An intelligent control method and device for wellbore pressure based on physical constraints

Abstract

The present invention provides an intelligent control method and device for wellbore pressure based on physical constraints. The method includes: obtaining the managed pressure drilling data during drilling; constructing an input vector according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening adjustment model, where the throttle valve opening adjustment model is trained using constraint conditions, a neural network, and the historical managed pressure drilling data of the completed well, and the constraint conditions are established from the correlation relationship between constraint parameters and the throttle valve opening; inputting the input vector into the throttle valve opening adjustment model to obtain a predicted throttle valve opening; inputting the predicted throttle valve opening into the bottomhole pressure calculation model to obtain a predicted calculated bottomhole pressure; determining whether the difference between the predicted calculated bottomhole pressure and the target bottomhole pressure is within a preset range, if not, updating the input vector and returning to execute the prediction process, if so, adjusting the throttle valve according to the predicted throttle valve opening. The present invention can effectively improve the stability and generalization ability of the throttle valve opening adjustment model.

Description

Technical Field

[0001] This document relates to the technical field of oil and gas well drilling, and particularly to an intelligent control method and device for wellbore pressure based on physical constraints. Background Art

[0002] With the continuous deepening of oil and gas exploration and development, the focus of oil and gas exploration and development has shifted to deep and ultra-deep formations. However, deep formations often have the characteristics of high temperature, high formation pressure, and narrow safety drilling fluid density window. If the drilling process measures are improper, complex downhole conditions such as overflow and gas invasion are likely to occur. Therefore, accurately controlling the wellbore pressure and making it slightly higher than the formation pressure is the key technology for safe drilling in deep high-temperature and high-pressure formations.

[0003] Traditional wellbore pressure control is realized based on algorithms such as PID controllers. This wellbore pressure control method has problems of low control accuracy, poor stability, and low reliability under complex formation conditions. Moreover, due to its relatively fixed control parameters and inability to adaptively adjust, its robustness is not strong, and overshoot phenomena are likely to occur, which greatly increases the risk of well loss.

[0004] In recent years, the advantages of artificial intelligence technology in the field of oil and gas exploration and development have gradually emerged and have been widely applied to working condition diagnosis, parameter optimization, etc., and good results have been achieved. However, due to the lack of constraints of mechanism models and traditional empirical knowledge, there are still deficiencies in its stability and reliability. Summary of the Invention

[0005] This document is used to solve the problem that when existing artificial intelligence is applied to wellbore pressure control, the constraint parameters related to the throttle valve opening in the pressure control drilling data are not analyzed, and the constraint conditions are not established, so there are problems of poor stability and reliability in wellbore pressure control.

[0006] To solve the above technical problems, the first aspect of this document provides an intelligent control method for wellbore pressure based on physical constraints, including:

[0007] Obtain the pressure control drilling data during drilling;

[0008] Construct an input vector according to the pressure control drilling data during drilling and the input parameters of the throttle valve opening adjustment model, where the input parameters include the constraint parameters and non-constraint parameters of the throttle valve opening, the non-constraint parameters at least include the target bottom hole pressure and the calculated bottom hole pressure, and the throttle valve opening adjustment model is trained using constraint conditions, neural networks, and historical pressure control drilling data of the completed reference well, and the constraint conditions are established by the correlation relationship between the constraint parameters and the throttle valve opening;

[0009] Execute the following prediction process: input the input vector into the throttle valve opening adjustment model to obtain the predicted throttle valve opening; input the predicted throttle valve opening into the bottom hole pressure calculation model to obtain the predicted calculated bottom hole pressure;

[0010] Judge whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within the preset range. If not, update the input vector and continue to execute the prediction process. If so, adjust the throttle valve according to the predicted throttle valve opening.

[0011] As a further embodiment of this article, the determination process of the input parameters of the throttle valve opening adjustment model includes:

[0012] Obtain the historical pressure control drilling data and throttle valve opening of the completed reference well;

[0013] Process the abnormal values and missing values of the historical pressure control drilling data and throttle valve opening;

[0014] According to the processed pressure control drilling data and throttle valve opening, calculate the correlation between each parameter in the pressure control drilling data and the throttle valve opening;

[0015] Select the parameters with a correlation greater than the predetermined value;

[0016] Analyze the variation law of the selected parameters and the throttle valve opening;

[0017] Take the parameters whose variation law is to follow the change of the bottom hole pressure as the constraint parameters, and take the parameters whose variation law is non-follow-up change as the non-constraint parameters.

[0018] As a further embodiment of this article, the training process of the throttle valve opening adjustment model includes:

[0019] According to the historical pressure control drilling data of the completed reference well, construct multiple groups of sample data, and each group of sample data includes the input parameter values and the true values of the throttle valve opening after corresponding regulation;

[0020] Take the constraint parameters and non-constraint parameters of the throttle valve opening as the input, and take the throttle valve opening after regulation as the output to establish a neural network model;

[0021] According to the sample data and the neural network model, construct an error loss function;

[0022] According to the correlation relationship between the constraint parameters of the throttle valve opening and the throttle valve opening, establish a constraint condition;

[0023] According to the error loss function and the constraint condition, solve the parameters in the neural network model.

[0024] In a further embodiment of this article, establishing a constraint condition according to the correlation between the constraint parameter of the throttle valve opening and the throttle valve opening includes:

[0025] Based on the multiphase flow mechanism in the wellbore annulus and / or the variation law of the constraint parameter of the throttle valve opening with the throttle valve opening, determine the correlation between the constraint parameter of each throttle valve opening and the throttle valve opening;

[0026] Using the throttle valve opening expression of the neural network model, convert the correlation between the constraint parameter of each throttle valve opening and the throttle valve opening into an inequality constraint condition.

