Intelligent fault diagnosis method for digital twin-driven offshore oil well control equipment

Abstract

The invention belongs to the technical field of ocean engineering equipment detection, and particularly relates to an intelligent fault diagnosis method for digital twin-driven offshore oil well control equipment. According to the intelligent fault diagnosis method, the continuous optimization of the fault diagnosis model is realized, and the long-term stable and reliable effect of the fault diagnosis process is ensured. The intelligent fault diagnosis method for the digital twin-driven offshore oil well control equipment comprises the following steps: screening to obtain a fault diagnosis model used in an intelligent fault diagnosis process, and updating and adjusting the fault diagnosis model; acquiring real-time operation data of the offshore oil well control equipment to be subjected to fault diagnosis, and performing initial fault diagnosis on the offshore oil well control equipment by using the screened fault diagnosis model used in the intelligent fault diagnosis process; according to the result of the initial fault diagnosis, performing parameter updating on the digital twin model corresponding to the offshore oil well control equipment to be subjected to fault diagnosis; and based on the digital twinborn model after parameter updating, performing fault verification and re-diagnosis on the offshore oil well control equipment to be subjected to fault diagnosis.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of marine engineering equipment detection, and particularly relates to a digital twin driven intelligent fault diagnosis method for marine oil well control equipment. BACKGROUND

[0002] As an important source of oil and gas supply in China, marine oil and gas is used to close the wellhead in an emergency when well overflow or blowout occurs in the process of underwater oil and gas exploitation, so as to prevent major production accidents and is the last line of defense to protect marine oil and gas production. In order to ensure its safe and stable operation, technical personnel based on fault diagnosis technology, through monitoring the sensing data of marine oil well control equipment, can judge the equipment state in real time and alarm in time when fault occurs. It is worth noting that the design life of marine oil well control equipment is usually 25-30 years and is in a long-term underwater service state, so the fault diagnosis technology needs to maintain the stability of the diagnosis performance for a long time, and the diagnosis model should be self-optimized and iterated with the change of marine oil well control equipment.

[0003] However, after further research, the inventors found that the existing marine oil well control equipment diagnosis method mostly relies on a single model for implementation, and it is difficult to realize the continuous optimization and improvement of the diagnosis model. With the continuous operation of marine oil well control equipment, its diagnosis ability will gradually be lost until it can no longer meet the needs of on-site diagnosis. In the absence of effective monitoring and diagnosis, potential faults are likely to occur in the marine oil production site, which may lead to the failure of well control equipment and cause serious safety accidents. Therefore, it is particularly necessary for those skilled in the art to design a new digital twin driven intelligent fault diagnosis method. SUMMARY

[0004] The application provides a digital twin driven intelligent fault diagnosis method for marine oil well control equipment, which includes the steps of selection and updating of fault diagnosis model, initial fault diagnosis, and fault verification and re-diagnosis based on digital twin model. Through the above steps, the intelligent fault diagnosis method realizes the continuous optimization of the fault diagnosis model and guarantees the long-term stable and reliable effect of the fault diagnosis process.

[0005] To solve the above technical problems, the application adopts the following technical solutions: The digital twin driven intelligent fault diagnosis method for marine oil well control equipment includes the following steps: Step S1: screening the fault diagnosis model used in the intelligent fault diagnosis process and updating and adjusting the fault diagnosis model; Step S2: Collect real-time operation data of the offshore oil well control equipment to be fault diagnosed, and perform initial fault diagnosis on the collected data using the fault diagnosis model selected in step S1; Step S3: According to the result of the initial fault diagnosis, update the parameters of the digital twin model corresponding to the offshore oil well control equipment to be fault diagnosed; Based on the updated digital twin model, perform fault verification and re-diagnosis on the offshore oil well control equipment to be fault diagnosed.

