Intelligent control method for oil pipe elevator

Abstract

The present invention discloses an intelligent control method for a tubing elevator, and specifically relates to the technical field of tubing elevator control. The method collects operating status data of the tubing elevator, extracts two key characteristic parameters, namely, locking action delay time and microtension fluctuation, and constructs a comprehensive judgment model. The accuracy coefficient of the control system's judgment of dangerous states is output in a polynomial regression manner. The control logic is optimized in real time based on the coefficient and a dynamically adjusted judgment threshold, ensuring that tubing lifting is automatically prohibited and an alarm is issued when the judgment is inaccurate, and that continued operation is allowed when the judgment is accurate, thereby effectively avoiding the risk of misjudgment such as closed but not locked, and improving the judgment accuracy and operational safety of the tubing elevator control system under complex working conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of oil tubing elevator control, and in particular to an intelligent control method for oil tubing elevators. Background Art

[0002] Intelligent tubing elevator control uses intelligent technologies (such as sensors, control systems, and automated programs) to automatically monitor and control the operation of tubing elevators during oil drilling or workover operations, improving operational safety and efficiency. This control system can monitor the elevator's status in real time, determine whether it is fully closed, and automatically open and close the elevator when necessary, reducing human error and ensuring reliability during the tubing lifting and lowering process.

[0003] The existing technology has the following shortcomings:

[0004] In the existing technology, when the control system determines whether the tubing elevator is in a safe state, if the logical rules are not well designed or the state interlocking mechanism fails, it may mistakenly identify a dangerous state such as "closed but not locked" as a safe state, causing the system to incorrectly perform key actions such as lifting the tubing, which in turn may cause the tubing to fall off, equipment damage, or wellhead accidents. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent control method for an oil pipe elevator to solve the deficiencies in the background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a method for intelligently controlling a tubing elevator, comprising:

[0007] Acquiring operating status data of the tubing elevator, including the elevator opening and closing status, locking status, tubing tension data, and control instruction records;

[0008] Extracting key characteristic parameters from the operating state data includes the locking action delay time ΔTlock and the micro-tension fluctuation ΔFprelock before the locking state changes;

[0009] The ΔTlock and ΔFprelock are input into the comprehensive judgment model of the control system for training, and the model output is the accuracy coefficient of the control system's logical judgment of dangerous states;

[0010] The control logic is adjusted based on the comprehensive judgment results. If the judgment is inaccurate, the control system prohibits the execution of the pipe lifting action and issues an alarm signal; if the judgment is accurate, the pipe lifting or lowering operation instructions are allowed to continue;

[0011] The key characteristic parameters and comprehensive judgment results are recorded in the control system database, and the judgment threshold of the accuracy coefficient is dynamically adjusted according to historical data to optimize the judgment accuracy of the control system.

[0012] Preferably, in the step of obtaining the opening and closing state of the elevator, the position sensor is a switch, a Hall element, a photoelectric encoder or an angle sensor, which is used to output a binary signal or an angular displacement indicating whether the elevator is in an open or closed state.

[0013] Preferably, the method for extracting the locking action delay time ΔTlock is as follows: collecting the sensor output signal S(t) during the locking process with a sampling frequency of fs, performing noise suppression processing on the original signal S(t), and recording the processed signal as ; Use numerical methods to derive the smoothed signal and obtain the instantaneous rate of change: ; Where Δt = 1 / fs; Find two significant peak points in the derivative sequence D(t): Tstart: The first time the derivative exceeds the set positive threshold Dth, it indicates that the locking action has started; Tlock: The first time the derivative drops to the negative threshold -Dth or returns to the stable zone, it indicates that the locking action is completed; Calculate the locking action delay time: ΔTlock = Tlock - Tstart; ΔTlock is the locking action delay time.

[0014] Preferably, the method for extracting the micro-tension fluctuation ΔFprelock before the locking state changes is: the control system records the locking feedback completion time Tlock; in the time interval [Tstart, Tlock), the real-time data stream F(t) of the oil pipe tension sensor is continuously collected; the maximum and minimum values ​​of the tension data in the time interval are counted; the micro-tension fluctuation ΔFprelock is calculated, and the expression is: ΔFprelock=max(F(t))-min(F(t)), t∈[Tstart, Tlock); where max(F(t)) and min(F(t)) represent the maximum and minimum values ​​of the tension sensor in the interval.

