Hydraulic remote opening and closing method and system for stratum isolation valve

Abstract

The invention provides a hydraulic remote opening and closing method and system for a stratum isolation valve, and the method comprises the steps: obtaining underground sensor data, and building an underground tubular column topological structure model; constructing a graph neural network for the fluid flow data and the pipeline network, and calculating an optimal opening and closing path; real-time pressure compensation calculation is conducted on the optimal opening and closing path, it is ensured that the valve can be dynamically adjusted during execution, and an optimized control signal is obtained; and converting the control signal into an actual hydraulic opening and closing action, monitoring the state of the valve in real time, monitoring the pressure and the flow of each node through real-time feedback, comparing the pressure and the flow with the target pressure and the target flow, and if the comparison value exceeds a preset threshold value, recalculating a new control signal, and finely adjusting the hydraulic valve. The optimal opening and closing path is intelligently calculated, and a real-time pressure compensation mechanism is combined, so that hydraulic control is more accurate and reliable.

Description

Technical Field

[0001] The present invention belongs to the field of formation isolation valves, and particularly relates to a hydraulic remote opening and closing method and system for formation isolation valves. Background Art

[0002] During the development and production of oil and gas wells, a formation isolation valve is a key downhole control tool, mainly used in operations such as cementing, fracturing, acidizing, and fluid drainage. The main function of the formation isolation valve is to open and close the liquid passage downhole through remote control, so as to achieve the isolation of different areas inside the wellbore or the purpose of fluid circulation.

[0003] However, in a complex downhole environment, there are still many technical problems in the existing formation isolation valve opening and closing system, which affect its actual application effect. First of all, the current hydraulic control method is easily affected by complex downhole conditions, resulting in low reliability of opening and closing. For example, after cementing or fracturing operations, there are usually high-viscosity fluids or solid-phase particle depositions in the well, which may lead to a decrease in hydraulic transmission efficiency, resulting in problems such as slip sleeve jamming or opening and closing failure.

[0004] In addition, due to the gradient change of formation pressure in different well sections, it is difficult to adapt to the differences in different formation pressures by using a control method of triggering opening and closing with a fixed pressure, which may cause the slip sleeve to not fully close or seal failure, affecting the smooth progress of subsequent operations. Secondly, there are great limitations in the optimization of the opening and closing sequence of the existing hydraulic control system.

[0005] Generally, the opening and closing sequence of the formation isolation valve is controlled according to a preset pressure value or the fluid flow direction. However, the downhole string structure is relatively complex, and the interaction between different slip sleeves, packers, and hydraulic channels is not fixed, resulting in the hydraulic signal being easily affected by factors such as flow resistance and pressure drop during transmission, thereby reducing the accuracy and execution efficiency of hydraulic control. In some cases, if the opening and closing sequence of the slip sleeve is unreasonable, it may also lead to excessive energy loss during the hydraulic transmission process, and even cause interference between the slip sleeves, making the opening and closing operation unable to proceed smoothly.

[0006] In addition, after the cementing operation is completed, the sealing mechanism of the slip sleeve may be affected by cement backflow, resulting in incomplete sealing or damage to the sealing structure. If the slip sleeve sealing ring cannot close due to cement backflow, it may cause damage to the cement layer in the cementing section, forming a bypass channel for the fluid and affecting the isolation effect. In some high-temperature and high-pressure wells or complex well conditions, the traditional formation isolation valve may also face the problem of reduced sealing performance due to formation pressure fluctuations, thereby affecting the effectiveness of subsequent acidizing, fracturing, or fluid drainage operations.

[0007] In addition, most current remote hydraulic control systems rely on hydraulic signals applied on the ground for operation, but the real-time status information feedback from the downhole is insufficient, making it impossible for ground control personnel to accurately judge the actual opening and closing status of the sliding sleeve. If problems such as sliding sleeve jamming and hydraulic signal attenuation occur downhole, it is difficult for ground personnel to adjust parameters in a timely manner, and it may be necessary to lower additional work strings for fault handling, which not only increases the construction cost but also raises the operation risk.