[0006] Preferably, the process of selecting the fault diagnosis model used in the intelligent fault diagnosis process in step S1 includes the following steps: Step S101: Select a plurality of fault diagnosis models as candidates, and initialize the model parameters; Step S102: Collect the operation data of the offshore oil well control equipment to be fault diagnosed within one year to generate a test set; Step S103: Input the data in the test set obtained in step S102 into each candidate fault diagnosis model; compare the diagnosis results of each candidate fault diagnosis model with the actual data labels respectively to obtain the diagnosis accuracy of each candidate fault diagnosis model; Step S104: Select the fault diagnosis model with a diagnosis accuracy greater than the accuracy threshold from the candidate fault diagnosis models.

[0007] Preferably, the process of selecting the fault diagnosis model used in the intelligent fault diagnosis process in step S1 further includes the following steps: Step S105: Collect the operation history data of the offshore oil well control equipment to be fault diagnosed to construct a model training set; Step S106: Divide the fault diagnosis models with an accuracy greater than the accuracy threshold selected in step S104 into model-based fault diagnosis models and data-driven fault diagnosis models according to types; Train the data-driven fault diagnosis models in the model training set; Step S107: Calculate the diagnosis accuracy of each data-driven fault diagnosis model and the diagnosis accuracy of the model-based fault diagnosis model respectively; Step S108: Select the fault diagnosis model with the highest diagnosis accuracy.

[0008] Preferably, the process of updating and adjusting the fault diagnosis model used in the intelligent fault diagnosis process in step S1 includes the following steps: Step S109: Determine whether the diagnosis accuracy of the fault diagnosis model with the highest diagnosis accuracy selected in step S108 meets the fault diagnosis threshold requirement; If the conditions are met, the fault diagnosis model with the highest diagnostic accuracy is selected as the fault diagnosis model used in the intelligent fault diagnosis process. If the conditions are not met, the model-based fault diagnosis model with the highest diagnostic accuracy is selected as the fault diagnosis model used in the intelligent fault diagnosis process. Its conditional probability is adjusted and the diagnostic accuracy is recalculated until its diagnostic accuracy meets the fault diagnosis threshold requirements.

[0009] Preferably, step S3 specifically includes the following steps: Step S301: Based on the results of the initial fault diagnosis, update the parameters of the digital twin model corresponding to the offshore oil well control equipment to be diagnosed. Among them, the fault parameters of the digital twin model The update rule is: if If this indicates that valve CVx is in a fault state; if If x = 1, 2, ..., t, then it means that valve CVx is in a normal state; x satisfies: x = 1, 2, ..., t; Step S302: Run the digital twin model to obtain the on / off status of each valve in the offshore oil well control equipment to be diagnosed, and update the control parameters. Among them, the control parameters of the digital twin model The update rule is: if valve Cx is in the open state, then let If valve Cx is in the closed state, then let ; Step S303: Obtain the diagnostic results of the digital twin model and calculate the oil pressure value of the offshore oil well control equipment to be diagnosed. Among them, the oil pressure values ​​of the offshore oil well control equipment to be diagnosed for faults must meet the following requirements: ; In the formula, This represents the pipeline pressure before the i-th valve; This indicates the density of crude oil, obtained from well logging data. It is the acceleration due to gravity; This indicates the height of the pipeline center before the i-th valve, with the wellhead as the reference height; This represents the pipeline pressure after the i-th valve; This indicates the height of the pipeline center after the i-th valve, with the wellhead as the reference height; This indicates the pressure loss caused by the i-th valve, which can be obtained by consulting the valve inspection report. These are the control parameters for the i-th valve; Step S304: comparing the actual pressure collected by the offshore well control equipment to be fault diagnosed with the pressure calculated by the digital twin model; Wherein, the error value between the actual pressure collected by the offshore well control equipment to be fault diagnosed and the pressure calculated by the digital twin model satisfies: ; In the formula, is the error value between the actual pressure of the i valve outlet collected by the offshore well control equipment to be fault diagnosed and the pressure calculated by the digital twin model; is the actual pressure collected by the i valve outlet of the offshore well control equipment to be fault diagnosed; Step S305: judging whether the error value between the actual pressure collected by the offshore well control equipment to be fault diagnosed and the pressure calculated by the digital twin model satisfies the diagnostic error requirement; If yes, output the diagnostic result of the digital twin model; Step S306: if no, obtain error diagnostic information, and mark the valve number currently diagnosed as fault as f; Step S307: further judging whether the fault diagnostic model used in the intelligent fault diagnosis process is a model-based fault diagnostic model or a data-driven fault diagnostic model; If the diagnostic model currently used is a model-based fault diagnostic model, set the state CVf of the valve with number f to normal, and infer fault again through the model-based fault diagnostic model; If the diagnostic model currently used is a data-driven diagnostic model, set the feedback state CVfs of the valve with number f to normal, and infer fault again through the data-driven diagnostic model; After the inference, return to step S301, and continue to run the subsequent steps.

