Underwater equipment fault dependence identification method based on digital twinning
Through the digital twin-based method, the fault dependence relationship of complex underwater equipment is identified and updated in real time, and the problem of difficulty in quantifying the multi-level fault dependence relationship in the existing technology is solved, real-time tracking of the evolution trend of fault dependence is achieved, the uncertainty of fault propagation is reduced, and the accuracy of fault prediction and maintenance is improved.
Patent Information
- Application Number
- CN202411755420.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-03-25
AI Technical Summary
It is difficult for the prior art to effectively quantify and characterize the multi-level fault dependence relationship of complex underwater equipment, and track the evolution trend of fault dependence constraints in real time, resulting in greater uncertainty and randomness in the fault propagation process.
Using a digital twin method, the fault dependency relationship of underwater equipment is identified and updated in real time through target equipment requirements analysis, physical entity data acquisition and processing, multi-level fault dependency identification model modeling, fault dependency evolution model modeling under submarine environmental factors, digital twin model construction, and model virtual and real consistency verification and optimization.
Real-time identification and update of fault dependencies of complex underwater equipment is realized, which reduces the uncertainty and randomness of fault propagation and improves the accuracy and reliability of fault prediction and maintenance.
Smart Images
Figure CN119691504A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of marine engineering and relates to a method for identifying fault dependencies of complex underwater equipment based on digital twins. Background Art
[0002] Fault dependency is the inherent dependency between units and the root cause of cascading fault propagation events. Fault propagation is the manifestation of the impact of fault dependency on the system. The study of fault dependency mechanism and propagation process is the basis for preventing cascading failures in the system and ensuring its safe and stable operation and maintenance. With the development of Industry 4.0, the complexity of various systems continues to increase, and the dependence between subsystems and components within the system is increasing. In recent years, a series of numerical models have been proposed and applied to characterize the impact of fault dependency on component status, including multivariate joint distribution-based models, copula models, and degradation rate interaction models. These models simulate the random degradation process of complex systems and help understand the underlying mechanism of component degradation and failure when fault dependency exists.
[0003] In order to further study the law of cascading fault propagation caused by fault dependence, experts and scholars have adopted a variety of models to simulate the dynamic process of fault propagation, such as the sand pile model, OPA model, topological model, etc. In the field of underwater production systems, research on fault dependence has just started. At present, the models that characterize the dynamic process of fault propagation in underwater production systems include Markov model, Bayesian model, CASCADE model, etc. For the heterogeneous fault dependency mode of underwater transmission systems, the Markov model "Zhao, YX, et al., Condition-based maintenance for a multi-component system subject to heterogeneous failure dependences. Reliability Engineering & System Safety, 2023.239." is used to characterize the state transfer matrix of a multi-component system affected by heterogeneous fault dependence. Taking into account the influence of degradation interactions and uncertainties, "Shao, XY, et al., Remaining useful life prediction considering degradation interactions of subsea Christmas tree: A multi-stage modeling approach. Ocean Engineering, 2022. 264." A multi-stage model framework based on Bayesian network is proposed to describe the accelerated degradation process of underwater oil production trees and evaluate the remaining useful life (Remaining Useful Life, RUL) of the system. The CASCADE model "Zhao, YX, et al., Cascading failure analysis of multistate loading dependent systems with application in an overloading piping network. Reliability Engineering & System Safety, 2023. 231." is suitable for load-dependent systems and is used to simulate the cascading fault propagation process of underwater manifold systems subject to load-type fault dependence. The above studies have analyzed the fault dependency relationship and the fault propagation process to a certain extent, but they have simplified the quantification of the impact of fault dependency and ignored the multi-level fault dependency mapping relationship within the system.In addition, the inherent uncertainty of fault dependency and the randomness of fault propagation are still constraints to fault dependency research. The impact of complex seabed environmental factors on the evolution of fault dependency during fault propagation exacerbates this uncertainty and randomness. Therefore, the identification of fault dependency cannot be tracked and updated in real time.
[0004] Digital twins integrate multi-source data such as physical models, sensor updates, and operation history to build a virtual digital model that is highly similar to the physical system, mirroring the physical entity into the virtual space to create a digital twin. This twin can fully perceive and monitor the status and operation of the physical entity, and help track and update the evolution of the fault dependency relationship of various complex underwater equipment in real time.