[0008] Therefore, there are still many technical problems in the existing hydraulic remote opening and closing technology of formation isolation valves in terms of hydraulic opening and closing reliability, optimization of opening and closing sequence, sealing guarantee, and remote feedback mechanism. There is an urgent need for a more intelligent, more adaptable and self-adjustable opening and closing method and system to improve the stability and sealing integrity of formation isolation valves under complex well conditions. Summary of the Invention

[0009] The object of the present invention is to propose a hydraulic remote opening and closing method and system for formation isolation valves. By intelligently calculating the optimal opening and closing path and combining with a real-time pressure compensation mechanism, the hydraulic control is made more accurate and reliable.

[0010] In a first aspect, an embodiment of the present invention provides a hydraulic remote opening and closing method for a formation isolation valve, the method comprising:

[0011] Obtain downhole sensor data and establish a downhole string topology structure model; wherein, the downhole sensor data includes: fluid flow data and pipeline network;

[0012] For the fluid flow data and pipeline network, construct a graph neural network and calculate the optimal opening and closing path; wherein, the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing status of each node and the pressure of each node; wherein, each node represents a valve;

[0013] Perform real-time pressure compensation calculation on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution, and obtain an optimized control signal; the control signal includes pressure, flow distribution and valve status;

[0014] Convert the control signal into actual hydraulic opening and closing actions, and monitor the valve status in real time. By comparing the pressure and flow rate of each node monitored through real-time feedback with the target pressure and flow rate, if the comparison value exceeds the preset threshold, recalculate a new control signal and fine-tune the hydraulic valve.

[0015] Among them, the graph neural network uses the current pressure of each node, the flow rate of each node, and the opening and closing state of the valve as nodes, performs weighted graph convolution modeling on the current pressure of each node and the flow rate of each node, and updates through the adjacency matrix; the adjacency matrix represents the pipeline connection relationship between nodes.

[0016] Furthermore, for the edges of the downhole string topology structure model, the edge weights are calculated based on the length and inner diameter of the hydraulic channel.

[0017] Furthermore, the downhole string topology structure model has a dynamic update mechanism, and the edge weights of the connection edges of the topology graph are updated through the downhole sensor data collected in real time.

[0018] Furthermore, the graph neural network is expressed as:

[0019]

[0020] Among them, represents the feature vector of node at the -th layer; is the set of neighbor nodes of node ; and are the weight matrix and bias term of the -th layer respectively; is a correction term for the node pressure, flow rate and opening and closing state, used to introduce the influence of the fluid state.

[0021] Furthermore, during the path optimization of the fluid state by the graph neural network, corrections based on the fluid state are simultaneously performed; the corrections based on the fluid state include:

[0022] It is calculated by weighted summation of the pressure difference between the current pressure and the normalized pressure, the difference between the current flow rate and the normalized flow rate, and the semi-closed state respectively;

[0023] Among them, after each round of graph convolution update, the node feature vector output by the graph neural network will be used for the optimization of the target pressure; the optimization of the target pressure is based on the objective function, and the objective function avoids unnecessary losses caused by excessive opening and closing by minimizing the pressure difference of each node.

[0024] Furthermore, before performing the real-time pressure compensation calculation for the optimal opening and closing path, the pressure difference between the pressure of the node and the target pressure is obtained; an adaptive pressure compensation term is designed based on the pressure difference, and the adaptive pressure compensation term includes the pressure difference, the opening and closing state, the current flow rate of the node, and the normalized flow rate.

[0025] Further, the real-time pressure compensation for the optimal opening and closing path includes:

[0026] Adjust the flow rate of each node through an adjustment coefficient according to the adaptive pressure compensation term to obtain a new flow rate value;

[0027] According to the pressure difference and the adaptive pressure compensation term, use the Sigmoid function to smoothly adjust the opening and closing state of the valve to obtain a new opening and closing state of the valve.

[0028] Further, the comparison of the pressure and flow rate of each node monitored through real-time feedback with the target pressure and flow rate includes:

[0029] Calculate the pressure deviation and flow rate deviation of each node; the pressure deviation is the difference between the node pressure after compensation adjustment and the target pressure; the flow rate deviation is the difference between the node flow rate value after compensation adjustment and the target flow rate value.