[0010] More preferably, the model-based fault diagnostic model in step S106 satisfies: ; In the formula, represents the probability of the x valve being fault under the condition that , , , represents the probability of the i sensor being in the current state under the condition that the x valve is fault, and the value is obtained through expert decision; represents the probability of the x valve being fault, represents the probability of the i sensor being in the current state under the condition that the x valve is normal; Probability of the xth valve being normal.

[0011] The application provides a digital twin driven intelligent fault diagnosis method for offshore oil well control equipment.

[0012] The digital twin driven intelligent fault diagnosis method for offshore oil well control equipment has at least the following technical advantages: the digital twin driven intelligent fault diagnosis method for offshore oil well control equipment provided by the application can realize the evolution of the diagnosis model with the change of the offshore oil well control equipment by selecting a suitable fault diagnosis model and implementing updating and adjustment; in addition, the digital twin driven intelligent fault diagnosis method for offshore oil well control equipment provided by the application further performs initial diagnosis and fault verification and re-diagnosis based on the digital twin model, and the fault diagnosis result is more reliable compared with the prior art, thereby providing strong technical support for long-term stable diagnosis of the offshore oil well control equipment. BRIEF DESCRIPTION OF DRAWINGS

[0013] The accompanying drawings are used to provide a further understanding of the application, and constitute a part of the specification, and are used to explain the application together with embodiments of the application, and do not constitute a limitation on the application. In the following drawings: Figure 1 A flowchart of the digital twin driven intelligent fault diagnosis method for offshore oil well control equipment provided by the application; Figure 2 An architecture diagram of model-based fault diagnosis; Figure 3 An architecture diagram of data-driven fault diagnosis. DETAILED DESCRIPTION

[0014] The application provides a digital twin driven intelligent fault diagnosis method for offshore oil well control equipment, which comprises the steps of selecting and updating a fault diagnosis model, initial fault diagnosis, and fault verification and re-diagnosis based on a digital twin model.

[0015] As shown in Figure 1 The digital twin driven intelligent fault diagnosis method for offshore oil well control equipment provided by the application specifically comprises the following steps: Step S1: screening a fault diagnosis model used in the intelligent fault diagnosis process and updating and adjusting the fault diagnosis model.

[0016] As a more preferred embodiment of the application, the process of screening the fault diagnosis model used in the intelligent fault diagnosis process in step S1 specifically comprises the following steps: Step S101: selecting a plurality of fault diagnosis models as candidates and initializing model parameters of the fault diagnosis models.

[0017] It is worth noting that the purpose of selecting the fault diagnosis model is to exclude obviously unsuitable fault diagnosis models, thereby reducing the burden of the fault diagnosis model screening process.

[0018] Step S102: collecting operation data of the offshore oil well control equipment to be fault diagnosed within one year to generate a test set.

[0019] On the basis of completing step S101, step S102 is further implemented. Step S102 first collects operation data of the offshore oil well control equipment to be fault diagnosed within one year; wherein the operation data at least includes fault data and normal operation data of the offshore oil well control equipment. Then, based on the operation data of the offshore oil well control equipment to be fault diagnosed within one year, a test set is constructed.