[0015] Preferably, the ΔTlock and ΔFprelock are input into a comprehensive judgment model of the control system for training, and the model output is the accuracy coefficient of the control system's logical judgment of dangerous states, specifically including:

[0016] ΔTlock and ΔFprelock are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy coefficient label of the control system logic judgment of the dangerous state as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy coefficient labels of all control system logic judgments of the dangerous state as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy coefficient of the control system logic judgment of the dangerous state is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0017] Preferably, the control logic is adjusted according to the comprehensive judgment result. For inaccurate judgment, the control system prohibits the execution of the tubing lifting action and issues an alarm signal; for accurate judgment, the lifting or lowering operation instruction is allowed to continue, specifically including:

[0018] Compare the accuracy coefficient of the dangerous state judged by the control system logic obtained with the preset threshold. If the accuracy coefficient of the dangerous state judged by the control system logic is greater than or equal to the preset threshold, it indicates that the accuracy of the dangerous state judged by the control system logic is high, which is an accurate judgment; if the accuracy coefficient of the dangerous state judged by the control system logic is less than the preset threshold, it indicates that the accuracy of the dangerous state judged by the control system logic is low, which is an inaccurate judgment.

[0019] Preferably, the key characteristic parameters and the comprehensive judgment result are recorded in the control system database, and the judgment threshold of the accuracy coefficient is dynamically adjusted according to historical data, specifically including:

[0020] The key characteristic parameters ΔTlock and ΔFprelock in each tubing elevator control process, the judgment result, and the accuracy coefficient y output by the model ,

[0028] ,

[0027] are written into the database together; continuously accumulate historical data to form a training set;

[0021] Cluster all the recorded accuracy coefficients y pred initially set to two fuzzy classes:

[0022] Class C1: Judged as an inaccurate sample group;

[0023] Class C2: Judged as an accurate sample group;

[0024] Use the AFCM algorithm for iterative clustering to optimize the objective function: ; where u ij is the membership degree of the i-th sample to the j-th cluster center; m is the fuzzy index; c j is the cluster center of the j-th class; N is the total number of samples;

[0025] After the clustering converges, extract two cluster centers c1 < c2, and calculate their average value as the new accuracy judgment threshold; <未见对应原文,猜测为换行符>Update the calculated new accuracy judgment threshold to the control system to replace the original static setting threshold.

[0027] The beneficial effects of the present invention are:

[0028] 1. By introducing two key characteristic parameters: locking action delay time and microtension fluctuation, this invention significantly improves the tubing elevator control system's ability to identify dangerous conditions such as "closed but not locked." By inputting these characteristics into a polynomial regression model for training, a confidence coefficient for the control system's judgment accuracy is output. The system then makes intelligent judgments based on this coefficient and a preset threshold. When an inaccurate condition is identified, the system automatically blocks the tubing elevator operation and issues an alarm, effectively preventing tubing detachment, equipment damage, or wellhead accidents caused by misjudgment, thereby ensuring operational safety.

[0029] 2. This invention utilizes the AFCM (Adaptive Fuzzy C-Means Clustering) algorithm to analyze historical judgment data, dynamically optimizing judgment thresholds and enhancing the control system's adaptability under varying operating conditions. This mechanism not only overcomes the rigidity and poor generalization of traditional threshold settings, but also enables the control system to self-learn and iteratively optimize precision, significantly improving the robustness and accuracy of judgment logic and transforming elevator operation control from "rule-based judgment" to "intelligent judgment." BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a mind map of an intelligent control method for oil pipe elevators. DETAILED DESCRIPTION

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0032] See also Figure 1 As shown, the intelligent control method of a tubing elevator described in this embodiment includes:

[0033] Acquiring operating status data of the tubing elevator, including the elevator opening and closing status, locking status, tubing tension data, and control instruction records;

[0034] Extracting key characteristic parameters from the operating state data includes the locking action delay time ΔTlock and the micro-tension fluctuation ΔFprelock before the locking state changes;

[0035] The ΔTlock and ΔFprelock are input into the comprehensive judgment model of the control system for training, and the model output is the accuracy coefficient of the control system's logical judgment of dangerous states;

[0036] The control logic is adjusted based on the comprehensive judgment results. If the judgment is inaccurate, the control system prohibits the execution of the pipe lifting action and issues an alarm signal; if the judgment is accurate, the pipe lifting or lowering operation instructions are allowed to continue;

[0037] The key characteristic parameters and comprehensive judgment results are recorded in the control system database, and the judgment threshold of the accuracy coefficient is dynamically adjusted according to historical data to optimize the judgment accuracy of the control system.