[0030] Further, if the comparison value exceeds the preset threshold, recalculate a new control signal and fine-tune the hydraulic valve, including:

[0031] The new control signal Constrain based on the influence of the flow rate on the control signal, and the calculation is:

[0032]

[0033] Wherein, is the maximum pressure of the system, is the maximum flow rate of the system; , , is the adjustment coefficient, which controls the influence degree of each term; is the regularization term, which is designed as an additional term related to the pressure and flow rate deviation, and is used to balance the feedback effects of pressure and flow rate; is the new opening and closing state of the valve;

[0034] Wherein, the regularization term Adjust its influence degree by introducing the product of pressure and flow rate;

[0035] Wherein, the actual opening and closing degree of the hydraulic valve is output using a smooth non-linear model, and the new control signal is mapped through the Sigmoid function To the opening and closing state.

[0036] In a second aspect, an embodiment of the present invention provides a hydraulic remote opening and closing system for a formation isolation valve, and the system includes:

[0037] A sensor data acquisition unit for acquiring downhole sensor data and establishing a downhole string topology model; wherein, the downhole sensor data includes: fluid flow data and a pipeline network;

[0038] A graph neural network construction unit for constructing a graph neural network for the fluid flow data and the pipeline network and calculating an optimal opening and closing path; wherein, the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing state of each node and the pressure of each node; wherein, each node represents a valve;

[0039] A pressure compensation unit for performing real-time pressure compensation calculation on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution and obtain an optimized control signal; the control signal includes pressure, flow distribution, and valve state;

[0040] A hydraulic valve adjustment unit for converting the control signal into an actual hydraulic opening and closing action, monitoring the valve state in real time, comparing the pressure and flow rate of each node monitored through real-time feedback with the target pressure and flow rate, and if the comparison value exceeds a preset threshold, recalculating a new control signal and finely adjusting the hydraulic valve;

[0041] Wherein, the graph neural network uses the current pressure of each node, the flow rate of each node, and the opening and closing state of the valve as nodes, performs weighted graph convolution modeling on the current pressure of each node and the flow rate of each node, and is updated through an adjacency matrix; the adjacency matrix represents the pipeline connection relationship between nodes.

[0042] The beneficial technical effects of the present invention are at least as follows:

[0043] First of all, the present invention uses a graph neural network to construct a dynamic topology model for downhole strings, sliding sleeves, packers, and hydraulic channels, and calculates the optimal opening and closing sequence based on real-time monitoring data. By establishing an association model between downhole tools and hydraulic control, it can effectively predict the impact of different opening and closing sequences on the overall pressure transmission, thereby optimizing the opening and closing sequence of the sliding sleeve, improving the success rate of opening and closing, and reducing operation errors caused by hydraulic signal attenuation or interference between sliding sleeves.

[0044] Secondly, through the adaptive pressure compensation control system, the present invention monitors the real-time downhole pressure changes, liquid flow rates, and the force on the sealing ring during the opening and closing process of the sliding sleeve, and dynamically adjusts the hydraulic pressure through an adaptive control algorithm, so as to ensure appropriate pressure compensation in different formation pressure environments and prevent problems such as sliding sleeve jamming or seal failure caused by insufficient pressure. In addition, the present invention also adopts remote intelligent monitoring and feedback control technology to obtain key data such as the opening and closing state, pressure distribution, and seal integrity of the downhole sliding sleeve in real time at the ground control end, and dynamically adjusts the opening and closing strategy based on an intelligent control algorithm. When it is detected that the sliding sleeve is not fully closed or the hydraulic signal is abnormal, the system can automatically adjust the pressure compensation parameters or recalculate the opening and closing path to ensure the safety and stability of the opening and closing operation.

[0045] Finally, in order to address the impact of cement backflow on the sliding sleeve sealing structure after cementing, the present invention adopts an intelligent check mechanism combined with pressure compensation sealing design, and optimizes the opening and closing sequence based on a graph neural network, enabling the system to predict the possible cement backflow trend and adjust the sliding sleeve sealing structure in advance to prevent the backflow from impacting the sealing ring.

[0046] Meanwhile, during subsequent operations, the formation pressure changes are adjusted in real time through the adaptive pressure compensation system, enabling the sealing structure to always remain stable and reliable during long-term operation. The present invention comprehensively applies key technologies such as graph neural network optimization of the opening and closing path, adaptive pressure compensation control, remote intelligent monitoring and feedback control, etc., to achieve intelligent hydraulic remote opening and closing of the formation isolation valve in complex downhole environments, which can effectively improve the reliability of the opening and closing system, reduce energy consumption, enhance the sealing performance, and lower the construction cost and operation risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0048] Figure 1 It is a flowchart of the hydraulic remote opening and closing method of the formation isolation valve of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The embodiments of the present invention are described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0050] In one embodiment, as Figure 1As shown in the figure, a hydraulic remote opening and closing method for a formation isolation valve is provided; the method includes the following steps:

[0051] S1. Obtain downhole sensor data and establish a downhole string topology model; wherein, the downhole sensor data includes: fluid flow data and pipeline network.