[0027] In a further embodiment of this article, the constraint parameters of the throttle valve opening include: bottom hole pressure error, drilling fluid flow rate, and drilling fluid density;

[0028] The unconstrained parameters of the throttle valve opening further include: vertical depth and drilling fluid viscosity.

[0029] In a further embodiment of this article, using the mathematical expression of the throttle valve opening of the neural network model to convert the correlation between the constraint parameter of each throttle valve opening and the throttle valve opening into an inequality constraint condition includes:

[0030] Establish an inequality constraint condition using the following formula:

[0031]

[0032]

[0033]

[0034]

[0035] Wherein, A i , B i , C i , D i are the constraint conditions under each sample data, u = F(W1,...,W j ,...,W m ,b1,...,b j ,...,b m ) is the throttle valve opening expression output by the neural network model, ρ is the drilling fluid density, m is the number of layers of the neural network model, W1,...,W j ,...,W m are the weights of each layer in each neural network model, b1,...,b j ,...,b m are the biases of each layer in each neural network model, q is the drilling fluid flow rate, and e is the bottom hole pressure error.

[0036] In a further embodiment of this article, solving the parameters in the neural network model according to the error loss function and the constraint conditions includes:

[0037] Converting the system of equations composed of the error loss function and the constraint conditions into an unconstrained equation;

[0038] Using an intelligent optimization algorithm to solve the unconstrained equation to determine the parameters in the neural network model.

[0039] The second aspect of this article provides an intelligent wellbore pressure control device based on physical constraints, including:

[0040] An acquisition unit for acquiring the managed pressure drilling data during drilling;

[0041] A processing unit for constructing an input vector according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening adjustment model, where the input parameters include the constraint parameters and unconstrained parameters of the throttle valve opening, and the unconstrained parameters at least include the target bottom hole pressure and the calculated bottom hole pressure. The throttle valve opening adjustment model is trained using the constraint conditions, neural network, and historical managed pressure drilling data of the completed well. The constraint conditions are established by the correlation relationship between the constraint parameters and the throttle valve opening;

[0042] A prediction unit for performing the following prediction process: inputting the input vector into the throttle valve opening adjustment model to obtain a predicted throttle valve opening; inputting the predicted throttle valve opening into the bottom hole pressure calculation model to obtain a predicted calculated bottom hole pressure;

[0043] An execution unit for determining whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within a preset range. If not, updating the input vector and having the prediction unit continue to execute the prediction process. If so, adjusting the throttle valve according to the predicted throttle valve opening.

[0044] The third aspect of this article provides a computer device, including a memory, a processor, and a computer program stored on the memory. When the computer program is run by the processor, it executes the instructions of the method according to any one of the foregoing embodiments.

[0045] The fourth aspect of this article provides a computer storage medium, on which a computer program is stored. When the computer program is run by the processor of a computer device, it executes the instructions of the method according to any one of the foregoing embodiments.

[0046] The intelligent control method and device for wellbore pressure based on physical constraints provided in this paper analyze the historical pressure control drilling data of the completed reference well to determine the input parameters related to the throttle valve opening, which can reduce the dimension of the input data, reduce the input of irrelevant data, and improve the training efficiency of the throttle valve opening adjustment model. By dividing the input parameters into constraint parameters and non-constraint parameters of the throttle valve opening, constructing constraint conditions using the constraint parameters, and training the throttle valve opening adjustment model using the constraint conditions, neural network, and historical pressure control drilling data of the completed reference well, the throttle valve opening adjustment model has a certain interpretability, effectively improving the stability and generalization ability of the throttle valve opening adjustment model. Applying the throttle valve opening adjustment model to the wellbore pressure test during drilling, an input vector is constructed according to the pressure control drilling data during drilling and the input parameters of the throttle valve opening adjustment model; the throttle valve opening prediction and bottom hole pressure prediction processes are executed; it is judged whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within the preset range. If not, the input vector is updated and the prediction process is returned. If so, the throttle valve is adjusted using the predicted throttle valve opening, and the throttle valve opening can be accurately predicted using the throttle valve opening adjustment model established by the completed reference well, thereby realizing the efficient and accurate control of the wellbore pressure under complex formation conditions, which has important guiding significance for reducing drilling risks and achieving safe drilling.

[0047] To make the above and other purposes, features, and advantages of this paper more obvious and understandable, the following specific preferred embodiments are given in conjunction with the accompanying drawings and described in detail as follows. Brief Description of the Drawings

[0048] To more clearly illustrate the technical solutions in the embodiments of this paper or in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of this paper. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 Shows the structural diagram of the intelligent control system for wellbore pressure based on physical constraints in the embodiments of this paper;

[0050] Figure 2 Shows the flowchart of the method for determining the throttle valve opening adjustment model in the embodiments of this paper;

[0051] Figure 3 Shows the structural diagram of the neural network model in the embodiments of this paper;

[0052] Figure 4 Shows the flowchart of the parameter solution process of the neural network model in the embodiments of this paper;

[0053] Figure 5Shows the process flow chart for determining the input parameters of the throttle valve opening adjustment model in the embodiments of this article;

[0054] Figure 6 Shows the process flow chart of the intelligent wellbore pressure control method based on physical constraints in the embodiments of this article;

[0055] Figure 7 Shows the process flow chart of the intelligent wellbore pressure control device based on physical constraints in the embodiments of this article;

[0056] Figure 8 Shows the structural diagram of the computer device in the embodiments of this article.