[0020] Step S103: inputting data in the test set obtained in step S102 into each candidate fault diagnosis model; comparing diagnosis results of each candidate fault diagnosis model with actual data labels respectively to obtain diagnosis accuracy of each candidate fault diagnosis model.

[0021] On the basis of completing step S102, step S103 is further implemented. In step S103, the diagnostic accuracy of each candidate fault diagnosis model is obtained by comparing the diagnostic result of each candidate fault diagnosis model with the actual data label. Further, the diagnostic accuracy of the fault diagnosis model refers to the percentage of the amount of data correctly diagnosed to the total amount of data. Therefore, the higher the diagnostic accuracy of the fault diagnosis model, the more accurate the diagnostic result of the fault diagnosis model.

[0022] Step S104: selecting a fault diagnosis model with a diagnostic accuracy greater than the accuracy threshold from the candidate fault diagnosis models.

[0023] It should be noted that, in order to improve the frequency of screening the fault diagnosis model, the test time can be further increased as a condition for adjusting the fault diagnosis model. For example, for the fault diagnosis model with a diagnostic accuracy greater than the accuracy threshold and a test time less than the set threshold, the model adjustment can be omitted. For the fault diagnosis model with a test time greater than the set threshold, regardless of the previous diagnostic accuracy, the model adjustment should be performed again to determine whether the diagnostic accuracy of the fault diagnosis model still meets the requirements.

[0024] As a preferred embodiment of the present application, the process of selecting the fault diagnosis model used in the intelligent fault diagnosis process in step S1 further includes the following steps: Step S105: collecting the operation history data of the offshore oil well control equipment to be diagnosed and constructing a model training set.

[0025] The operation history data of the offshore oil well control equipment to be diagnosed should at least include the fault data and the normal operation data of the offshore oil well control equipment. Based on the above operation history data of the offshore oil well control equipment to be diagnosed, the model training set can be constructed.

[0026] Step S106: dividing the fault diagnosis model with an accuracy greater than the accuracy threshold selected in step S104 into a model-based fault diagnosis model and a data-driven fault diagnosis model according to the type.

[0027] The data-driven fault diagnosis model in the model training set is trained.

[0028] On the basis of completing step S105, step S106 is further implemented. It should be noted that in step S106, the fault diagnosis model selected in step S104 is first classified into a model-based fault diagnosis model and a data-driven fault diagnosis model.

[0029] For example, the model-based fault diagnosis model can be a neural network model, and the data-driven fault diagnosis model can be a support vector machine model.Figure 2 As shown, the fault diagnosis process of the model-based fault diagnosis model satisfies the following: ; In the formula, Indicates that in the known , , Under the given conditions, the probability of the x-th valve failing; This represents the probability that the i-th sensor is in its current state given that the x-th valve fails; this value is obtained through expert decision-making. This represents the probability that the x-th valve will fail. This represents the probability that the i-th sensor is in its current state given that the x-th valve is functioning normally. This represents the probability that the x-th valve is functioning correctly.

[0030] It is worth noting that this model-based fault diagnosis model does not require training with a training set.

[0031] For details on data-driven fault diagnosis models, please refer to, for example... Figure 3 As shown. Specifically, Figure 3 In the diagram, PT represents pressure sensor data, acquired through data collection from sensors in offshore oil well control equipment; CV represents valves in offshore oil well control equipment, whose status includes normal and fault conditions. is Let represent the feedback state of the i-th valve, which includes unknown and normal states, and h represent the hidden layer nodes. It is worth noting that since the data-driven fault diagnosis model implements its diagnostic process through the basic principles of Bayesian regularized neural networks, it needs to be trained using a training set.

[0032] Step S107: Calculate the diagnostic accuracy of each data-driven fault diagnosis model and the diagnostic accuracy of each model-based fault diagnosis model.

[0033] Step S108: Select the fault diagnosis model with the highest diagnostic accuracy.

