure 2 It can be seen that the total charge of perforation has the most significant linear effect on the downhole perforation impact load, followed by the initial pressure of the wellbore and the explosion space of the wellbore, while the detonation interval time and the formation pressure have a smaller linear effect on the downhole perforation impact load. This is consistent with other research results in this field, and further verifies the reliability of the data set generated by the numerical simulation in the present invention.

[0077] Since data normalization has an important impact on neural network training and performance, it can accelerate the training speed, improve the generalization ability of the model, and avoid gradient disappearance or explosion. Taking the Sigmoid function commonly used in neural networks as an example, when the input value is too large or too small, the derivative of the Sigmoid function is close to zero, causing the gradient to disappear and thus unable to converge. As a preferred embodiment, the present invention uses a deviation standardization (Min-Max Normalization) method in step S103 to normalize the data set, and linearly transforms the numerical value to between 0 and 1, and calculates as follows:

[0078]

[0079] Among them, x is the current principal component value, x min is the minimum value of the parameter category to which the principal component belongs, x max is the maximum value of the parameter category to which the principal component belongs.

[0080] After normalization, the scale differences between different features are eliminated, avoiding the dominant influence of certain features on the network weight update and improving the stability of training.

[0081] As a specific embodiment, in step S104, the normalized perforation data set is divided into a training set, a validation set, and a test set according to a ratio. In practice, 210 sets of data are used to form a training set for model training. 45 sets of data are used to form a validation set for optimizing the number of hidden layer neurons, connection weights, and neuron biases of the neural network. 45 sets of data are used to form a test set for testing the model effect.

[0082] As a preferred embodiment, in step S105, a perforation impact load analysis model based on a neural network is constructed, including:

[0083] Based on BP neural network, an initial perforation impact load analysis model with input layer, hidden layer and output layer is constructed;

[0084] The weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm; wherein the fitness function of each particle is the loss function of the initial perforation impact load analysis model;

[0085] Forward propagation and back propagation of the model weights and biases output by the particle swarm optimization algorithm are performed, a loss function is calculated, and model parameters are updated iteratively through gradient descent;

[0086] When the number of iterations reaches the preset total number of iterations, a fully trained perforation impact load analysis model is obtained.

[0087] like Figure 3 As shown, Figure 3 The network topology diagram of the initial perforation impact load analysis model based on the BP neural network, which contains input layer, hidden layer and output layer, is shown. The BP neural network uses the back propagation method to update the connection weights between neurons and the bias of neurons. Through continuous iterative operations, the loss function is minimized, and the performance of the artificial neural network is improved.

[0088] Specifically, the loss function of the perforation impact load analysis model is:

[0089]

[0090] in, is the predicted value, y i is the actual value, and n is the number of samples.

[0091] The weight update formula from input layer neuron i to hidden layer neuron j is:

[0092]

[0093] in, is the loss function L with respect to weight ω ij The partial derivative of , indicating that the loss function is in ω ij The rate of change in direction, η is the learning rate, O i is the output of the ith neuron.

[0094] Similarly, the bias update formula is:

[0095]

[0096] Hidden layer error δ j for:

[0097]

[0098] in, is the predicted output of the jth output layer neuron (i.e., the output of the activation function), O j is the actual output (i.e., target value) of the jth output layer neuron, is the derivative of the sigmoid activation function, which indicates the sensitivity of the output of the output layer neurons to the input. Represents the difference between the predicted output and the actual output.

[0099] In order to optimize the performance of the BP neural network, the present invention uses the PSO optimization method (particle swarm optimization) to optimize the initial parameters of the BP neural network, and uses the optimized BP neural network for training. Figure 4 As shown, Figure 4 The algorithm flow chart of the perforation impact load analysis model based on the PSO-BP neural network provided by the present invention is shown. By combining the global search capability of the PSO algorithm and the characteristics of the BP neural network in quickly and efficiently searching for local optimal solutions, the PSO-BP neural network can converge the loss function to the global minimum point during the training process.

[0100] As a preferred embodiment, in the process of training the perforation impact load analysis model, the weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm, including:

[0101] According to the fitness function value of the best particle in each round of iteration, a fitness curve reflecting the convergence status in the particle swarm optimization process is drawn.

