A fault mode risk grading method and system for an offshore oil underwater electric control tree

Abstract

A fault mode risk grading method and system of an offshore oil underwater electric control tree belong to the technical field of safety monitoring and intelligent operation and maintenance of offshore oil underwater electric control trees. The control behavior, state feedback, barrier function and failure consequences of the underwater electric control tree are structurally associated, and multi-source objective evidence is fused for risk perception, weight calculation, fault mode disposal priority generation and risk grading output, so as to improve the integrity of underwater electric control tree fault mode risk identification, the objectivity of risk evaluation and the engineering implementability of grading results.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring and intelligent operation and maintenance technology for subsea electrical control trees in marine oil and gas, specifically to a fault mode risk classification method and system for subsea electrical control trees in marine oil and gas. Background Technology

[0002] Subsea electrical control trees (EDMs) are key equipment in subsea production systems for deepwater oil and gas development, responsible for wellhead production control, annular control, chemical injection, pressure monitoring, hydraulic actuation, remote shutdown, and safety isolation. Subsea EDMs typically operate in high-pressure, low-temperature, highly corrosive environments, requiring long-distance remote control and presenting significant maintenance challenges. Their valve operation, hydraulic pressure supply, communication feedback, pressure maintenance, and emergency shutdown functions directly impact wellhead safety, production continuity, and marine environmental safety.

[0003] During the operation of an underwater electronically controlled tree (ECMA-MAS), faults do not always manifest as the failure of a single device. For example, control commands may be issued but the device fails to respond correctly, valve position feedback may appear normal but pressure holding capacity may be insufficient, status feedback may be delayed, frozen, or drifting, or there may be timing anomalies in the emergency shutdown process. All of these can lead to the barrier function not being effectively verified. Such faults often involve the interrelationships between control behavior, the device being executed, status feedback, barrier function, and the consequences of failure. If judgments are made based solely on a single alarm signal, a single device status, or human experience, it is easy to result in incomplete identification of barrier degradation faults, unclear explanation of risk sources, and ambiguous identification of the appropriate action.

[0004] Existing methods for fault analysis and risk assessment of underwater production systems typically employ fault mode and impact analysis (FMEA), risk matrices, expert scoring, alarm statistics, or single monitoring data analysis to identify and rank equipment faults. While these methods are effective in general equipment risk assessment, their application to fault mode risk classification in underwater electrical control trees (ECLS) suffers from the following shortcomings: First, fault modes are often recorded as equipment anomaly names or alarm events, lacking a structured correlation with control behavior, status feedback, barrier functions, and failure consequences, making it difficult to identify hidden faults where control execution is normal but barrier functions cannot be effectively verified. Second, risk assessment results tend to rely on a single source of evidence or subjective weighting, making it difficult to simultaneously utilize operational monitoring evidence, functional verification evidence, and fault handling evidence, leading to data gaps, evidence deviations, or isolated false alarms that significantly impact the assessment results. Third, assessment results are often expressed as risk scores or rankings, lacking outputs on risk level boundaries, dominant risk factors, and subsequent verification and handling signals, making it difficult to transform risk assessment results into technical outputs for barrier verification, control feedback verification, and remote recovery review.

[0005] Therefore, it is necessary to propose a failure mode risk classification method and system for underwater electronic control trees (ECUs). This method would structurally correlate the control behavior, state feedback, barrier functions, and failure consequences of the ECU, and integrate multi-source objective evidence for risk perception, weight calculation, failure mode handling priority generation, and risk classification output. This would improve the completeness of ECU failure mode risk identification, the objectivity of risk assessment, and the engineering feasibility of the classification results. Summary of the Invention

[0006] The present invention aims to address the above-mentioned problems by providing a method and system for fault mode risk classification of marine oil subsea electrical control trees.

[0007] The technical solution of the present invention: A fault mode risk classification method for subsea electrical control trees in marine oilfields, such as Figure 1 As shown, it includes the following steps: S1: Construct an underwater electrical control tree control-feedback-barrier mapping model.

[0008] Identify the structured relationships between control behaviors, execution objects, safety constraints, status feedback, barrier functions, and failure consequences involved in the operation of the underwater electronic control tree, and form an underwater electronic control tree control-feedback-barrier mapping model.