[0034] In a preferred embodiment of the present invention, the process of updating and adjusting the fault diagnosis model used in the intelligent fault diagnosis process in step S1 specifically includes the following steps: Step S109: Determine whether the diagnostic accuracy of the fault diagnosis model with the highest diagnostic accuracy obtained from the screening meets the fault diagnosis threshold requirement. If the conditions are met, the fault diagnosis model with the highest diagnostic accuracy is selected as the fault diagnosis model used in the intelligent fault diagnosis process. If not, the model-based fault diagnosis model with the highest diagnosis accuracy is selected as the fault diagnosis model used in the intelligent fault diagnosis process, its conditional probability is adjusted, and the diagnosis accuracy is recalculated until its diagnosis accuracy meets the fault diagnosis threshold requirement.

[0035] On the basis of completing step S106, steps S107, S108, and S109 are further implemented. It is worth noting that in the process of screening, updating, and adjusting the fault diagnosis model, in order to obtain a fault diagnosis model whose fault diagnosis capability best meets the marine oil well control equipment, the diagnosis accuracy of each fault diagnosis model is first calculated, and the fault diagnosis model with the highest diagnosis accuracy is selected. Then, further fault diagnosis determination is made on the fault diagnosis model with the highest diagnosis accuracy, so as to determine whether the fault diagnosis model with the highest diagnosis accuracy meets the fault diagnosis threshold requirement of the marine oil well control equipment. Further, in the case where the diagnosis accuracy of the fault diagnosis model with the highest diagnosis accuracy does not meet the fault diagnosis threshold requirement, an alternative solution is given, that is, the model-based fault diagnosis model with the highest diagnosis accuracy is reselected as the fault diagnosis model used in the intelligent fault diagnosis process, and the model-based fault diagnosis model with the highest diagnosis accuracy is updated and adjusted until the diagnosis accuracy of the model-based fault diagnosis model with the highest diagnosis accuracy meets the fault diagnosis threshold requirement.

[0036] Up to now, the fault diagnosis model used in the intelligent fault diagnosis process is selected through steps S101-S109, which can be used for specific fault diagnosis of the subsequent marine oil well control equipment.

[0037] Step S2: Collecting real-time operation data of the marine oil well control equipment to be fault diagnosed, and using the selected fault diagnosis model used in the intelligent fault diagnosis process to perform fault preliminary diagnosis.

[0038] Step S3: According to the result of the fault preliminary diagnosis, updating the parameters of the digital twin model corresponding to the marine oil well control equipment to be fault diagnosed. Based on the parameter-updated digital twin model, performing fault verification and re-diagnosis on the marine oil well control equipment to be fault diagnosed.

[0039] As a more preferred embodiment of the present application, the step S3 specifically includes the following steps: Step S301: According to the result of the fault preliminary diagnosis, updating the parameters of the digital twin model corresponding to the marine oil well control equipment to be fault diagnosed.

[0040] The update rule of the fault parameters of the digital twin model is that if the fault diagnosis result of the marine oil well control equipment to be fault diagnosed is not consistent with the fault diagnosis result of the digital twin model corresponding to the marine oil well control equipment to be fault diagnosed, the fault parameters of the digital twin model are updated. the update rule of the fault parameters of the digital twin model is that if the fault diagnosis result of the marine oil well control equipment to be fault diagnosed is not consistent with the fault diagnosis result of the digital twin model corresponding to the marine oil well control equipment to be fault diagnosed, the fault parameters of the digital twin model are updated. , then it indicates that the valve CVx is in a fault state; if , then it indicates that the valve CVx is in a normal state; x satisfies: x = 1, 2, … t.

[0041] Step S302: running the digital twin model to obtain the opening and closing state of each valve in the offshore oil well control equipment to be diagnosed, and updating the control parameters.

[0042] The control parameters of the digital twin model are updated according to the following rules: if the valve Cx is in an open state, then ; if the valve Cx is in a closed state, then . .