[0052] Specifically, the downhole sensor data includes information such as pressure, sliding sleeve position, sealing state, pipeline layout, and hydraulic channel flow rate. This data is sourced from downhole physical sensors (such as pressure sensors, temperature sensors, displacement sensors, etc.), has time series characteristics, and is collected in real time through downhole equipment. This data not only records the current state of the equipment but also provides the connection relationships between the equipment and the key parameters of fluid flow. These sensor data form the basis for subsequent opening and closing path optimization and pressure compensation. To effectively convert this information into the data structure for system control, a model reflecting the physical layout and state of the downhole string system needs to be constructed.

[0053] To solve the problem of mutual influence between equipment in downhole string control, we introduced graph theory and constructed a topology model based on a dynamic weighted graph. This topological graph reflects the connection relationships between equipment and the pressure distribution in the hydraulic channels and can be dynamically updated according to the opening and closing states of the equipment.

[0054] Specifically, our topology model has the following characteristics:

[0055] Equipment node modeling: Each piece of equipment in the downhole string (such as sliding sleeves, packers, hydraulic pumps, etc.) is modeled as a node in the graph. Each node not only represents the position and state of the equipment but also contains the operating characteristics of the equipment (such as hydraulic requirements, opening and closing time limits, control signals, etc.). Through the collected sensor data, we can update the state information of these equipment in real time.

[0056] Connection edge modeling: The edges of the graph represent the hydraulic connection relationships between equipment. Different from traditional static connection graphs, our edge weights not only include the physical connectivity between equipment but also incorporate the physical characteristics of the hydraulic channels, including factors such as pressure drop, flow rate, and temperature change. The specific edge weight calculation formula is:

[0057]

[0058] Wherein, is the length of the hydraulic channel between equipment and equipment ; is the inner diameter of the hydraulic channel; is the pressure drop between equipment and equipment ; It is the temperature influence of the hydraulic channel, reflecting the influence of the viscosity change of the hydraulic oil with temperature on the flow. It is the weight coefficient, controlling the influence of various factors on the connection strength.

[0059] The length and inner diameter of the hydraulic channel determine the resistance of fluid flow and affect the pressure drop between devices. The pressure drop and temperature directly affect the energy loss and flow rate change in the hydraulic system, thus affecting the pressure and flow rate required for device opening and closing.

[0060] To adapt to the real-time changes during the opening and closing process of devices in the downhole environment, we designed a dynamic topology graph update mechanism. Whenever the opening and closing state of a device changes, the connection relationship between devices and the corresponding hydraulic channel characteristics will change. Therefore, we need to update the connection edges of the topology graph through the sensor data collected in real time to reflect the current system state.

[0061] The mathematical model of this dynamic update mechanism is as follows:

[0062]

[0063] Among them, represents the current moment device and device the connection strength between them; represents device and device the change in pressure drop between them; is the update rate control coefficient. During the opening and closing process of the device, the pressure drop of the hydraulic channel will change, which in turn affects the connection strength between devices. Through this update formula, we can update the topology structure of the system at each time step to reflect the pressure and flow rate changes between downhole devices in real time.

[0064] The downhole string topology structure diagram obtained through the above modeling method. This diagram not only includes the connection relationship between devices, but also takes into account the physical characteristics of the pipeline (such as length, inner diameter, pressure drop, etc.) and temperature effects to form a comprehensive and dynamically updated topology graph. This topology graph will be updated in real time according to the opening and closing state of the device to ensure that the system can make optimized decisions according to the actual situation. Whenever there is an opening or closing action of a device, the system can quickly respond through this graph, adjust the optimization strategy, and ensure the efficiency and accuracy of the opening and closing operation.

[0065] S2. For the fluid flow data and the pipeline network, construct a graph neural network to calculate the optimal opening and closing path; among them, the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing state of each node and the pressure of each node; among them, each node represents a valve.