[0038] In the intelligent control method for tubing elevators, the system first acquires various status data during the elevator's operation to achieve a comprehensive assessment of its safe and operable status. This step relies on a sensor network and real-time data acquisition system, specifically including the following key data types:

[0039] Elevator opening and closing status data: Install a position sensor (such as a proximity switch, Hall element, photoelectric encoder or angle sensor) on the elevator opening and closing mechanism;

[0040] Collection target: Real-time monitoring of whether the elevator is currently in the "open" or "closed" state;

[0041] Technical function: To judge whether the elevator has completed the closing operation, which is the prerequisite for the subsequent locking and pipe lifting operations;

[0042] Data format: can be binary (0 / 1), angle value (°), displacement (mm), etc.

[0043] Sampling frequency: usually 10~100 Hz, to meet the real-time requirements of the motion process.

[0044] Locking state data detection method: set a limit switch, travel switch or magnetic proximity sensor on the locking mechanism (such as wedge block, claw);

[0045] Collection target: Determine whether the elevator has completed locking engagement and whether it is "locked in place";

[0046] Technical function: The locking state is directly related to whether the oil pipe can be firmly clamped, which is the key condition for judging the safety of pipe lifting;

[0047] Data characteristics: At the moment of locking, the sensor state flips edgewise, and the system can synchronously record the time;

[0048] Data output: Usually a binary signal (locked / unlocked), which can also be combined with the action delay to generate the characteristic parameter ΔTlock.

[0049] Oil pipe tension data detection method: Install a tension sensor or strain gauge under the elevator or on the load-bearing component connected to the oil pipe;

[0050] Collection objective: Measure the vertical tension or pressure currently applied to the oil pipe to reflect whether it is in a stressed state;

[0051] Technical function: Determine whether the oil pipe has been lifted before the elevator is locked, or whether the oil pipe is subjected to abnormal force during the locking action, and assist in identifying incorrect operations;

[0052] Data output: Continuous data stream in N or kg, which can be used to calculate microtension fluctuations ΔFprelock;

[0053] Key feature extraction: If the tension data fluctuates significantly before locking is completed, it may be a false lock or premature loading.

[0054] Control instruction recording data collection method: PLC or host computer automatically records all issued control instructions, including action type, trigger time, and execution module;

[0055] Collection content: for example, "closing command issuance time", "locking command execution time", "pipe lifting command triggering time", etc.

[0056] Technical function: Used to compare time with sensor feedback, such as extracting the locking action delay time (ΔTlock) and verifying whether the system response conforms to the expected logic;

[0057] Data format: command logs with timestamps, usually stored in JSON structure, CSV or database records;

[0058] Auxiliary functions: can perform backtracking analysis, support fault tracing and system learning.

[0059] All of the above data is transmitted to the central controller (PLC or embedded controller) via an industrial bus (such as CAN, MODBUS, RS485) or wirelessly. The control system sets up a cache and filtering mechanism to remove jitter signals and erroneous data to ensure the accuracy of status judgment. The data can be synchronously stored in a local database or uploaded to a remote platform through an edge gateway to achieve remote monitoring and analysis.

[0060] The key characteristic parameters extracted from the operating state data include the locking action delay time ΔTlock and the micro-tension fluctuation ΔFprelock before the locking state changes, specifically including:

[0061] The method for extracting the locking action delay time ΔTlock is:

[0062] The sensor output signal S(t) during the locking process is collected, such as the analog signal (voltage, resistance, etc.) from the locking limit switch or the tension change curve, with a sampling frequency of fs. Noise suppression is performed on the original signal S(t). Common methods include: moving average filtering (MAF); Gaussian filtering or median filtering; and wavelet soft threshold denoising (suitable for scenarios with blurred signal edges). The processed signal is recorded as: ; Use numerical methods to derive the smoothed signal and obtain the instantaneous rate of change: Where Δt = 1 / fs. Find two significant peaks in the derivative sequence D(t): Tstart: The first time the derivative exceeds the set positive threshold Dth, indicating the start of locking; Tlock: The first time the derivative drops to the negative threshold -Dth or returns to the stable zone, indicating the completion of locking. Calculate the locking delay time: ΔTlock = Tlock - Tstart; ΔTlock is the locking delay time.