[0057] Explanation of the reference symbols in the drawings:

[0058] 100, the first computing device;

[0059] 110, the client;

[0060] 120, the database;

[0061] 130, the second computing device;

[0062] 310, the input layer;

[0063] 320, the first hidden layer;

[0064] 330, the second hidden layer;

[0065] 340, the output layer;

[0066] 710, the acquisition unit;

[0067] 720, the processing unit;

[0068] 730, the prediction unit;

[0069] 740, the execution unit;

[0070] 802, the computer device;

[0071] 804, the processor;

[0072] 806, the memory;

[0073] 808, the drive mechanism;

[0074] 810, the input / output module;

[0075] 812, the input device;

[0076] 814, the output device;

[0077] 816, the presentation device;

[0078] 818, the graphical user interface;

[0079] 820, Network interface;

[0080] 822, Communication link;

[0081] 824, Communication bus. Detailed implementation manners

[0082] Next, the technical solutions in the embodiments of the present disclosure will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present disclosure.

[0083] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment including a series of steps or units does not necessarily need to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or equipment.

[0084] This specification provides method operation steps as described in the embodiments or flowcharts, but based on routine or non-creative labor, it may include more or fewer operation steps. The step sequences listed in the embodiments are only one way among the execution sequences of numerous steps, and do not represent the only execution sequence. When actually executed in a system or device product, it can be executed in the order of the embodiments or the method shown in the drawings, or executed in parallel.

[0085] It should be noted that the pressure control drilling data involved in the present disclosure are all information and data authorized by the user or fully authorized by all parties.

[0086] In an embodiment of the present disclosure, an intelligent wellbore pressure control system based on physical constraints is provided, as Figure 1 shown, the intelligent wellbore pressure control system based on physical constraints includes: a first computing device 100, a client 110, a database 120, and a second computing device 130.

[0087] The first computing device 100 is used to analyze the historical pressure control drilling data of the completed reference well to determine the input parameters related to the throttle valve opening. Among them, the input parameters include the constraint parameters and non-constraint parameters of the throttle valve opening. The non-constraint parameters at least include the target bottom hole pressure and the calculated bottom hole pressure. According to the historical pressure control drilling data of the completed reference well, multiple groups of sample data are constructed. Each group of sample data includes the input parameter values and the true values of the throttle valve opening after corresponding regulation. Taking the constraint parameters and non-constraint parameters of the throttle valve opening as inputs and the throttle valve opening after regulation as outputs, a neural network model is established. The structure of the neural network model is adjusted using the grid search algorithm or the random search algorithm. An error loss function is constructed based on the sample data and the neural network model. A constraint condition is established according to the correlation between the constraint parameters of the throttle valve opening and the throttle valve opening. According to the error loss function and the constraint condition, the parameters in the neural network model are solved. The neural network model with fixed parameter values is the throttle valve opening adjustment model.

[0088] Among them, the constraint parameters of the throttle valve opening refer to the parameters that have a certain response law with the throttle valve opening. For example, they include the bottom hole pressure error (target bottom hole pressure - calculated bottom hole pressure), drilling fluid flow rate, and drilling fluid density. The non-constraint parameters of the throttle valve opening refer to the parameters that are correlated with the throttle valve opening, but the correlation between them and the throttle valve opening is irregular. For example, they include the target bottom hole pressure, calculated bottom hole pressure, vertical depth, and drilling fluid viscosity.

[0089] The correlation between the constraint parameters and the throttle valve opening is a following relationship. Among them, the following relationship includes a direct proportional relationship and an inverse proportional relationship. The direct proportional relationship means that when the constraint parameter increases, the throttle valve opening also increases, and when the constraint parameter decreases, the throttle valve opening also decreases. The inverse proportional relationship means that when the constraint parameter decreases, the throttle valve opening increases instead, and when the constraint parameter increases, the throttle valve opening decreases instead. Specifically, in implementation, the correlation between the constraint parameters and the throttle valve opening can be represented by taking the partial derivative of the throttle valve opening expression with respect to the constraint parameters. If the partial derivative of the throttle valve opening expression with respect to the constraint parameters is greater than 0, it indicates a direct proportional relationship between the constraint parameters and the throttle valve opening. If the partial derivative of the throttle valve opening expression with respect to the constraint parameters is less than 0, it indicates an inverse proportional relationship between the constraint parameters and the throttle valve opening.

[0090] The client 110 is used to obtain the pressure control drilling data during drilling and send it to the second computing device 130.

[0091] Specifically, in implementation, the client 110 can be a dedicated on-site device or a software program installed in an intelligent terminal. The user can input the pressure control drilling data during drilling in the client 110, or the client can automatically obtain the pressure control drilling data during drilling from the database storing the pressure control drilling data. This article does not limit the method for the client 110 to obtain the pressure control drilling data during drilling.