[0005] How to quantitatively characterize the multi-level fault dependency of underwater complex equipment and use digital twin technology to track the evolution trend of fault dependency constrained by uncertainty in real time requires in-depth exploration. Due to the constraints of uncertainty in fault propagation, scarcity of historical data, and difficulty in controlling experimental conditions, most current studies have simplified the fault dependency between components. How to quantitatively characterize the multi-level fault dependency of underwater complex equipment is an issue that needs to be explored in depth; at the same time, seabed environmental factors and the operation of complex equipment itself will lead to dynamic changes in the status of the overall equipment and key components, which puts forward comprehensive and real-time requirements for fault dependency identification technology. How to establish a digital twin model of underwater complex equipment and track the evolution trend of fault dependency of underwater complex equipment constrained by the uncertainty of the underwater complex environment during its operation also requires more detailed and in-depth exploration. Summary of the invention
[0006] In order to overcome the defects of the prior art, the present invention provides a method for identifying fault dependencies of complex underwater equipment based on digital twins, which includes six steps: target equipment demand analysis, physical entity data collection and processing, multi-level fault dependency relationship identification model modeling, fault dependency evolution model modeling under the constraints of seabed environmental factors, digital twin model construction, and model virtual-real consistency verification and optimization.
[0007] The technical solution of the present invention:
[0008] A method for identifying underwater equipment fault dependency based on digital twins, the steps are as follows:
[0009] S1. Target equipment requirements analysis
[0010] Identify the target equipment, which are underwater oil trees, blowout preventers, and umbilical cables; analyze the working principle and structure of the target equipment; among them,
[0011] Identify key components and their functions in the target equipment. Key components include valves, pipelines, filters, connectors, and dependencies between key components, including structural dependencies and functional dependencies;
[0012] Define the operating indicators of the target equipment, including pressure, temperature, and flow, and determine the normal operating range of each key component;
[0013] Analyze the geometric features and physical characteristics of key components in the target equipment and define the boundary conditions for the normal operation of the target equipment;
[0014] Extract key failure modes and influencing factors based on actual historical failure data of target equipment and OREDA database, identify failure modes of key components through failure mode effect analysis (FMEA); use fault tree analysis (FTA) to construct fault trees of key components, common cause failures and cascading failures in target equipment, and clarify failure modes and their dependencies;
[0015] S2. Physical entity data collection and processing
[0016] Collect the physical structure and functional characteristic parameters of the target equipment, including geometric shape, functional description and performance parameters. The collected geometric shape, functional description and performance parameter data are the basic parameters of the structural layer entity and functional characteristic layer entity in the physical entity of the digital twin model; arrange sensors on the physical structure of the target equipment to collect the operating status data of the physical object in real time, including the operating status data, fault data and maintenance data of the physical object. The operating status data are the basic parameters of the operating status entity and fault layer entity in the physical entity of the digital twin model, so that the digital twin model is synchronized with the actual status of the physical structure; collect marine environment data Environmental data, including seawater temperature, pressure, flow rate, dissolved oxygen and salinity. Marine environmental data are the basic parameters of the marine environmental factor layer entity in the physical entity of the digital twin model; the basic parameters of the above physical entities are the input for building the digital twin model, pre-processing the collected status data and environmental data, removing noise, outliers and duplicate data in the data, integrating data from different sources and in different formats to form a unified data set; based on the above data, the physical entities of the digital twin model are formed, including the physical structure layer entity, functional characteristic layer entity, operating status entity, fault layer entity and marine environmental factor layer entity of the target equipment;
[0017] S3. Modeling of multi-level fault dependency identification model
[0018] Based on the identified fault modes and fault dependencies, the Bayesian network model is used to build the fault dependency model relationship model of the system. The component-level dependencies are clarified in matrix form to indicate the degree of fault dependency between components 1 to n. The dependency relationship is expressed as:
[0019]
[0020] in, Indicates that the state is x when environmental factors are not considered j Component C j For component C i The degree of fault dependence; ij is the cascade strength, indicating that the fault starts from component C j Propagate to component C i possibility; The impact degree is determined by component C j The degradation degree of is determined; According to formula (1), the heterogeneous fault dependency values with different properties and values between components are calculated;