[0063] When the locking action delay is significantly longer than the normal threshold (e.g., >1.5 seconds), it typically indicates a slow or blocked response of the locking mechanism, potentially caused by mechanical jamming, unstable oil pressure, signal delays, and other factors. In this case, if the control system still determines the status as "safe" and allows the pipe lifting operation, its logic may be at risk of misjudging the delayed response. This means it fails to identify the potentially dangerous state represented by the abnormal delay, resulting in low accuracy and insufficient system perception of the fault condition.

[0064] Conversely, if the locking action delay is abnormally short (e.g., <0.1 second), it may indicate premature triggering of the locking feedback signal, a false lock or sensor misreporting. If the system then determines "locking complete" and quickly allows the pipe to be lifted, there is a risk of premature misjudgment and false triggering. This indicates that the control system's dynamic recognition of state changes is inadequate, failing to effectively verify the authenticity of the locking action, resulting in reduced accuracy in dangerous state judgments. Therefore, a reasonable ΔTlock is an important reference indicator for evaluating the robustness of control logic.

[0065] The method for extracting the microtension fluctuation ΔFprelock before the locking state changes is:

[0066] The control system records the locking feedback completion time Tlock;

[0067] In the time interval [Tstart, Tlock), the real-time data stream F(t) of the oil pipe tension sensor is continuously collected;

[0068] The maximum and minimum values ​​of the tension data within the time interval are counted; the micro-tension fluctuation ΔFprelock is calculated to reflect whether the oil pipe is prematurely stressed or not locked firmly during the locking process. The expression is: ΔFprelock = max(F(t)) - min(F(t)), t∈[Tstart, Tlock); where max(F(t)) and min(F(t)) represent the maximum and minimum values ​​of the tension sensor within the interval.

[0069] Significantly large microtension fluctuations (ΔFprelock) typically indicate that the tubing has begun to bear weight or experience stress changes before fully locking. This could be due to premature tubing lifting before the locking mechanism is fully engaged, or abnormal conditions such as elevator slippage or uneven force distribution. If the control system still determines the status as "safe" in this situation, it lacks the ability to identify abnormal force distribution, creating the risk of misjudgment and failing to effectively prevent dangerous events such as tubing detachment caused by false or partial locking.

[0070] Conversely, if ΔFprelock is abnormally small or close to zero, two possible scenarios could exist: first, the locking process is smooth, with no tension disturbances, indicating a truly safe state; second, the tension sensor is not functioning properly or is installed in a location that fails to sense actual micro-deformations, resulting in "distorted" sensor data. In the latter case, the system may rely on invalid data to make "safe" judgments, leading to misjudgments of safety in data blind spots and reducing the accuracy of dangerous state judgments. Therefore, any abnormal magnitude of microtension fluctuations may indicate a vulnerability in the system's judgment mechanism, requiring multi-dimensional verification in conjunction with other characteristic parameters.

[0071] The ΔTlock and ΔFprelock are input into the comprehensive judgment model of the control system for training. The model output is the accuracy coefficient of the control system's logical judgment of dangerous states, specifically including:

[0072] ΔTlock and ΔFprelock are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy coefficient label of the control system logic judgment of the dangerous state as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy coefficient labels of all control system logic judgments of the dangerous state as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy coefficient of the control system logic judgment of the dangerous state is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0073] The present invention can select comparative analysis of "manual control logic system vs. the present invention system under certain operating scenarios (or historical working conditions)" or experimental data of "system accuracy indicators before and after optimization" to verify the feasibility of the present invention:

[0074] Table 1: Comparison of control system judgment accuracy

[0075]

[0076] The control logic is adjusted based on the comprehensive judgment results. If the judgment is inaccurate, the control system prohibits the execution of the pipe lifting action and issues an alarm signal; if the judgment is accurate, the pipe lifting or lowering operation instructions are allowed to continue, including:

[0077] The obtained accuracy coefficient of the control system logic judgment of the dangerous state is compared with the preset threshold. If the accuracy coefficient of the control system logic judgment of the dangerous state is greater than or equal to the preset threshold, it means that the control system logic judgment of the dangerous state is high in accuracy, which is an accurate judgment; if the accuracy coefficient of the control system logic judgment of the dangerous state is less than the preset threshold, it means that the control system logic judgment of the dangerous state is low in accuracy, which is an inaccuracy judgment.