[0092] The controlled pressure drilling data includes but is not limited to the target bottomhole pressure, the calculated bottomhole pressure, conventional logging data, measurement-while-drilling data, and drilling fluid property parameters, etc. Among them, the target bottomhole pressure is the seismic true pressure, and the calculated bottomhole pressure is the bottomhole pressure calculated after inputting the current choke valve opening into the bottomhole pressure calculation model. The controlled pressure drilling data can be obtained from drilling logging equipment and measurement-while-drilling equipment, as shown in Table 1 specifically. The controlled pressure drilling data shown in Table 1 is only part of the data, and this article does not limit the specific parameters included in the controlled pressure drilling data. In specific implementation, the client 110 can also directly obtain parameters according to the input vector of the choke valve opening regulation model and send them to the second calculation device 130.

[0093] Table 1

[0094]

[0095]

[0096] The choke valve opening regulation model and its parameters, as well as the bottomhole pressure calculation model and its parameters, are stored in the database 120.

[0097] Among them, the structure of the choke valve opening regulation model can adopt the structure of the existing neural network model, and this article does not limit its specific network structure.

[0098] The bottomhole pressure calculation model is an annulus multiphase flow mechanism model. For specific reference to the prior art, it will not be elaborated in this article.

[0099] The second calculation device 130 receives the controlled pressure drilling data during drilling sent by the client 110; constructs an input vector according to the controlled pressure drilling data during drilling and the input parameters of the choke valve opening regulation model; and performs the following prediction process: calls the choke valve opening regulation model from the database 120, inputs the input vector into the choke valve opening regulation model to obtain the predicted choke valve opening; calls the bottomhole pressure calculation model from the database 120, inputs the predicted choke valve opening into the bottomhole pressure calculation model to obtain the predicted calculated bottomhole pressure; determines whether the difference between the predicted calculated bottomhole pressure and the target bottomhole pressure is within the preset range. If not, updates the input vector (that is, replaces the original calculated bottomhole pressure with the current predicted calculated bottomhole pressure, and updates the parameters in the input vector related to the calculated bottomhole pressure, such as the calculated bottomhole pressure and the bottomhole pressure error), and returns to execute the prediction process. If so, adjusts the choke valve using the current predicted choke valve opening.

[0100] In specific implementation, the first calculation device 100 and the second calculation device 130 can be the same device, including but not limited to smart terminals, tablet computers, desktop computers, servers, etc.

[0101] The client 110 can be a desktop computer, a tablet computer, a laptop computer, a smart phone, a digital assistant, a smart wearable device, etc. Among them, the smart wearable device can include a smart bracelet, a smart watch, smart glasses, a smart helmet, etc. Of course, the client is not limited to the above-mentioned electronic devices with a certain entity, and it can also be software running on the above-mentioned electronic devices.

[0102] The database 120 can include a Random Access Memory (RAM), and can also include a non-volatile memory, such as at least one disk memory. Optionally, the database 120 can also be at least one storage device located far from the aforementioned computing device.

[0103] In this embodiment, by analyzing the historical managed pressure drilling data of the completed reference well, the input parameters related to the throttle valve opening are determined, which can reduce the dimension of the input data, reduce the input of irrelevant data, and improve the training efficiency of the throttle valve opening adjustment model. By dividing the input parameters into the constraint parameters and non-constraint parameters of the throttle valve opening, the constraint conditions are constructed using the constraint parameters, and the throttle valve opening adjustment model is trained using the constraint conditions, neural network, and the historical managed pressure drilling data of the completed reference well, so that the throttle valve opening adjustment model has a certain interpretability, effectively improving the stability and generalization ability of the throttle valve opening adjustment model. Applying the throttle valve opening adjustment model to the wellbore pressure test during drilling, an input vector is constructed according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening adjustment model; the throttle valve opening prediction and bottom hole pressure prediction processes are executed; it is judged whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within a preset range. If not, the input vector is updated and the prediction process is returned. If so, the throttle valve is adjusted using the predicted throttle valve opening, which can accurately predict the throttle valve opening using the throttle valve opening adjustment model established by the completed reference well, and then realize the efficient and accurate control of the wellbore pressure under complex formation conditions, which has important guiding significance for reducing drilling risks and achieving safe drilling.

[0104] In one embodiment of this article, a method for determining a throttle valve opening adjustment model is provided, as Figure 2 shown, including:

[0105] Step 210, analyze the historical managed pressure drilling data at multiple sampling moments of the completed reference well, and determine the input parameters of the throttle valve opening adjustment model.

[0106] Among them, the input parameters include constraint parameters and unconstrained parameters related to the throttle valve opening. The unconstrained parameters at least include the target bottom-hole pressure and the calculated bottom-hole pressure. In a specific embodiment, the constraint parameters of the throttle valve opening include the bottom-hole pressure error, the drilling fluid flow rate, and the drilling fluid density. The unconstrained parameters of the throttle valve opening also include the vertical depth and the drilling fluid viscosity.

[0107] Step 220: Construct multiple groups of sample data according to the historical pressure control drilling data of the completed reference well. Each group of sample data includes the input parameter values and the true values of the throttled valve opening after regulation.

[0108] Among them, the completed reference well is a completed well with similar geological conditions to the well to be analyzed. The historical pressure control drilling data is referred to Table 1 shown above. The true value of the throttled valve opening after regulation in each group of sample data is the throttle valve opening corresponding to the calculated bottom-hole pressure in the input parameters being regulated to the target bottom-hole pressure, which can be determined according to historical regulations.

[0109] Step 230: Use the constraint parameters and unconstrained parameters of the throttle valve opening as the input, and the throttled valve opening after regulation as the output to establish a neural network model.