[0009] The control behavior refers to the operational behavior issued by the control system during the operation of the underwater electrical control tree, including one or more of the following: production control, annular control, chemical injection control, hydraulic control, and emergency shutdown control; the execution object refers to the equipment or circuit directly affected by the control behavior, including one or more of the following: production main valve, production wing valve, annular valve, chemical injection valve, and hydraulic control circuit.

[0010] The safety constraints refer to the control requirements, response requirements, or status requirements that must be met under the corresponding operating conditions to ensure the effectiveness of the underwater electrically controlled tree barrier function. The safety constraints include one or more of the following: the valve should act according to the preset control command; the valve should generate corresponding status feedback after acting; the pressure should be maintained after shutting down or isolating; and the barrier should not be confirmed as effective based on a single feedback signal when feedback is abnormal.

[0011] The status feedback refers to the operating status information returned by the underwater electronic control tree after the control behavior is executed, including one or more of the following: valve position status, pressure parameters, temperature parameters, flow parameters, hydraulic parameters, communication status, and alarm information. The status feedback is used to determine whether the control behavior has been executed correctly and to form an objective basis for verifying the effectiveness of the barrier function.

[0012] The barrier function refers to the function of the subsea electrical control tree in maintaining wellhead safety, production isolation, annular isolation, pressure maintenance, injection isolation, flow regulation, or emergency shutdown; the failure consequences refer to the consequences caused by the failure of the barrier function or its inability to be effectively verified, including one or more of the following: unexpected release of wellhead fluid, oil and gas leakage, runaway production pressure, runaway annular pressure, abnormal discharge of chemical agents, inability to use remote shutdown, equipment damage, threat to personnel safety, and marine environmental pollution.

[0013] Through the above processing, each control action is made to correspond to its execution object, safety constraints, status feedback, barrier function, and failure consequences.

[0014] S2. Identify fault modes based on the underwater electrical control tree control-feedback-barrier mapping model.

[0015] Based on the underwater electrical control tree control-feedback-barrier mapping model formed in step S1, the fault modes that can lead to abnormal barrier function or failure during the operation of the underwater electrical control tree are identified, forming a set of fault modes with mapping attributes.

[0016] Specifically, for each control action, fault modes are identified from four aspects: control action execution status, control action timing, status feedback effectiveness, and barrier function verification results, forming a fault mode set of the underwater electronic control tree.

[0017] Furthermore, when a control action is not executed, the execution object is incorrect, or the execution result does not meet safety constraints, it is identified as a control execution type failure mode; when a control action is executed too early, too late, in the wrong order, or the duration does not meet the preset requirements, it is identified as a control timing type failure mode; when status feedback is missing, delayed, frozen, drifting, abrupt, or inconsistent with the expected response, it is identified as a feedback anomaly type failure mode; when a control action has been executed and the status feedback shows normal, but it cannot prove that the corresponding barrier function is effective, it is identified as a barrier function degradation type failure mode.

[0018] Furthermore, for each identified fault mode, its corresponding control behavior, execution object, safety constraints, status feedback, barrier function, and failure consequences are retained to form a set of fault modes with mapping attributes.

[0019] S3. Construct a multi-source risk perception matrix for underwater electrical control trees.

[0020] Based on the fault mode set of the underwater electronic control tree formed in step S2, the operation monitoring risk perception matrix, the function verification risk perception matrix, and the fault handling risk perception matrix are generated from three evidence sources: operation monitoring evidence, function verification evidence, and fault handling evidence, respectively, forming a multi-source risk perception matrix.

[0021] The operational monitoring evidence is used to characterize the abnormal performance of the fault mode during actual operation; the functional verification evidence is used to characterize the abnormal performance of the fault mode during functional verification; and the fault handling evidence is used to characterize the handling performance of the fault mode during alarm handling, remote reset, valve retesting, maintenance and replacement, production shutdown recovery, or external intervention.

[0022] For each source of evidence, a risk evidence value is generated according to the same risk factor dimensions; the risk factors include abnormal exposure, barrier impact, detection difficulty, recovery difficulty, and consequence severity.