[0043] Step S303: obtaining the diagnosis result of the digital twin model and calculating the oil pressure value of the offshore oil well control equipment to be diagnosed.

[0044] On the basis of completing steps S301 and S302, step S303 is further implemented. It is worth noting that the oil pressure value of the offshore oil well control equipment to be diagnosed satisfies: .

[0045] In the formula, represents the pipeline pressure before the i th valve; represents the crude oil density, which is obtained according to the logging data; is the acceleration of gravity; represents the height of the pipeline center before the i th valve, with the wellhead as the reference height; represents the pipeline pressure after the i th valve; represents the height of the pipeline center after the i th valve, with the wellhead as the reference height; represents the pressure loss caused by the i th valve, which is given by consulting the valve detection report; is the control parameter of the i th valve.

[0046] Step S304: comparing the actual pressure collected by the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model.

[0047] On the basis of completing step S303, step S304 is further implemented. The error value between the actual pressure collected by the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model satisfies: .

[0048] In the formula, an error value between the actual pressure collected from the i-th valve outlet of the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model; the actual pressure collected from the i-th valve outlet of the offshore oil well control equipment to be diagnosed.

[0049] Step S305: determining whether the error value between the actual pressure collected from the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model meets the diagnostic error requirement.

[0050] On the basis of completing step S304, step S305 is further implemented. It needs to be noted that, for the error between the actual pressure collected from the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model, if the error value meets the diagnostic error requirement, it means that the pressure calculated by the digital twin model can be used for diagnosis and reflects the actual pressure collected from the offshore oil well control equipment to be diagnosed. At this time, the diagnostic result calculated by the digital twin model is output, and the process of intelligent fault diagnosis of the offshore oil well control equipment based on the digital twin driving is completed.

[0051] Step S306: if not, error diagnosis information is obtained, and the valve number currently diagnosed as faulty is marked as f.

[0052] Step S307: further determining whether the fault diagnosis model used in the intelligent fault diagnosis process is a model-based fault diagnosis model or a data-driven fault diagnosis model. If the currently used diagnostic model is a model-based fault diagnosis model, the state CVf of the valve with number f is set to normal, and the fault is inferred again by the model-based fault diagnosis model. If the current diagnostic model is a data-driven diagnostic model, the feedback state CVfs of the valve with number f is set to normal, and the fault is inferred again by the data-driven diagnostic model. After the inference is completed, return to step S301 and continue to run the subsequent steps.

[0053] On the basis of completing step S305, steps S306 and S307 are further implemented. For the case that the error between the actual pressure collected from the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model cannot meet the diagnostic error requirement, it means that the pressure calculated by the digital twin model cannot be directly used for diagnosis and reflect the actual pressure collected from the offshore oil well control equipment to be diagnosed. At this time, the fault needs to be verified and diagnosed again.

[0054] Specifically, the process of verifying and re-diagnosing the fault can be described as: obtaining the error diagnosis information, and first marking the valve number currently diagnosed as a fault as f. After marking is completed, based on the different types of fault diagnosis models used in the intelligent fault diagnosis process, the process of fault verification and re-diagnosis is also different.

[0055] Specifically, if the currently used diagnosis model is a model-based fault diagnosis model, the state CVf of the valve numbered f is set to normal, and the fault is inferred again through the model-based fault diagnosis model; if the current diagnosis model is a data-driven diagnosis model, the feedback state CVfs of the valve numbered f is set to normal, and the fault is inferred again through the data-driven diagnosis model.

[0056] Then, after the inference is completed, return to step S301 and continue to run the subsequent steps until the error value between the pressure calculated by the digital twin model and the actual pressure value obtained from the offshore oil well control equipment to be diagnosed meets the diagnosis error requirement. At this time, the diagnosis result calculated by the digital twin model is output, and the intelligent fault diagnosis process of the offshore oil well control equipment driven by the digital twin is completed.