[0110] Among them, the neural network model can select an existing CNN neural network model. Among them, the input of the input layer includes the constraint parameters and unconstrained parameters of the throttle valve opening, and the number of hidden layers can be selected according to requirements. Specifically, when implementing, the structure of the neural network model can be adjusted using the grid search algorithm or the random search algorithm. The specific implementation processes of the grid search algorithm and the random search algorithm can refer to the existing technology and will not be elaborated here.

[0111] In one embodiment, as Figure 3 shown, the established neural network model includes an input layer 310, a first hidden layer 320, a second hidden layer 330, and an output layer 340. The activation functions of the nodes in each layer can be the same or different. Assuming that the activation functions of the nodes in each layer are the same and all are f, the corresponding mathematical expression of the neural network model is:

[0112] u pred = f(W3 × f(W2 × f(W1 × X + b1) + b2) + b3);

[0113] Among them, u pred is the predicted throttle valve opening; f is the activation function; W1, W2, and W3 are the weights between each layer; b1, b2, and b3 are the biases.

[0114] Step 240: Construct an error loss function according to the sample data and the neural network model, as shown in the following formula:

[0115]

[0116] Among them, MAE is the mean absolute error; n is the amount of data; u i,pred is the predicted opening of the i-th throttle valve; u i,true is the actual opening of the i-th throttle valve.

[0117] Step 250: Establish constraint conditions based on the correlation between the constraint parameters of the throttle valve opening and the throttle valve opening.

[0118] When implementing this step, first determine the correlation between the constraint parameters of each throttle valve opening and the throttle valve opening based on the mechanism of multiphase flow in the wellbore annulus and / or the variation law of the constraint parameters of the throttle valve opening and the throttle valve opening; then, use the mathematical expression of the throttle valve opening of the neural network model to convert the correlation between the constraint parameters of each throttle valve opening and the throttle valve opening into inequality constraint conditions.

[0119] Among them, the mechanism of multiphase flow in the wellbore annulus can refer to the prior art and will not be elaborated here. The variation law of the constraint parameters of the throttle valve opening and the throttle valve opening can be determined by fitting the curve of the constraint parameters of the historical throttle valve opening and the throttle valve opening of the completed reference well.

[0120] Taking the constraint parameters including bottom hole pressure error, drilling fluid flow rate, and drilling fluid density as an example, the constraint conditions established in step 250 include the constraint conditions expressed by the following formulas:

[0121]

[0122]

[0123]

[0124]

[0125] Among them, A i , B i , C i , D i are the constraint conditions for each sample data. A i means that when the bottom hole pressure is constant, the opening of the throttle valve increases as the drilling fluid density ρ increases. B i means that when the bottom hole pressure is constant, the opening of the throttle valve increases as the drilling fluid flow rate q increases. C i means that the opening of the throttle valve increases as the absolute value of the bottom hole pressure error e increases and e is negative. D i means that the opening of the throttle valve decreases as the absolute value of the bottom hole pressure error e increases and e is positive.

[0126] u = F(W1,..., W j ,..., W m , b1,..., bj ,...,b m ) is the expression of the throttle valve opening degree output by the neural network model, m is the number of layers of the neural network model, W1,...,W j ,...,W m are the weights of each layer in each neural network model, b1,...,b j ,...,b m are the biases of each layer in each neural network model, and e is the bottom hole pressure error.

[0127] Step 260, solve the parameters in the neural network model according to the error loss function and the constraint conditions.

[0128] Specifically, as Figure 4 shown, the implementation process of this step includes:

[0129] Step 410, convert the system of equations composed of the error loss function and the constraint conditions into an unconstrained equation. Specifically, use the penalty function interior point method to convert the system of equations composed of the error loss function and the constraint conditions into an unconstrained equation expressed by the following formula:

[0130]

[0131] Among them, W, b, and λ are the weight matrix, bias matrix, and constraint term weight matrix in the neural network model respectively, n is the sample data volume, u i,pred is the predicted throttle valve opening degree value, u i,true is the true value of the throttle valve opening degree, λ1, λ2, λ3, and λ4 are the weights in the constraint term weight matrix, A i , B i , C i and D i are the constraint terms under the i-th sample data, and the calculation formulas of each constraint term refer to the foregoing embodiments.

[0132] Step 420, use an intelligent optimization algorithm to solve the unconstrained equation to determine the parameters in the neural network model. Among them, the parameters in the neural network model include weights and biases, that is, W, b. In implementation, an intelligent optimization algorithm such as a genetic algorithm can be selected to determine the parameters in the neural network model.

[0133] This embodiment converts the constrained problem into an unconstrained problem, which can reduce the complexity of the solution and improve the solution efficiency.

[0134] In one embodiment of this article, as Figure 5 shown, after analyzing the historical pressure control drilling data at multiple sampling moments of the reference well in the above step 210, the input parameters determined for the throttle valve opening degree adjustment model include:

[0135] Step 510: Obtain the historical managed pressure drilling data and choke valve opening of the completed drilling reference well.

[0136] Among them, the managed pressure drilling data at each moment corresponds to a choke valve opening, that is, the choke valve opening at that moment.

[0137] Step 520: Process the abnormal values and missing values in the historical managed pressure drilling data and choke valve opening.

[0138] Among them, the processing of abnormal values is, for example, to analyze whether there are abnormal values in each parameter of the managed pressure drilling data. If there are, the corresponding values are deleted or limited within a predetermined range. The processing of missing values is, for example, to fill in the missing parameter values.

[0139] Step 530: Calculate the correlation between each parameter in the managed pressure drilling data and the choke valve opening according to the processed managed pressure drilling data and choke valve opening.