[0066] Specifically, the pipeline network topology and fluid state information output in the previous stage will be used as the input for this step. This information includes:

[0067] Pipeline network topology: represented by an adjacency matrix to indicate the pipeline connection relationship between node and node .

[0068] Node information:

[0069] represents the current pressure of each node (valve or pipeline).

[0070] represents the flow rate of each node.

[0071] is the opening and closing state of the valve, , where indicates that the valve is open, indicates that the valve is closed.

[0072] Target pressure: , and this value may come from external requirements, system design requirements, or feedback information from dynamic regulation.

[0073] To calculate the optimal opening and closing path and target pressure, we use a graph neural network (GNN) to model the pipeline network. The advantage of a graph neural network is that it can capture complex dependencies in the network structure through information propagation between nodes and can adaptively adjust the opening and closing states of valves. In each layer of the graph neural network, the feature vectors of nodes (including pressure, flow rate, opening and closing states, etc.) are updated through the adjacency matrix . Our goal is to design a new graph convolutional layer that can not only update the basic information of nodes but also consider the impact of fluid flow on the entire pipeline network.

[0074] The present invention introduces a weighted graph convolutional model based on flow rate and pressure, and the formula is as follows:

[0075]

[0076] Wherein, represents the feature vector of node at the -th layer, including information such as the pressure, flow rate, and opening and closing states of the node. is the set of neighbor nodes of node . and are respectively the The weight matrix and bias term of the layer. is a correction term for node pressure, flow rate, and opening / closing state, used to introduce the influence of fluid state. The role of this correction term is to adjust the calculation of node features, enabling the optimization process to consider the impact of fluid flow on path selection.

[0077] To ensure the role of fluid state in path optimization, we designed the following fluid state correction function :

[0078]

[0079] where and are adjustment coefficients that determine the influence of pressure and flow rate in path optimization. and are the average pressure and flow rate of the entire pipe network, used for normalization. is an adjustment coefficient used to balance the influence of the opening / closing state on the network. The influence range of the opening / closing state is mapped to to realize the contribution of valve opening or closing to the network.

[0080] After each round of graph convolution update, the node feature vector output by the graph neural network will be further used for the optimization of the target pressure. To optimize the opening / closing path and pressure field, we designed an objective function aiming to minimize the pressure difference of each node and avoid unnecessary losses caused by excessive opening / closing.

[0081] Objective function is defined as:

[0082]

[0083] where is the pressure of node after being optimized by the graph neural network. is the target pressure value, which may vary according to external demands and system dynamic regulation. is the opening / closing state of node (binary, or ). is the regularization coefficient, used to control the influence of the opening / closing state on the system pressure optimization.

[0084] This objective function considers two aspects:

[0085] Pressure optimization: We hope that the pressure of each node is as close as possible to the target pressure , thus ensuring the pressure stability of the entire system.

[0086] Optimization of opening and closing path: By adding a penalty term for the opening and closing state, ensure that the selection of the opening and closing path does not change too frequently to avoid system instability.

[0087] Through the above graph convolution update, correction function, objective function, and optimization strategy, we can train the graph neural network to obtain the optimal opening and closing path and target pressure. During the training process, we use the standard gradient descent method (such as the Adam optimizer) to adjust the weights and biases in the network to minimize the objective function , thereby obtaining the optimal opening and closing state of each node and target pressure.

[0088] The final output results are:

[0089] The opening and closing state of each node: that is, the opening and closing decision of each valve , with values or .

[0090] The pressure of each node: the calculated node pressure , whose value is close to the predetermined target pressure .

[0091] In this way, the graph neural network can effectively learn the optimal opening and closing path from the complex pipeline topology, while ensuring that the pressure of the entire system remains within a stable range during operation and reaches the target pressure value.

[0092] S3. Perform real-time pressure compensation calculation on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution and obtain the optimized control signal; the control signal includes pressure, flow distribution, and valve state.

[0093] Specifically, in step two, we have determined the opening and closing state of each node , and made a preliminary valve state setting according to the control objectives (such as pressure, flow, etc.). However, the valve state setting is based on static data or preliminary assumptions, and may not be able to fully adapt to the real-time changing pressure requirements, especially in cases of dynamic load, environmental changes, etc. At this time, the role of step three is to dynamically calculate the real-time pressure compensation to ensure that the valve can be dynamically adjusted during execution and achieve the optimal opening and closing path, thereby optimizing the working state of the entire system, that is, providing more accurate and stable valve control.