[0057] The present application provides an intelligent fault diagnosis method for offshore oil well control equipment driven by digital twin. Specifically, the intelligent fault diagnosis method for offshore oil well control equipment driven by digital twin comprises the following steps: screening the fault diagnosis model used in the intelligent fault diagnosis process and updating and adjusting it; collecting real-time operation data of the offshore oil well control equipment to be diagnosed, and using the screened fault diagnosis model used in the intelligent fault diagnosis process to perform initial fault diagnosis; updating the parameters of the digital twin model corresponding to the offshore oil well control equipment to be diagnosed according to the result of the initial fault diagnosis; and performing fault verification and re-diagnosis on the offshore oil well control equipment to be diagnosed based on the digital twin model after parameter updating.

[0058] The intelligent fault diagnosis method for offshore oil well control equipment driven by digital twin has at least the following technical advantages: the intelligent fault diagnosis method for offshore oil well control equipment driven by digital twin provided by the present application selects a suitable fault diagnosis model and updates and adjusts it, so that the diagnosis model evolves with the changes of the offshore oil well control equipment; in addition, the intelligent fault diagnosis method for offshore oil well control equipment driven by digital twin further performs initial fault diagnosis and fault verification and re-diagnosis based on the digital twin model, so that the fault diagnosis result is more reliable compared with the prior art, and provides strong technical support for long-term stable diagnosis of offshore oil well control equipment.

[0059] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A digital-twin-driven intelligent fault diagnosis method for offshore oil well control equipment, characterized in that, The method comprises the following steps: Step S1: screening a fault diagnosis model used in the intelligent fault diagnosis process, and updating and adjusting the fault diagnosis model; Step S2: collecting real-time operation data of the offshore oil well control equipment to be fault diagnosed, and performing initial fault diagnosis on the real-time operation data using the screened fault diagnosis model used in the intelligent fault diagnosis process; Step S3: updating parameters of a digital twin model corresponding to the offshore oil well control equipment to be fault diagnosed according to a result of the initial fault diagnosis; and performing re-diagnosis on the offshore oil well control equipment to be fault diagnosed based on the digital twin model after the parameter updating. The process of screening the fault diagnosis model used in the intelligent fault diagnosis process in the step S1 specifically comprises the following steps: Step S101: selecting a plurality of fault diagnosis models as candidates, and initializing model parameters of the fault diagnosis models; Step S102: collecting operation data of the offshore oil well control equipment to be fault diagnosed in one year to generate a test set; Step S103: inputting data in the test set obtained in the step S102 into each candidate fault diagnosis model; comparing diagnosis results of each candidate fault diagnosis model with actual data labels respectively to obtain diagnosis accuracy of each candidate fault diagnosis model; Step S104: screening a fault diagnosis model with diagnosis accuracy greater than an accuracy threshold from the candidate fault diagnosis models.

2. The digital-twin-driven intelligent fault diagnosis method for offshore oil and gas well control equipment according to claim 1, characterized in that, The process of screening the fault diagnosis model used in the intelligent fault diagnosis process in the step S1 specifically further comprises the following steps: Step S105: collecting operation history data of the offshore oil well control equipment to be fault diagnosed to construct a model training set; Step S106: dividing the fault diagnosis model with the accuracy greater than the accuracy threshold screened in the step S104 into a model-based fault diagnosis model and a data-driven fault diagnosis model according to types; training the data-driven fault diagnosis model in the model training set; Step S107: calculating diagnosis accuracy of each data-driven fault diagnosis model and diagnosis accuracy of the model-based fault diagnosis model respectively; Step S108: screening a fault diagnosis model with the highest diagnosis accuracy.