[0140] During implementation, the Pearson correlation coefficient shown in the following formula can be used to calculate the correlation between each parameter in the managed pressure drilling data and the choke valve opening:

[0141]

[0142] Among them, ρ XY is the Pearson correlation coefficient; Cov(X,Y) is the covariance between each parameter and the output; D(X) and D(Y) are the variances of the input parameter and the output of the choke valve opening adjustment model respectively.

[0143] Step 540: Screen out the parameters with a correlation greater than a predetermined value.

[0144] Among them, the predetermined value can be selected according to the actual situation. For example, it is 0.5, and the specific value is not limited in this article.

[0145] Step 550: Analyze the variation law between the screened parameters and the choke valve opening.

[0146] During implementation, the screened parameters can be used as the abscissa and the choke valve opening as the ordinate to draw a curve, and then the variation law between the screened parameters and the choke valve opening can be determined.

[0147] Step 560: Use the parameters whose variation law is to follow the change of the bottom hole pressure as the constraint parameters, and use the parameters whose variation law is non-following as the non-constraint parameters.

[0148] In the training stage and application stage of the choke valve opening adjustment model, its input parameters are all standardized and normalized data. During specific implementation, it can be realized according to the following standardization formula and normalization formula:

[0149] Standardization formula:

[0150]

[0151] Normalization formula:

[0152]

[0153] In the formula, X standardization is the standardized feature; X i is the original feature; μ is the mean of the original features; σ is the standard deviation of the original features; X normalization is the normalized feature; X min is the minimum value of the original feature; X max is the maximum value of the original feature.

[0154] This embodiment can reduce the dimension of the input parameters, reduce the input of useless and irrelevant data, and improve the training efficiency of the throttle valve opening adjustment model. First, correlation analysis is performed to optimize the input parameters. To reduce the influence of the dimension unit and order of magnitude on the training effect, the parameters are made dimensionless by standardization or normalization.

[0155] Based on the establishment of the throttle valve opening adjustment model, it can be used for intelligent control of wellbore pressure based on physical constraints. In one embodiment of this article, an intelligent control method for wellbore pressure based on physical constraints is also provided, as Figure 6 shown, including:

[0156] Step 610, obtain the managed pressure drilling data during drilling.

[0157] Among them, the well under drilling is the well being dug within a certain range from the reference well after completion, or the well being dug with similar geological conditions to the reference well after completion.

[0158] Step 620, construct an input vector according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening adjustment model.

[0159] Among them, the arrangement order of the input vector is the same as the order of the input parameters during the training of the throttle valve opening adjustment model.

[0160] Step 630, perform the following prediction process: input the input vector into the throttle valve opening adjustment model to obtain the predicted throttle valve opening; input the predicted throttle valve opening into the bottom hole pressure calculation model to obtain the predicted calculated bottom hole pressure.

[0161] Step 640, determine whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within the preset range. If not, execute step 650; if so, execute step 660.

[0162] Step 650, update the input vector and return to execute the prediction process.

[0163] Step 660, adjust the throttle valve by using the finally obtained predicted throttle valve opening degree.

[0164] Among them, the preset range can be determined according to the control accuracy, and the specific value thereof is not limited in this article.

[0165] Taking the input parameters including bottom hole pressure error, calculated bottom hole pressure, target bottom hole pressure, drilling fluid flow rate, drilling fluid density, vertical depth and drilling fluid viscosity as an example, to update the input vector is to recalculate the bottom hole pressure error according to the predicted calculated bottom hole pressure and the target bottom hole pressure, and replace the original bottom hole pressure error with the recalculated bottom hole pressure error, and use the predicted calculated bottom hole pressure to replace the original bottom hole pressure error.

[0166] This embodiment combines the managed pressure drilling data, the managed pressure drilling experience knowledge (converted into constraint conditions) and the artificial intelligence algorithm, and can realize the adaptive and precise control of the wellbore pressure, thereby avoiding the occurrence of drilling risks, reducing the non-production time and reducing the drilling cost.

[0167] Based on the same inventive concept, this article also provides an intelligent control device for wellbore pressure based on physical constraints, as described in the following embodiments. Since the principle of solving problems by the intelligent control device for wellbore pressure based on physical constraints is similar to that of the intelligent control method for wellbore pressure based on physical constraints, therefore, the implementation of the intelligent control device for wellbore pressure based on physical constraints can refer to the intelligent control method for wellbore pressure based on physical constraints, and the repeated parts will not be described again. Specifically, as Figure 7 shown, the intelligent control device for wellbore pressure based on physical constraints includes:

[0168] An acquisition unit 710, configured to acquire the managed pressure drilling data during drilling;

[0169] A processing unit 720, configured to construct an input vector according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening degree adjustment model, where the input parameters include the constraint parameters and non-constraint parameters of the throttle valve opening degree, and the non-constraint parameters at least include the target bottom hole pressure and the calculated bottom hole pressure, and the throttle valve opening degree adjustment model is trained by using constraint conditions, a neural network and the historical managed pressure drilling data of the completed reference well, and the constraint conditions are established by the correlation relationship between the constraint parameters and the throttle valve opening degree;

[0170] A prediction unit 730, configured to perform the following prediction process: input the input vector into the throttle valve opening degree adjustment model to obtain a predicted throttle valve opening degree; input the predicted throttle valve opening degree into the bottom hole pressure calculation model to obtain a predicted calculated bottom hole pressure;

[0171] An execution unit 740 is configured to determine whether the difference between the predicted bottomhole pressure and the target bottomhole pressure is within a preset range. If not, the input vector is updated, and the prediction process is returned for execution. If so, the throttle valve is adjusted according to the finally obtained predicted throttle valve opening degree.