[0021] The equipment-level fault dependency within the target equipment is represented as a matrix
[0022]
[0023] in, Indicates the components C in different states when environmental factors are not considered j For component C i The degree of fault dependence; E represents the unit matrix;
[0024] S4. Modeling of fault-dependent evolution under the constraints of seabed environmental factors
[0025] Identify key environmental factors and use a multivariate linear regression model to analyze the impact of environmental variables, including temperature, pressure, flow rate, dissolved oxygen, and salinity, on the probability of failure and failure dependency, and establish a failure dependency model subject to uncertainty:
[0026]
[0027] in, is the degree of fault dependence between components constrained by the uncertainty of environmental factors after considering environmental factors; E1,…,E n are environmental factors; β0,…,β n is the regression coefficient;∈ is the error term;
[0028] Adjust the conditional probability table of the Bayesian network model nodes according to environmental factors so that the Bayesian network model reflects the impact of environmental uncertainty factors on the fault dependency between components; for each component C i , its failure depends on the front components and environmental factors, and the fault propagation CPT is expressed as:
[0029]
[0030] Among them, P(C i =1|Parent(C i ), E) means that under the condition of environmental uncertainty factor E and given the state of the preceding component, component C i The probability of failure; Parent(C i ) represents component C i The state collection of the parent node or predecessor component; represents the degree of fault dependence under the constraints of the environmental factor set E; f is a function that maps the impact of environmental factors and the state of the preceding components to the component C. i The function of the failure probability needs to be determined according to the actual situation and available data. The Bayesian network model is used to calculate the path of fault propagation. Combined with the time series analysis ARIMA model, see formula (5), the dynamic impact of environmental factors on the failure mode and fault propagation trajectory over time is studied. A time-varying fault dependency evolution model of key components constrained by environmental uncertainty is constructed. Through simulation, the fault evolution process of underwater target equipment under different environmental factors is simulated to verify the accuracy and reliability of the fault dependency evolution model.
[0031] Y t =c+φ1Y t-1 +φ2Y t-2 +…+φ p Y t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t (5)
[0032] Among them, Y t is the sequence value at time t, c is a constant term, φ and θ are the coefficients of the autoregressive and moving average parts respectively, and ∈ is the error term;
[0033] S5. Construction of digital twin model
[0034] Based on the data set formed in step S2, the virtual entity of the digital twin model, i.e., the twin model, is constructed using modeling techniques and tools; Siemens NX is used to establish a geometric twin model of underwater complex equipment, including the structural layer and functional characteristic layer of underwater complex equipment; the environmental uncertainty factor data and the equipment operation status are integrated, and the digital twin model of the marine environment layer is established using Python tools; according to the influence of environmental uncertainty factors on the component operation status, failure mode and failure dependency, a digital twin model of the failure dependency formation and evolution over time, i.e., the operation status layer and the failure layer twin model, is established; API technology is used to complete the data transmission between different digital twin models, the physical data is mapped to the virtual entity, the physical and virtual merged data of the digital twin of the underwater complex equipment is used for fault dependency identification, and the fault dependency evolution data is mapped in real time; after completing the data mapping, the fault data in the digital twin model is subjected to real-time dynamic simulation analysis to identify the dependency and mutual influence between the faults, and to compare them with the physical data;
[0035] S6. Model virtual-real consistency verification and optimization
[0036] According to the physical entity of the target equipment and the virtual entity established in step S5, i.e., the twin model, a virtual-real consistency state reference point is proposed, a consistency state reference point matrix is developed, the virtual-real inconsistency between the physical entity and the virtual entity is described, and an inconsistency judgment rule based on the reference point matrix is established; various fault propagation scenarios that may occur in underwater complex equipment are simulated in the digital twin model, and the operation data of the physical entity is compared with the corresponding data in the digital twin model to verify the accuracy of the digital twin model in fault dependency identification; the deviation between the two is checked according to the inconsistency judgment rule, and the parameters of the digital twin model are adjusted according to the results of the virtual-real consistency verification, and the parameters of the digital twin model are iteratively optimized through the optimization algorithm; if it is found that the digital twin model has structural deficiencies, the structure of the digital twin model is improved; the actual data and the model data are fused through data fusion technology to improve the real-time performance and accuracy of the model.