[0078] Table 2: Comparison of accident rates such as oil pipe detachment and false alarms

[0079]

[0080] The key characteristic parameters and comprehensive judgment results are recorded in the control system database, and the judgment threshold of the accuracy coefficient is dynamically adjusted according to historical data to optimize the judgment accuracy of the control system, including:

[0081] The key characteristic parameters (ΔTlock and ΔFprelock) of each tubing elevator control process are compared with the system judgment result (accurate or inaccurate) and the accuracy coefficient y of the model output. pred Write them into the database together; continue to accumulate historical data to form a training set that can be used for model correction.

[0082] The AFCM algorithm is selected to dynamically adjust the judgment threshold of the accuracy coefficient based on historical data. AFCM can assign different membership degrees to samples belonging to multiple clusters, which is suitable for dealing with problems such as fuzzy judgment boundaries and continuously changing states. It has a center update mechanism and the ability to dynamically adjust the fuzzy coefficient, and can adapt to the judgment distribution characteristics under different historical working conditions.

[0083] The accuracy coefficient y for all records pred Perform clustering, initially set to two fuzzy classes:

[0084] Class C1: The sample group judged as inaccurate;

[0085] Class C2: judged as the accuracy sample group;

[0086] The AFCM algorithm is used for iterative clustering to optimize the objective function: ; where u ij is the membership degree of the i-th sample to the j-th clustering center; m is the fuzzy exponent (usually taken as 1.5 - 2.5); c j is the clustering center of the j-th class; N is the total number of samples.

[0087] After the clustering converges, two clustering centers c1 < c2 are extracted, and their average value is calculated as the new accuracy judgment threshold. The expression is: .

[0088] The calculated new accuracy judgment threshold is updated to the control system judgment module to replace the original static set threshold; in each subsequent control operation, this updated threshold is used to judge whether the accuracy coefficient of the current model output meets the standard, and decide whether to perform the pipe lifting operation; every certain operation cycle (such as daily / every 100 operations), the AFCM clustering is re-run for an adaptive threshold optimization.

[0089] Table 3: Dynamic optimization effect of accuracy coefficient

[0090]

[0091] This embodiment provides an intelligent control method for a tubing elevator. By collecting key characteristic parameters during the operation of the elevator, including the locking action delay time and the micro-tension fluctuation before the change of the locking state, a comprehensive feature vector is constructed and input into a polynomial regression model, and the accuracy coefficient of the control system's judgment of the dangerous state is output; combined with this coefficient and the dynamically updated judgment threshold, an adaptive adjustment of the control logic is realized, ensuring that the pipe lifting is blocked and an alarm is issued when the accuracy is insufficient, and improving the system's recognition ability of abnormal states. In addition, the fuzzy C-means clustering algorithm is used to analyze the historical judgment data, and the judgment threshold of the accuracy coefficient is dynamically optimized, enhancing the intelligent judgment accuracy and operation safety of the control system under complex working conditions.

[0092] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0093] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0094] It should be understood that the term "and / or" herein 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 three situations: A exists alone, A and B exist at the same time, and B exists alone, where A and B may be singular or plural. In addition, the character " / " herein generally indicates that the objects associated with each other are in an "or" relationship, but it may also indicate an "and / or" relationship, which can be understood by referring to the context. A person of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0095] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An intelligent control method for a tubing elevator, characterized by: include: Acquiring operating status data of the tubing elevator, including the elevator opening and closing status, locking status, tubing tension data, and control instruction records; Extracting key characteristic parameters from the operating state data includes the locking action delay time ΔTlock and the micro-tension fluctuation ΔFprelock before the locking state changes; The ΔTlock and ΔFprelock are input into the comprehensive judgment model of the control system for training, and the model output is the accuracy coefficient of the control system's logical judgment of dangerous states; The control logic is adjusted based on the comprehensive judgment results. If the judgment is inaccurate, the control system prohibits the execution of the pipe lifting action and issues an alarm signal; if the judgment is accurate, the pipe lifting or lowering operation instructions are allowed to continue; The key characteristic parameters and comprehensive judgment results are recorded in the control system database, and the judgment threshold of the accuracy coefficient is dynamically adjusted according to historical data to optimize the judgment accuracy of the control system.