[0172] In this embodiment, by analyzing the historical pressure control drilling data of the completed reference well, the input parameters related to the throttle valve opening degree are determined, which can reduce the dimension of the input data, reduce the input of irrelevant data, and improve the training efficiency of the throttle valve opening degree adjustment model. By dividing the input parameters into constraint parameters and non-constraint parameters of the throttle valve opening degree, constraint conditions are constructed using the constraint parameters, and the throttle valve opening degree adjustment model is trained using the constraint conditions, neural network, and historical pressure control drilling data of the completed reference well, so that the throttle valve opening degree adjustment model has a certain interpretability, effectively improving the stability and generalization ability of the throttle valve opening degree adjustment model. Applying the throttle valve opening degree adjustment model to the wellbore pressure test during drilling, an input vector is constructed according to the pressure control drilling data during drilling and the input parameters of the throttle valve opening degree adjustment model; the throttle valve opening degree prediction and bottomhole pressure prediction processes are executed; it is determined whether the difference between the predicted calculated bottomhole pressure and the target bottomhole pressure is within a preset range. If not, the input vector is updated, and the prediction process is returned for execution. If so, the throttle valve is adjusted using the predicted throttle valve opening degree, which can accurately predict the throttle valve opening degree using the throttle valve opening degree adjustment model established by the completed reference well, and then realize the efficient and accurate control of the wellbore pressure under complex formation conditions, which has important guiding significance for reducing drilling risks and achieving safe drilling.

[0173] In one embodiment of this article, a computer device is further provided, such as Figure 8 As shown, the computer device 802 may include one or more processors 804, such as one or more central processing units (CPUs), and each processing unit may implement one or more hardware threads. The computer device 802 may also include any memory 806, which is used to store the code of the method described in any of the foregoing embodiments and related settings, data, and other such information. Non-limiting, for example, the memory 806 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory devices, hard disks, optical discs, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent a fixed or removable component of the computer device 802. In one case, when the processor 804 executes the associated instructions stored in any memory or combination of memories, the computer device 802 may perform any operation of the associated instructions. The computer device 802 also includes one or more drive mechanisms 808 for interacting with any memory, such as a hard disk drive mechanism, an optical disc drive mechanism, etc.

[0174] The computer device 802 may also include an input / output module 810 (I / O) for receiving various inputs (via the input device 812) and for providing various outputs (via the output device 814). A specific output mechanism may include a presentation device 816 and an associated graphical user interface 818 (GUI). In other embodiments, the input / output module 810 (I / O), the input device 812, and the output device 814 may not be included, and it may only be a computer device in the network. The computer device 802 may also include one or more network interfaces 820 for exchanging data with other devices via one or more communication links 822. One or more communication buses 824 couple the components described above together.

[0175] The communication link 822 may be implemented in any way, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 822 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0176] Corresponding to Figure 2 、 Figure 5 - Figure 6 In the method, embodiments of the present invention also provide a computer-readable storage medium on which a computer program is stored, and when the computer program is run by a processor, the steps of the above method are executed.

[0177] Embodiments of the present invention also provide a computer-readable instruction, wherein when the processor executes the instruction, the program therein causes the processor to execute the method as shown in Figure 2 、 Figure 5 - Figure 6 shown.

[0178] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0179] It should also be understood that in the embodiments of the present invention, the term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in the present invention generally represents an "or" relationship between the associated objects before and after.

[0180] Those of ordinary skill in the art will appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of the examples have been generally described according to their functions in the above description. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this article.

[0181] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0182] In the several embodiments provided in this article, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the couplings, direct couplings, or communication connections shown or discussed with each other can be indirect couplings or communication connections through some interfaces, devices, or units, and can also be electrical, mechanical, or other forms of connection.

[0183] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments in this article.

[0184] In addition, the functional units in the various embodiments of this article can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0185] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of this article, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this article. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0186] Specific embodiments are used in this article to elaborate on the principles and implementation manners of this article. The description of the above embodiments is only used to help understand the method and its core idea of this article; at the same time, for those of ordinary skill in the art, according to the idea of this article, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation on this article.