[0094] Measurement of pressure difference: Calculate the pressure difference of each node , i.e., the difference between the pressure of the node and the target pressure:

[0095]

[0096] Where is the target pressure set by the system, usually obtained based on design requirements or real-time monitoring data.

[0097] The innovative adaptive pressure compensation term designed in the present invention , considering various factors such as the pressure difference of the node, the opening / closing state, and the flow difference. The compensation term formula is:

[0098]

[0099] Where, is the adjustment coefficient to balance the influence of the opening / closing state on the compensation term. is the opening / closing state ( or ). is the adjustment coefficient to control the influence of the flow difference on the compensation term. is the flow rate of node , is the average flow rate of all nodes. This compensation term design can dynamically adjust the pressure according to the flow difference of each node, thus ensuring the balance of the pressure in the pipe network system.

[0100] Flow adjustment: Adjust the flow rate of each node through the compensation term , and the new flow rate value is:

[0101]

[0102] Where is the adjustment coefficient to control the influence degree of the compensation term on the flow adjustment.

[0103] This formula ensures that the node flow rate is adjusted accordingly according to the pressure compensation, so that the pressure gradually approaches the target value.

[0104] Valve control: According to the pressure difference of each node and the compensation term , use the Sigmoid function to smoothly adjust the opening / closing state of the valve:

[0105]

[0106] Where is the threshold parameter used to control when to adjust the valve state. The Sigmoid function ensures that the opening or closing of the valve is smooth rather than a sudden switch.

[0107] This control rule automatically adjusts the valve opening degree according to the pressure difference, thereby further optimizing the pressure distribution.

[0108] Output: Optimized pressure, flow distribution, and valve status

[0109] Through compensation calculation, the updated pressure, flow, and valve opening / closing status of each node are output:

[0110] Optimized pressure , the node pressure adjusted through compensation.

[0111] Adjusted flow , the flow value updated based on pressure compensation.

[0112] Valve opening / closing status , the valve status dynamically adjusted according to the pressure difference.

[0113] S4. Convert the control signal into an actual hydraulic opening / closing action, and monitor the valve status in real time. By comparing the pressure and flow of each node monitored through real-time feedback with the target pressure and flow, if the comparison value exceeds the preset threshold, recalculate a new control signal and fine-tune the hydraulic valve.

[0114] Specifically, step 4 is to convert the aforementioned optimized calculation results into actual hydraulic opening / closing actions and monitor the valve status in real time.

[0115] Among them, based on the optimized data passed in from the previous step, first calculate the pressure deviation and flow deviation of each node, where the pressure deviation is calculated from the difference between the current pressure and the target pressure of the node:

[0116]

[0117] Among them, is the current pressure of the optimized node , and is the set target pressure.

[0118] Similarly, the flow deviation is calculated as:

[0119]

[0120] Among them, is the optimized flow value, and

[0121] is the target flow value. Since the valve control signal of the hydraulic system Not only depends on these deviations, but also needs to consider the influence of flow rate on the control signal. Therefore, we added a special regularization term in the generation of the control signal to improve the response of the control signal. The final control signal is given by the following formula:

[0122]

[0123] where is the maximum pressure of the system, is the maximum flow rate of the system; , , are adjustment coefficients to control the influence degree of each term; is the regularization term, designed as an additional term related to the pressure and flow rate deviations, used to balance the feedback effects of pressure and flow rate. This regularization term adjusts its influence degree by introducing the product of pressure and flow rate, and the formula is as follows:

[0124]

[0125] where is the adjustment coefficient of the regularization term to ensure that the combination of pressure and flow rate has an appropriate influence on the control signal.

[0126] According to the control signal calculated above, the hydraulic system controls the opening and closing degree of the valve. The actual opening and closing degree of the hydraulic valve adopts a smooth non-linear model, and maps the control signal to the opening and closing state through the Sigmoid function:

[0127]

[0128] where is the Sigmoid activation function to ensure that the change of the valve state is smooth and does not occur suddenly. The function form is:

[0129]

[0130] where is the adjustment coefficient to control the influence strength of the control signal on the opening and closing degree of the valve. The adjustment coefficient can be adjusted according to the response speed and accuracy of the hydraulic system.