[0037] Beneficial effects of the invention: The underwater equipment fault dependency identification method based on digital twin proposed in the invention can effectively identify and update the key fault modes and main fault dependencies of underwater equipment in real time, provide information reference for subsequent fault prediction and maintenance of underwater equipment, avoid fault escalation, and reduce maintenance costs. Through virtual-real consistency verification, the accuracy and reliability of the underwater equipment digital twin model are improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 This is the basic flow chart of the fault dependency identification method for complex underwater equipment based on digital twins.
[0039] Figure 2 This is an illustration of the fault dependency matrix. DETAILED DESCRIPTION
[0040] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.
[0041] A method for identifying underwater equipment fault dependency based on digital twins, the steps are as follows:
[0042] S1. Target equipment requirements analysis
[0043] Identify the target underwater equipment, taking the underwater blowout preventer as an example;
[0044] Analyze the working principle and structure of the target equipment. The underwater blowout preventer is a key safety device for underwater wellheads, used to control the wellhead pressure during drilling or oil production to prevent blowout accidents. Its main processes include closing the wellhead, sealing the wellbore, and controlling the wellbore pressure;
[0045] The key components and functions of the underwater blowout preventer are: the gate assembly, which is used to close the wellhead to prevent fluid from blowing out; the piston assembly, which is used to drive the gate assembly to close or open; the control system, which is used to receive ground instructions and control the action of the piston assembly; the sealing assembly, which is used to ensure the internal sealing of the blowout preventer to prevent fluid leakage;
[0046] Identify the dependencies between components in the system. The gate assembly relies on the piston assembly to provide power to close or open. The control system relies on sensor feedback and ground commands to control the piston assembly. The sealing assembly relies on the correct closure of the gate assembly to maintain the seal. The control system relies on sensor data to make control decisions.
[0047] Define the operating indicators of the identification system, including the pressure in the wellbore, the temperature inside the blowout preventer, and the fluid flow through the blowout preventer, determine the normal range of each operating indicator, and define the boundary conditions for the normal operation of the identification system;
[0048] Extract key failure modes and influencing factors based on the actual historical failure data of the identification system and the OREDA database, and identify the failure modes of key components through failure mode impact analysis FMEA, including gate blocking, piston seal failure, sensor failure, control command transmission failure and other failure modes; construct a fault tree for the gate assembly, piston assembly, control system and sealing components, identify common cause failures and cascading failures, such as control system failure causing the piston assembly to fail to move, which in turn causes the gate assembly to fail to close, and clarify the failure modes and their dependencies;
[0049] S2. Physical entity data collection and processing
[0050] Sensors are arranged on the blowout preventer to collect data such as pressure, temperature, and fluid flow, which are used to collect the status data of the physical object in real time, and to update the digital twin model so that the digital twin model is synchronized with the actual status of the physical object, including the operation data, fault data, and maintenance data of the physical object. The fault data includes the fault type, fault time, and fault cause. Sensors are arranged to collect relevant parameters of the marine environment at the same time, including seawater temperature, pressure, flow rate, dissolved oxygen, and salinity, remove noise, outliers, and duplicate data in the data, ensure the accuracy and completeness of the data, pre-process the collected status data and environmental data, and integrate data from different sources and formats to form a unified data set. Based on the above data, the physical entities of the digital twin model mainly include the structural layer entities, functional layer entities, operation state entities, fault layer entities, and environmental factor layer entities of the target equipment.