Claims

1. An intelligent control method for wellbore pressure based on physical constraints, characterized in that, Including: Obtaining the managed pressure drilling data during drilling; Constructing an input vector according to the managed pressure drilling data during drilling and the input parameters of the throttle valve opening adjustment model, wherein the input parameters include the constraint parameters and non-constraint parameters of the throttle valve opening, and the non-constraint parameters at least include the target bottom hole pressure, calculated bottom hole pressure, vertical depth, and drilling fluid viscosity; the constraint parameters of the throttle valve opening include: bottom hole pressure error, drilling fluid flow rate, and drilling fluid density; Performing the following prediction process: inputting the input vector into the throttle valve opening adjustment model to obtain a predicted throttle valve opening; inputting the predicted throttle valve opening into the bottom hole pressure calculation model to obtain a predicted calculated bottom hole pressure; Judging whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within a preset range. If not, updating the input vector and continuing to perform the prediction process. If so, adjusting the throttle valve according to the predicted throttle valve opening; Wherein, the bottom hole pressure calculation model is an annulus multiphase flow mechanism model; Wherein, the training process of the throttle valve opening adjustment model includes: Constructing multiple groups of sample data according to the historical managed pressure drilling data of the completed reference well, and each group of sample data includes input parameter values and the true value of the throttle valve opening after corresponding regulation; Taking the constraint parameters and non-constraint parameters of the throttle valve opening as inputs and the throttle valve opening after regulation as outputs to establish a neural network model; Constructing an error loss function according to the sample data and the neural network model; Establishing a constraint condition according to the correlation relationship between the constraint parameters of the throttle valve opening and the throttle valve opening; Solving the parameters in the neural network model according to the error loss function and the constraint condition; Wherein, establishing a constraint condition according to the correlation relationship between the constraint parameters of the throttle valve opening and the throttle valve opening includes: Based on the mechanism of wellbore annulus multiphase flow and / or the variation law of the constraint parameters of the throttle valve opening and the throttle valve opening, determining the correlation relationship between the constraint parameters of each throttle valve opening and the throttle valve opening; Using the mathematical expression of the throttle valve opening of the neural network model to convert the correlation relationship between the constraint parameters of each throttle valve opening and the throttle valve opening into an inequality constraint condition.

2. The method according to claim 1, wherein The determination process of the input parameters of the throttle valve opening adjustment model includes: Obtaining the historical managed pressure drilling data and throttle valve opening of the completed reference well; Performing outlier and missing value processing on the historical managed pressure drilling data and throttle valve opening; Calculating the correlation between each parameter in the managed pressure drilling data and the throttle valve opening according to the processed managed pressure drilling data and throttle valve opening; Selecting the parameters with a correlation greater than a predetermined value; Analyzing the variation law of the selected parameters and the throttle valve opening; Taking the parameters whose variation law is to follow the change of the bottom hole pressure as constraint parameters, and taking the parameters whose variation law is non-follow-up change as non-constraint parameters.

3. The method according to claim 2, characterized in that, Using the data expression of the throttle valve opening of the neural network model to convert the correlation relationship between the constraint parameters of each throttle valve opening and the throttle valve opening into an inequality constraint condition includes: Establishing an inequality constraint condition using the following formula: Among them, A i , B i , C i , D i are the constraint conditions under each sample data. u = F(W1,...,W j ,...,W m ,b1,...,b j ,...,b m ) is the expression of the throttle valve opening degree output by the neural network model. ρ is the density of the drilling fluid, m is the number of layers of the neural network model, W1,...,W j ,...,W m are the weights of each layer in each neural network model, b1,...,b j ,...,b m are the biases of each layer in each neural network model, q is the flow rate of the drilling fluid, and e is the bottom hole pressure error.

4. The method according to claim 2, wherein Solving the parameters in the neural network model according to the error loss function and the constraint condition includes: Convert the system of equations composed of the error loss function and the constraint conditions into an unconstrained equation; Use an intelligent optimization algorithm to solve the unconstrained equation to determine the parameters in the neural network model.

5. An intelligent control device for wellbore pressure based on physical constraints, characterized in that, Including: An acquisition unit for acquiring the managed pressure drilling data during drilling; A processing unit for constructing an input vector according to the managed pressure drilling data during drilling and the input parameters of the choke valve opening adjustment model, where the input parameters include the constraint parameters and unconstraint parameters of the choke valve opening, and the unconstraint parameters at least include the target bottom hole pressure, calculated bottom hole pressure, vertical depth, and drilling fluid viscosity; the constraint parameters of the choke valve opening include: bottom hole pressure error, drilling fluid flow rate, and drilling fluid density; A prediction unit for performing the following prediction process: inputting the input vector into the choke valve opening adjustment model to obtain a predicted choke valve opening; inputting the predicted choke valve opening into the bottom hole pressure calculation model to obtain a predicted calculated bottom hole pressure; An execution unit for determining whether the difference between the predicted calculated bottom hole pressure and the target bottom hole pressure is within a preset range. If not, update the input vector and continue to execute the prediction process by the prediction unit. If so, adjust the choke valve according to the predicted choke valve opening; Wherein, the bottom hole pressure calculation model is an annular multiphase flow mechanism model; Wherein, the training process of the choke valve opening adjustment model includes: Construct multiple groups of sample data according to the historical managed pressure drilling data of the completed reference well, and each group of sample data includes the input parameter values and the true values of the adjusted choke valve opening; Establish a neural network model with the constraint parameters and unconstraint parameters of the choke valve opening as inputs and the adjusted choke valve opening as outputs; Construct an error loss function according to the sample data and the neural network model; Establish constraint conditions according to the correlation between the constraint parameters of the choke valve opening and the choke valve opening; Solve the parameters in the neural network model according to the error loss function and the constraint conditions; Wherein, establishing constraint conditions according to the correlation between the constraint parameters of the choke valve opening and the choke valve opening includes: Based on the mechanism of annular multiphase flow in the wellbore and / or the variation law of the constraint parameters of the choke valve opening and the choke valve opening, determine the correlation between the constraint parameters of each choke valve opening and the choke valve opening; Use the mathematical expression of the choke valve opening of the neural network model to convert the correlation between the constraint parameters of each choke valve opening and the choke valve opening into inequality constraint conditions.

6. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that, When the computer program is run by the processor, it executes the instructions of the method according to any one of claims 1-4.

7. A computer storage medium, on which a computer program is stored, characterized in that, When the computer program is run by the processor of the computer device, it executes the instructions of the method according to any one of claims 1-4.

Images