[0131] The system monitors the pressure and flow rate of each node through real-time feedback. These values are compared with the target values to calculate the pressure deviation and flow rate deviation , and it is used to judge the stability of the system and adjust the control signal. The formula for calculating the deviation monitored in real time is as follows:

[0132]

[0133]

[0134] If the actual deviation or exceeds the set threshold, then a new control signal is recalculated according to the adjustment formula , and the hydraulic valve is finely adjusted.

[0135] Through this real-time adjustment, the system can respond to changes in flow rate and pressure, ensuring that the opening and closing state of the hydraulic valve always meets the optimal control target.

[0136] In one embodiment, a hydraulic remote opening and closing system for a formation isolation valve is provided. The system includes:

[0137] A sensor data acquisition unit for acquiring downhole sensor data and establishing a downhole string topology structure model; wherein, the downhole sensor data includes: fluid flow data and pipeline network;

[0138] A graph neural network construction unit for constructing a graph neural network for the fluid flow data and pipeline network and calculating the optimal opening and closing path; wherein, the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing state of each node and the pressure of each node; wherein, each node represents a valve;

[0139] A pressure compensation unit for performing real-time pressure compensation calculation on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution and obtain an optimized control signal; the control signal includes pressure, flow distribution, and valve state;

[0140] A hydraulic valve adjustment unit for converting the control signal into an actual hydraulic opening and closing action, and real-time monitoring of the valve state. By real-time feedback, the pressure and flow rate of each node are compared with the target pressure and flow rate. If the comparison value exceeds the preset threshold, a new control signal is recalculated, and the hydraulic valve is finely adjusted;

[0141] Wherein, the graph neural network uses the current pressure of each node, the flow rate of each node, and the opening and closing state of the valve as nodes, performs weighted graph convolution modeling on the current pressure of each node and the flow rate of each node, and is updated through an adjacency matrix; the adjacency matrix represents the pipeline connection relationship between nodes.

[0142] Unless otherwise specifically stated, the relative steps, numerical expressions, and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0143] If the described functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this 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 system described in various embodiments of the present application. The foregoing storage medium includes: various media 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 that can store program codes.

[0144] In the description of the present application, it should be noted that the orientation or positional relationship indicated by terms such as "upper" and "lower" is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the inventive product is customarily placed during use. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application.

[0145] In the description of the present application, it should also be noted that unless otherwise clearly specified and limited, the terms "set", "install", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of various embodiments of the present application.

Claims

1. A hydraulic remote opening and closing method for a formation isolation valve, characterized in that: The method comprises: Acquire downhole sensor data and establish a downhole pipe string topology model; wherein the downhole sensor data includes: fluid flow data and pipeline network; For the fluid flow data and pipeline network, a graph neural network is constructed to calculate the optimal opening and closing path; wherein the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing state of each node and the pressure of each node; wherein each node represents a valve; Perform real-time pressure compensation calculation on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution and obtain an optimized control signal; the control signal includes pressure, flow distribution and valve status; Convert the control signal into actual hydraulic opening and closing action, monitor the valve status in real time, and compare the pressure and flow of each node with the target pressure and flow through real-time feedback monitoring. If the comparison value exceeds the preset threshold, recalculate the new control signal and fine-tune the hydraulic valve; Among them, the graph neural network uses the current pressure of each node, the flow of each node and the opening and closing status of the valve as nodes, performs weighted graph convolution modeling with the current pressure of each node and the flow of each node, and updates it through the adjacency matrix; the pipeline connection relationship between the nodes of the adjacency matrix.

2. The hydraulic remote opening and closing method of the formation isolation valve according to claim 1 is characterized in that: The edges of the downhole tubular string topology model, wherein the edge weights are calculated based on the length and inner diameter of the hydraulic channel.

3. The hydraulic remote opening and closing method of the formation isolation valve according to claim 1 is characterized in that: The downhole tubular string topology model is a dynamic update mechanism, which updates the edge weights of the connection edges of the topology graph through downhole sensor data collected in real time.