[0051] S3. Modeling of multi-level fault dependency identification model
[0052] Based on the identified fault modes and fault dependencies, the Bayesian network is used to build the fault dependency model relationship model of the system. The component-level dependency is clarified in the form of a matrix to indicate the degree of fault dependency between component 1 and component 2. The dependency is expressed as:
[0053]
[0054] Among them, the state of the component is x i ∈{0, 1, 2, 3}, 0 is a brand new state, 3 is an invalid state; represents the failure dependence of component C2 in state x2 on component C1 when environmental factors are not considered; γ 12 is the cascade strength, which indicates the possibility of fault propagation from component C2 to component C1. The value of cascade strength is between 0 and 1. is the degree of influence, which is determined by the degradation degree of component C2. When the state of component C2 is x2=0, it has no influence on other components. When the state of component C2 is x2=3, the degree of influence is 1. According to formula (1), the heterogeneous fault dependency values with different properties and different values between components are calculated.
[0055] According to the fault dependency between components, the fault dependency relationship of the equipment level in the underwater blowout preventer without considering environmental factors is calculated and expressed as a matrix
[0056]
[0057] in, Indicates the components C in different states when environmental factors are not considered j For component C i The degree of fault dependence;
[0058] S4. Modeling of fault-dependent evolution under the constraints of seabed environmental factors
[0059] Identify key environmental factors and use a multivariate linear regression model to analyze the impact of environmental variables, including temperature, pressure, flow rate, dissolved oxygen, and salinity, on the probability of failure and failure dependency, and establish a failure dependency model subject to uncertainty:
[0060]
[0061] in, is the degree of fault dependence between components constrained by the uncertainty of environmental factors after considering environmental factors; E1,…,E5 are environmental factors, representing the impact of temperature, pressure, flow rate, dissolved oxygen, and salt content on component C, respectively. i and component C j The influence of fault dependencies between n is the regression coefficient;∈ is the error term;
[0062] Adjust the conditional probability table of the Bayesian network nodes according to environmental factors so that the Bayesian network model can reflect the impact of environmental factors on the fault dependency between components; for each component C i , its failure depends on the front components and environmental factors, and the fault propagation CPT is expressed as:
[0063]
[0064] Among them, P(C i =1|Parent(C i ), E) means that under the condition of environmental uncertainty factor E and given the state of the preceding component, component C i The probability of failure; Parent(C i ) represents component C i The state collection of the parent node or predecessor component; represents the degree of fault dependence under the constraints of the environmental factor set E; f is a function that maps the impact of environmental factors and the state of the preceding components to the component C. i The function of the failure probability needs to be determined according to the actual situation and available data. The path of fault propagation is calculated through the Bayesian network. Combined with the time series analysis ARIMA model, see formula (5), the dynamic influence of environmental factors on the failure mode and fault propagation trajectory over time is studied, and a time-varying failure dependency evolution model of key components affected by the environment is constructed. Through simulation, the failure evolution process of the underwater blowout preventer under different environmental factors is simulated to verify the accuracy and reliability of the model.
[0065] Y t =c+φ1Yt-1 +φ2Y t-2 +…+φ p Y t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t (10)
[0066] Among them, Y t is the sequence value at time t, c is a constant term, φ and θ are the coefficients of the autoregressive and moving average parts respectively, and ∈ is the error term;
[0067] S5. Construction of digital twin model
[0068] Based on the data collected by S2, the prototype of the digital twin model of the underwater blowout preventer is built using modeling technology and tools; Siemens NX is used to establish a geometric twin model of the underwater blowout preventer, including the dependency chain of the underwater blowout preventer structure layer, function layer, and fault layer; the environmental factor data and the operating status of the underwater blowout preventer are integrated, and the Python tool is used to establish a twin model of the marine environment. The twin model of the marine environment affects the underwater blowout preventer system status layer and fault layer twin model through the Bayesian network; according to the influence of environmental uncertainty factors on the operating status, failure mode and failure dependency of key components inside the underwater blowout preventer, a twin model of fault dependency formation and evolution over time is established; API technology is used to study the data between different digital twin models The transmission method and the mapping method of physical data to virtual entities are used to identify the fault dependency relationship of the physical and virtual merged data of the underwater blowout preventer digital twin, and the fault dependency evolution data is mapped in real time; the fault data in the digital twin model is simulated and analyzed in real time to identify the dependency relationship and mutual influence between faults, which is achieved through data analysis technology and compared with physical data; the fault evolution process of the blowout preventer under different environmental conditions is simulated through the digital twin model, and the key failure modes such as gate seal failure and piston sticking and their dependencies are identified, and the impact of environmental factors on the probability of failure and fault dependency is analyzed;
[0069] S6. Model virtual-real consistency verification and optimization
[0070] According to the state parameters of the physical and virtual entities of the underwater blowout preventer, a virtual-real consistency state reference point is proposed, a consistency reference point matrix is developed, the virtual-real inconsistency between the physical and virtual entities is described, and an inconsistency judgment rule based on the reference point matrix is established; various fault propagation scenarios that may occur in complex underwater equipment are simulated in the digital twin model, and the physical entity operation data is compared with the corresponding data in the digital twin model to verify the accuracy of the digital twin model in fault dependency identification; the deviation between the two is checked according to the inconsistency judgment rule, and the parameters of the digital twin model are adjusted according to the results of the virtual-real consistency verification, and the model parameters are iteratively optimized through the optimization algorithm to improve the accuracy of the model; if the digital twin model is found to have structural deficiencies, the model structure is improved; the actual data is fused with the model data through data fusion technology to improve the real-time and accuracy of the model.