2. The intelligent control method for a tubing elevator according to claim 1, characterized in that: In the step of obtaining the open and closed state of the elevator, the position sensor is a switch, a Hall element, a photoelectric encoder or an angle sensor, which is used to output a binary signal or an angular displacement indicating whether the elevator is in an open or closed state.

3. The intelligent control method for a tubing elevator according to claim 1, characterized in that: The method for extracting the locking action delay time ΔTlock is as follows: the sensor output signal S(t) of the locking process is collected with a sampling frequency of fs, and the original signal S(t) is subjected to noise suppression processing. The processed signal is recorded as ; Use numerical methods to derive the smoothed signal and obtain the instantaneous rate of change: ; Where Δt = 1 / fs; Find two significant peak points in the derivative sequence D(t): Tstart: The first time the derivative exceeds the set positive threshold Dth, it indicates that the locking action has begun; Tlock: The first time the derivative drops to the negative threshold -Dth or returns to the stable zone, it indicates that the locking action is completed; Calculate the locking action delay time: ΔTlock = Tlock - Tstart; ΔTlock is the locking action delay time.

4. The intelligent control method for a tubing elevator according to claim 3, characterized in that: The method for extracting the microtension fluctuation ΔFprelock before the locking state changes is as follows: the control system records the locking feedback completion time Tlock; continuously collects the real-time data stream F(t) from the oil pipe tension sensor during the time interval [Tstart, Tlock); and calculates the maximum and minimum values ​​of the tension data within the time interval. Calculate the microtension fluctuation ΔFprelock using the following expression: ΔFprelock = max(F(t)) - min(F(t)), t∈[Tstart, Tlock); where max(F(t)) and min(F(t)) represent the maximum and minimum values ​​of the tension sensor within the interval.

5. The intelligent control method for a tubing elevator according to claim 4, characterized in that: The ΔTlock and ΔFprelock are input into the comprehensive judgment model of the control system for training. The model output is the accuracy coefficient of the control system's logical judgment of dangerous states, specifically including: ΔTlock and ΔFprelock are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy coefficient label of the control system logic judgment of the dangerous state as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy coefficient labels of all control system logic judgments of the dangerous state as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy coefficient of the control system logic judgment of the dangerous state is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

6. The intelligent control method for a tubing elevator according to claim 5, characterized in that: The control logic is adjusted based on the comprehensive judgment results. If the judgment is inaccurate, the control system prohibits the execution of the pipe lifting action and issues an alarm signal; if the judgment is accurate, the pipe lifting or lowering operation instructions are allowed to continue, including: Compare the obtained accuracy coefficient of the control system's logical judgment of the dangerous state with the preset threshold. If the accuracy coefficient of the control system's logical judgment of the dangerous state is greater than or equal to the preset threshold, it indicates that the accuracy of the control system's logical judgment of the dangerous state is high, which is an accuracy judgment; if the accuracy coefficient of the control system's logical judgment of the dangerous state is less than the preset threshold, it indicates that the accuracy of the control system's logical judgment of the dangerous state is low, which is an inaccuracy judgment.

7. The intelligent control method for a tubing elevator according to claim 6, characterized in that: Record the key feature parameters and the comprehensive judgment results into the control system database, and dynamically adjust the judgment threshold of the accuracy coefficient according to historical data, specifically including: The key characteristic parameters ΔTlock and ΔFprelock in each tubing elevator control process are compared with the judgment results and the accuracy coefficient y of the model output pred Write them into the database together; continue to accumulate historical data to form a training set; The accuracy coefficient y for all records pred Perform clustering, initially set to two fuzzy classes: Class C1: The sample group judged as an inaccuracy sample group; Class C2: The sample group judged as an accuracy sample group; The AFCM algorithm is used for iterative clustering to optimize the objective function: Where u ij is the membership degree of the i-th sample to the j-th cluster center; m is the fuzzy index; c j is the cluster center of the jth class; N is the total number of samples; After clustering convergence, extract two cluster centers c1 < c2, and calculate their average value as the new accuracy judgment threshold; Update the calculated new accuracy judgment threshold to the control system to replace the original static setting threshold.

Images