4. The hydraulic remote opening and closing method of the formation isolation valve according to claim 1 is characterized in that: The graph neural network is expressed as: in, Representation Node In the The feature vector of the layer; Is a node The set of neighbor nodes of and They are The weight matrix and bias term of the layer; It is a correction term for node pressure, flow and opening and closing status, which is used to introduce the influence of fluid state.

5. The hydraulic remote opening and closing method of the formation isolation valve according to claim 4 is characterized in that: The graph neural network performs correction based on the fluid state while optimizing the path of the fluid state; The correction based on the fluid state includes: The calculation is performed by weighted summing the pressure difference between the current pressure and the normalized pressure, the difference between the current flow and the normalized flow, and the semi-closed state; Among them, after each round of graph convolution update, the node feature vector output by the graph neural network will be used to optimize the target pressure; the optimization of the target pressure is based on the objective function, which avoids unnecessary losses caused by excessive opening and closing by minimizing the pressure difference of each node.

6. The hydraulic remote opening and closing method of the formation isolation valve according to claim 5 is characterized in that: Before performing real-time pressure compensation calculation on the optimal opening and closing path, obtaining the pressure difference between the pressure of the node and the target pressure; An adaptive pressure compensation item is designed based on the pressure difference, and the adaptive pressure compensation item includes the pressure difference, the opening and closing state, the current flow of the node and the normalized flow.

7. The hydraulic remote opening and closing method of the formation isolation valve according to claim 6 is characterized in that: The real-time pressure compensation for the optimal opening and closing path includes: According to the adaptive pressure compensation term, the flow rate of each node is adjusted by adjusting the coefficient to obtain a new flow rate value; According to the pressure difference and the adaptive pressure compensation term, the Sigmoid function is used to smoothly adjust the valve opening and closing state to obtain a new valve opening and closing state.

8. The hydraulic remote opening and closing method of the formation isolation valve according to claim 7 is characterized in that: The real-time feedback monitoring of the pressure and flow rate of each node and the comparison with the target pressure and flow rate includes: The pressure deviation and flow deviation of each node are calculated; the pressure deviation is the difference between the node pressure adjusted by compensation and the target pressure; the flow deviation is the difference between the node flow value adjusted by compensation and the target flow value.

9. The hydraulic remote opening and closing method of the formation isolation valve according to claim 8, characterized in that: If the comparison value exceeds the preset threshold, a new control signal is recalculated and the hydraulic valve is fine-tuned, including: The new control signal Based on the constraints on the impact of flow on the control signal, the calculation is: in, is the maximum system pressure, is the maximum flow rate of the system; , , is the adjustment coefficient, which controls the influence of each item; is a regularization term, which is designed as an additional term related to the pressure and flow deviations to balance the feedback effects of pressure and flow; It is the new valve opening and closing state; Among them, the regularization term The degree of influence can be adjusted by introducing the product of pressure and flow; The actual switching degree of the hydraulic valve is outputted by a smooth nonlinear model, and a new control signal is mapped by a Sigmoid function. To the open and close state.

10. The hydraulic remote opening and closing system of the formation isolation valve is characterized by: The system comprises: A sensor data acquisition unit is used to acquire downhole sensor data and establish a downhole pipe string topology model; wherein the downhole sensor data includes: fluid flow data and pipeline network; A graph neural network construction unit is used to construct a graph neural network for the fluid flow data and the pipeline network, and calculate an optimal opening and closing path; wherein the objectives of the optimal opening and closing path include: pressure optimization and opening and closing path optimization; the optimal opening and closing path includes: the opening and closing state of each node and the pressure of each node; wherein each node represents a valve; The pressure compensation unit is used to perform real-time pressure compensation calculations on the optimal opening and closing path to ensure that the valve can be dynamically adjusted during execution to obtain an optimized control signal; the control signal includes pressure, flow distribution and valve status; A hydraulic valve adjustment unit, which is used to convert the control signal into an actual hydraulic opening and closing action, and monitor the valve status in real time, and compare the pressure and flow of each node with the target pressure and flow through real-time feedback monitoring. If the comparison value exceeds a preset threshold, a new control signal is recalculated and the hydraulic valve is fine-tuned; Among them, the graph neural network uses the current pressure of each node, the flow of each node and the opening and closing status of the valve as nodes, performs weighted graph convolution modeling with the current pressure of each node and the flow of each node, and updates it through the adjacency matrix; the pipeline connection relationship between the nodes of the adjacency matrix.

Images