3. The digital-twin-driven intelligent fault diagnosis method for offshore oil and gas well control equipment according to claim 2, characterized in that, The process of updating and adjusting the fault diagnosis model used in the intelligent fault diagnosis process in the step S1 specifically comprises the following steps: Step S109: judging whether the diagnosis accuracy of the fault diagnosis model with the highest diagnosis accuracy screened satisfies a fault diagnosis threshold requirement; if yes, selecting the fault diagnosis model with the highest diagnosis accuracy as the fault diagnosis model used in the intelligent fault diagnosis process; if no, selecting a model-based fault diagnosis model with the highest diagnosis accuracy as the fault diagnosis model used in the intelligent fault diagnosis process, adjusting a conditional probability of the model-based fault diagnosis model, and recalculating diagnosis accuracy of the model-based fault diagnosis model until the diagnosis accuracy of the model-based fault diagnosis model satisfies the fault diagnosis threshold requirement.

4. The digital-twin-driven intelligent fault diagnosis method for offshore oil and gas well control equipment according to claim 1, characterized in that, The step S3 specifically comprises the following steps: Step S301: updating parameters of a digital twin model corresponding to the offshore oil well control equipment to be fault diagnosed according to a result of the initial fault diagnosis; and Among them, the fault parameters of the digital twin model The update rule is: if If this indicates that valve CVx is in a fault state; if If x = 1, 2, ..., t, then it means that valve CVx is in a normal state; x satisfies: x = 1, 2, ..., t; Step S302: running the digital twin model to obtain the opening and closing state of each valve in the offshore oil well control equipment to be diagnosed and to update the control parameters; Wherein, the update rule of the control parameter of the digital twin model is: if the valve Cx state is open, then ; if the valve Cx state is closed, then ; if the valve Cx state is closed, then ; Step S303: obtaining the diagnosis result of the digital twin model and calculating the oil pressure value of the offshore oil well control equipment to be diagnosed; Wherein, the oil pressure value of the offshore oil well control equipment to be diagnosed meets: ; wherein, Pi-1represents the pressure of the pipeline before the ith valve; represents the density of the crude oil, which is obtained from the logging data; is the acceleration of gravity; represents the height of the center of the pipeline before the ith valve, with the wellhead as the reference height; represents the pressure of the pipeline after the ith valve; represents the height of the center of the pipeline after the ith valve, with the wellhead as the reference height; represents the pressure loss caused by the ith valve, which is given by consulting the valve test report; is the control parameter of the ith valve; Step S304: comparing the actual pressure collected from the offshore oil well control equipment to be diagnosed with the pressure calculated by the digital twin model; Wherein, the error value between the actual pressure collected by the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model satisfies: ; In the formula, an error value between the actual pressure of the i th valve outlet collected by the offshore oil well control equipment to be diagnosed for faults and the pressure calculated by the digital twin model; the actual pressure of the i th valve outlet collected by the offshore oil well control equipment to be diagnosed for faults. Step S305: determining whether the error value between the actual pressure collected from the offshore oil well control equipment to be diagnosed and the pressure calculated by the digital twin model meets the diagnosis error requirement; If yes, output the diagnosis result of the digital twin model; Step S306: if not, obtain the error diagnosis information and mark the valve number currently diagnosed as faulty as f; Step S307: further determining whether the fault diagnosis model used in the intelligent fault diagnosis process is a model-based fault diagnosis model or a data-driven fault diagnosis model; If the current diagnosis model is a model-based fault diagnosis model, set the state CVf of the valve with number f to normal and infer the fault again through the model-based fault diagnosis model; If the current diagnosis model is a data-driven diagnosis model, set the feedback state CVfs of the valve with number f to normal and infer the fault again through the data-driven diagnosis model; After the inference is completed, return to step S301 and continue to run the subsequent steps.

5. The digital twin driven intelligent fault diagnosis method for offshore oil and gas well control equipment according to claim 2, characterized in that, The model-based fault diagnosis model in step S106 has a fault diagnosis process that satisfies: ； wherein represents the probability of the xth valve being faulty given that , , the xth valve is faulty; represents the probability of the ith sensor being in the current state given that the xth valve is faulty, this value is obtained by expert decision; represents the probability of the xth valve being faulty, represents the probability of the ith sensor being in the current state given that the xth valve is not faulty; represents the probability of the xth valve not being faulty.

Images