[0102] When the number of neurons in the hidden layer is too small, it is difficult for the neural network to capture complex data relationships, resulting in underfitting of the model. When the number of neurons is too large, it is difficult for the PSO algorithm to obtain the best optimization result, thereby reducing the accuracy of the model; at the same time, too many neurons may cause the gradient to be unstable during the back propagation process, resulting in gradient disappearance or gradient explosion problems. When the number of neurons in the hidden layer is 20, the PSO-BP network shows good results on the training set and the validation set. Therefore, the perforation impact load analysis model proposed in this embodiment is selected to calculate the perforation impact load. The fitness curve of the PSO optimization process is shown as follows: Figure 5 As shown in the figure. The fitness curve is used to display and analyze the changing trend of the objective function or fitness function during the algorithm optimization process. It is an important tool to measure whether the optimization process is effective, whether the search converges, and the performance of the algorithm. Figure 5 It can be seen that with the increase of the number of iterations, the objective function shows a convergence state, which shows that the optimization process has a good effect.

[0103] In order to further verify the actual effect of the method of the present invention, three sets of empirical formulas for perforation impact load were selected for comparison with the perforation impact load analysis model, and verified by the measured data of two example wells. Through the calculation of example wells 1 and 2, the calculation results of the empirical formula and the neural network model were obtained, and the errors were obtained by comparison with the measured data. The specific results are listed in Table 2.

[0104] Table 2 Calculation results of perforation impact load by different methods

[0105]

[0106]

[0107] According to the comparative analysis results in Table 2, it can be seen that compared with the empirical formula method, the perforation impact load intelligent calculation model based on the PSO-BP neural network of the present invention has higher accuracy, and the comparison error is within a reasonable range.

[0108] This embodiment also provides an oil and gas well perforation impact load prediction system based on artificial intelligence, including:

[0109] A data set building module, used for building a perforation impact load data set based on a finite element simulation model considering multiple parameters;

[0110] An analysis module, used for performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient, obtaining the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain a principal component data set according to the primary and secondary relationships;

[0111] A normalization module, used for normalizing the principal component data set by using a deviation normalization method to eliminate scale differences between different features;

[0112] A data partitioning module is used to divide the normalized principal component data set into a training set, a validation set, and a test set;

[0113] A model building module is used to build a perforation impact load analysis model based on a neural network, train the model using the training set, optimize the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, and test the model effect using the test set to obtain a fully trained perforation impact load analysis model;

[0114] The prediction module is used to obtain the influencing parameters of the oil and gas wells and input the influencing parameters into the impact load analysis model to predict the perforation impact load.

[0115] like Figure 6As shown, the above-mentioned oil and gas well perforation impact load prediction method based on artificial intelligence, the present invention also provides an electronic device 600, which can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer and a server. The electronic device includes a processor 601, a memory 602 and a display 603.

[0116] In some embodiments, the memory 602 may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory 602 may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device. Further, the memory 602 may also include both an internal storage unit of the computer device and an external storage device. The memory 602 is used to store application software and various data installed on the computer device, such as program codes installed on the computer device. The memory 602 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a method program 604 for predicting the impact load of oil and gas wells perforation based on artificial intelligence is stored on the memory 602, and the method program 604 for predicting the impact load of oil and gas wells perforation based on artificial intelligence can be executed by the processor 601, thereby realizing a method for predicting the impact load of oil and gas wells perforation based on artificial intelligence in various embodiments of the present invention.

[0117] In some embodiments, the processor 601 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 602, such as executing an artificial intelligence-based oil and gas well perforation impact load prediction method program.

[0118] In some embodiments, the display 603 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 603 is used to display information on the computer device and to display a visual user interface. The components 601-603 of the computer device communicate with each other through a system bus.

[0119] This embodiment also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the artificial intelligence-based oil and gas well perforation impact load prediction method described in any of the above technical solutions is implemented.

[0120] The computer-readable storage medium and computing device provided according to the above-mentioned embodiments of the present invention can be implemented with reference to the specific description of the above-mentioned method for predicting the perforation impact load of oil and gas wells based on artificial intelligence according to the present invention, and have similar beneficial effects as the above-mentioned method for predicting the perforation impact load of oil and gas wells based on artificial intelligence, which will not be repeated here.

[0121] The artificial intelligence-based oil and gas well perforation impact load prediction method proposed in the present invention adopts a numerical simulation method to model the downhole perforation impact load calculation model, generates a perforation data set, and uses a PSO-BP neural network algorithm to construct a downhole perforation impact load calculation method, providing a more accurate, reliable and intelligent perforation impact load calculation method, and providing strong support for intelligent prediction of oil and gas well perforation.