[0071] The underwater equipment fault dependency identification method based on digital twin proposed in this paper can effectively identify and analyze the failure mode and dependency of underwater blowout preventers. Through simulation and virtual-real consistency verification, the accuracy and reliability of the model are improved. This method can be applied to other underwater equipment, such as underwater Christmas trees, umbilical cables, etc., to provide strong support for fault prediction and maintenance of underwater equipment.
Claims
1. A method for identifying underwater equipment fault dependency based on digital twins, characterized in that: Here are the steps: S1. Target equipment requirements analysis Identify the target equipment, which are underwater Christmas trees, blowout preventers, and umbilical cables; analyze the working principle and structure of the target equipment; among them, Identify key components and their functions in the target equipment. Key components include valves, pipelines, filters, connectors, and dependencies between key components, including structural dependencies and functional dependencies; Define the operating indicators of the target equipment, including pressure, temperature, and flow, and determine the normal operating range of each key component; Analyze the geometric features and physical characteristics of key components in the target equipment and define the boundary conditions for the normal operation of the target equipment; Extract key failure modes and influencing factors based on actual historical failure data of target equipment and OREDA database, identify failure modes of key components through failure mode effect analysis (FMEA); use fault tree analysis (FTA) to construct fault trees of key components, common cause failures and cascading failures in target equipment, and clarify failure modes and their dependencies; S2. Physical entity data collection and processing Collect the physical structure and functional characteristic parameters of the target equipment, including geometric shape, functional description and performance parameters. The collected geometric shape, functional description and performance parameter data are the basic parameters of the structural layer entity and functional characteristic layer entity in the physical entity of the digital twin model; arrange sensors on the physical structure of the target equipment to collect the operating status data of the physical object in real time, including the operating status data, fault data and maintenance data of the physical object. The operating status data are the basic parameters of the operating status entity and fault layer entity in the physical entity of the digital twin model, so that the digital twin model is synchronized with the actual status of the physical structure; collect marine environment data Environmental data, including seawater temperature, pressure, flow rate, dissolved oxygen and salinity. Marine environmental data are the basic parameters of the marine environmental factor layer entity in the physical entity of the digital twin model; the basic parameters of the above physical entities are the input for building the digital twin model, pre-processing the collected status data and environmental data, removing noise, outliers and duplicate data in the data, integrating data from different sources and in different formats to form a unified data set; based on the above data, the physical entities of the digital twin model are formed, including the physical structure layer entity, functional characteristic layer entity, operating status entity, fault layer entity and marine environmental factor layer entity of the target equipment; S3. Modeling of multi-level fault dependency identification model Based on the identified fault modes and fault dependencies, the Bayesian network model is used to build the fault dependency model relationship model of the system. The component-level dependencies are clarified in matrix form to indicate the degree of fault dependency between components 1 to n. The dependency relationship is expressed as: in, Indicates that the state is x when environmental factors are not considered j Component C j For component C i The degree of fault dependence; ij is the cascade strength, indicating that the fault starts from component C j Propagate to component C i possibility; The impact degree is determined by component C j The degradation degree of is determined; According to formula (1), the heterogeneous fault dependency values with different properties and values between components are calculated; The equipment-level fault dependency within the target equipment is represented as a matrix in, Indicates the components C in different states when environmental factors are not considered j For component C i The degree of fault dependence; E represents the unit matrix; S4. Modeling of fault-dependent evolution under the constraints of seabed environmental factors Identify key environmental factors and use a multivariate linear regression model to analyze the impact of environmental variables, including temperature, pressure, flow