[0122] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for predicting oil and gas well perforation impact load based on artificial intelligence, characterized in that: include: A perforation impact load data set is established based on a finite element simulation model that considers multiple parameters; Performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain the principal component data set according to the primary and secondary relationships; The principal component data set is normalized by using a deviation normalization method to eliminate scale differences between different features; Divide the normalized principal component data set into training set, validation set and test set; Constructing a perforation impact load analysis model based on a neural network, training the model using the training set, optimizing the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, testing the model effect using the test set, and obtaining a fully trained perforation impact load analysis model; The influencing parameters of the oil and gas well are obtained, and the influencing parameters are input into the impact load analysis model to predict the perforation impact load.

2. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 1 is characterized in that: The perforation impact load data set is established based on the finite element simulation model considering multiple parameters, including: Randomly sample historical prior data to obtain multiple sets of simulation calculation parameters of different dimensions; Establishing a finite element calculation model to calculate the perforation impact load corresponding to each group of simulation calculation parameters; Each set of simulation calculation parameters and its corresponding perforation impact load are collected to establish a perforation impact load data set.

3. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 1 is characterized in that: The perforation impact load data set is characterized by Pearson correlation coefficient to obtain the primary and secondary relationships of different parameters affecting the perforation impact load, including: The Pearson correlation coefficients between different parameters and perforation impact loads were calculated respectively, and the calculation formula is expressed as: Among them, r is the Pearson correlation coefficient, n is the number of samples in the data set, and X i is the parameter value of the i-th sample, Y i is the perforation impact load value of the i-th sample, is the sample mean of the parameter values, is the sample average of the perforation impact load values.

4. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 1, characterized in that: The principal component data set is obtained by screening according to the principal-secondary relationship, including: A heat map is drawn according to the calculation result of the Pearson correlation coefficient, in which different colors are used to distinguish the linear positive correlation and the linear negative correlation parameters, and different color depths are used to distinguish the strength of the correlation; The linear influence of different parameters on the perforation impact load is distinguished according to the color and the depth of the color.

5. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 1 is characterized in that: The principal component data set is normalized using a deviation normalization method, including: The formula for the normalization process is: Among them, x is the current principal component value, x min is the minimum value of the parameter category to which the principal component belongs, x max is the maximum value of the parameter category to which the principal component belongs.

6. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 1, characterized in that: Construct a perforation impact load analysis model based on neural network, including: Based on BP neural network, an initial perforation impact load analysis model with input layer, hidden layer and output layer is constructed; The weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm; wherein the fitness function of each particle is the loss function of the initial perforation impact load analysis model; Forward propagation and back propagation of the model weights and biases output by the particle swarm optimization algorithm are performed, a loss function is calculated, and model parameters are updated iteratively through gradient descent; When the number of iterations reaches the preset total number of iterations, a fully trained perforation impact load analysis model is obtained.

7. The artificial intelligence-based oil and gas well perforation impact load prediction method according to claim 6 is characterized in that: The weight and bias of the initial perforation impact load analysis model are optimized by a particle swarm optimization algorithm, including: According to the fitness function value of the best particle in each round of iteration, a fitness curve reflecting the convergence status in the particle swarm optimization process is drawn.

8. An oil and gas well perforation impact load prediction system based on artificial intelligence, characterized in that: include: A data set building module, used for building a perforation impact load data set based on a finite element simulation model considering multiple parameters; An analysis module, used for performing feature analysis on the perforation impact load data set according to the Pearson correlation coefficient, obtaining the primary and secondary relationships of different parameters affecting the perforation impact load, and screening to obtain a principal component data set according to the primary and secondary relationships; A normalization module, used for normalizing the principal component data set by using a deviation normalization method to eliminate scale differences between different features; A data partitioning module is used to divide the normalized principal component data set into a training set, a validation set, and a test set; A model building module is used to build a perforation impact load analysis model based on a neural network, train the model using the training set, optimize the number of neurons, connection weights and neuron biases in the hidden layer of the model using the validation set, and test the model effect using the test set to obtain a fully trained perforation impact load analysis model; The prediction module is used to obtain the influencing parameters of the oil and gas wells and input the influencing parameters into the impact load analysis model to predict the perforation impact load.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the artificial intelligence-based oil and gas well perforation impact load prediction method as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting oil and gas well perforation impact load based on artificial intelligence as described in any one of claims 1 to 7 is implemented.

Images