rate, dissolved oxygen, and salinity, on the probability of failure and failure dependency, and establish a failure dependency model subject to uncertainty: in, is the degree of fault dependence between components constrained by the uncertainty of environmental factors after considering environmental factors; E1,…,E n are environmental factors; β0,…,β n is the regression coefficient;∈ is the error term; Adjust the conditional probability table of the Bayesian network model nodes according to environmental factors so that the Bayesian network model reflects the impact of environmental uncertainty factors on the fault dependency between components; for each component C i , its failure depends on the front components and environmental factors, and the fault propagation CPT is expressed as: Among them, P(C i =1|Parent(C i ), E) means that under the condition of environmental uncertainty factor E and given the state of the preceding component, component C i The probability of failure; Parent(C i ) represents component C i The state collection of the parent node or predecessor component; represents the degree of fault dependence under the constraints of the environmental factor set E; f is a function that maps the impact of environmental factors and the state of the preceding components to the component C. i The function of the failure probability needs to be determined according to the actual situation and available data. The Bayesian network model is used to calculate the path of fault propagation. Combined with the time series analysis ARIMA model, see formula (5), the dynamic impact of environmental factors on the failure mode and fault propagation trajectory over time is studied. A time-varying fault dependency evolution model of key components constrained by environmental uncertainty is constructed. Through simulation, the fault evolution process of underwater target equipment under different environmental factors is simulated to verify the accuracy and reliability of the fault dependency evolution model. Y t =c+φ1Y t-1 +φ2Y t-2 +…+φ p Y t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t (5) Among them, Y t is the sequence value at time t, c is a constant term, φ and θ are the coefficients of the autoregressive and moving average parts respectively, and ∈ is the error term; S5. Construction of digital twin model Based on the data set formed in step S2, the virtual entity of the digital twin model, i.e., the twin model, is constructed using modeling techniques and tools; Siemens NX is used to establish a geometric twin model of underwater complex equipment, including the structural layer and functional characteristic layer of underwater complex equipment; the environmental uncertainty factor data and the equipment operation status are integrated, and the digital twin model of the marine environment layer is established using Python tools; according to the influence of environmental uncertainty factors on the component operation status, failure mode and failure dependency, a digital twin model of the failure dependency formation and evolution over time, i.e., the operation status layer and the failure layer twin model, is established; API technology is used to complete the data transmission between different digital twin models, the physical data is mapped to the virtual entity, the physical and virtual merged data of the digital twin of the underwater complex equipment is used for fault dependency identification, and the fault dependency evolution data is mapped in real time; after completing the data mapping, the fault data in the digital twin model is subjected to real-time dynamic simulation analysis to identify the dependency and mutual influence between the faults, and to compare them with the physical data; S6. Model virtual-real consistency verification and optimization According to the physical entity of the target equipment and the virtual entity established in step S5, i.e., the twin model, a virtual-real consistency state reference point is proposed, a consistency state reference point matrix is developed, the virtual-real inconsistency between the physical entity and the virtual entity is described, and an inconsistency judgment rule based on the reference point matrix is established; various fault propagation scenarios that may occur in underwater complex equipment are simulated in the digital twin model, and the operation data of the physical entity is compared with the corresponding data in the digital twin model to verify the accuracy of the digital twin model in fault dependency identification; the deviation between the two is checked according to the inconsistency judgment rule, and the parameters of the digital twin model are adjusted according to the results of the virtual-real consistency verification, and the parameters of the digital twin model are iteratively optimized through the optimization algorithm; if it is found that the digital twin model has structural deficiencies, the structure of the digital twin model is improved; the actual data and the model data are fused through data fusion technology to improve the real-time performance and accuracy of the model.