[0023] Furthermore, the original records from each source of evidence are converted into risk evidence values ​​of a uniform dimension according to a preset evidence level; the preset evidence levels include no evidence, weak evidence, moderate evidence, strong evidence, and invalid evidence, and the higher the evidence level, the greater the corresponding risk evidence value.

[0024] A risk perception matrix is ​​constructed with fault modes as rows and risk factors as columns for each evidence source: operation monitoring risk perception matrix, functional verification risk perception matrix, and fault handling risk perception matrix. These three risk perception matrices constitute a multi-source risk perception matrix.

[0025] S4. Generate evidence source weights and integrate them into the risk perception matrix.

[0026] Based on the operation monitoring risk perception matrix, functional verification risk perception matrix, and fault handling risk perception matrix formed in step S3, the weights of each evidence source are determined by combining data feature dimensionality reduction and consensus support. The three risk perception matrices are then fused to obtain a comprehensive risk perception matrix.

[0027] Assume the number of fault modes of the underwater electrical control tree is . m The number of risk factors is n The number of evidence sources is ,in . No. k The risk perception matrix corresponding to each source of evidence is represented as follows: ,in, Indicates the first k The first source of evidence i The failure mode is in the first j Risk evidence values ​​for each risk factor, and After processing in step S3, it is converted into a numerical value with uniform dimensions.

[0028] First, a data feature vector is constructed for each source of evidence. The data feature vector includes one or more of the following: data completeness rate, effective evidence coverage rate, and risk discrimination ratio. Specifically, the data completeness rate characterizes the proportion of non-missing risk evidence values ​​in the source of evidence; the effective evidence coverage rate characterizes the proportion of units in the source of evidence that can provide effective risk evidence; and the risk discrimination ratio characterizes the ability of the source of evidence to distinguish the differences in risk between different failure modes.

[0029] For the k Each source of evidence, its data integrity rate Valid evidence coverage and risk differentiation They are respectively , , .in, For the first k The number of non-missing units in each risk perception matrix For the first k The number of units in a risk perception matrix whose risk evidence value is greater than a preset valid evidence threshold. For the first k The first source of evidence j The average risk evidence value of each risk factor.

[0030] according to , and Constructing the feature matrix of evidence sources , No. k The line represents the first k The data feature vectors corresponding to each source of evidence. The feature matrix of the evidence sources. F Centralized processing, i.e. F Subtracting the average of the data in each column from the average of the data in the matrix yields the centered matrix. For the centralized matrix Standardization processing, i.e. Divide each data point by the standard deviation of the data in its corresponding column to obtain the standardized matrix. Based on the standardized matrix Constructing the covariance matrix For the covariance matrix Perform eigenvalue decomposition to obtain eigenvalues. and corresponding feature vectors ,in t Indicates the principal component index. Sort by eigenvalues ​​from largest to smallest, and select the cumulative contribution rate. Before reaching the preset threshold q Principal components, The cumulative contribution rate threshold is preferably 85%–95%. For the first... kFor each source of evidence, calculate its projection value along the directions of each principal component. ,in Represents the normalized matrix The first in k Row vectors. Further, the cumulative contribution rate of each principal component is used to calculate the first row vector. k Data feature scores of each source of evidence To avoid negative values ​​affecting subsequent weight allocation, ... Perform nonnegation processing. Feature scores of the data after nonnegation processing. Normalization is performed to obtain the data feature weights. .

[0031] Then, the consensus support for the evidence sources is calculated based on the similarity among the three risk perception matrices. For the first... k The first source of evidence and the first r Given several sources of evidence, calculate the matrix distance between them. Convert matrix distance into consensus support. ,in This represents the maximum value of the pairwise matrix distances among all evidence sources. For the... k Each source of evidence is used to calculate its total consensus support. and will Normalization, gaining consensus support for weights .

[0032] Then, a fusion coefficient was introduced. l Weighting data features and consensus support weights By merging, we obtain the first... k Weight of each source of evidence .in, When greater emphasis is placed on the data integrity, coverage, and discriminative power of the evidence sources themselves, improving... l Value; when greater emphasis is placed on consistency among sources of evidence, reduce l Values.

[0033] Finally, based on the weight of the evidence source The three risk perception matrices are weighted and fused together, and then... Fusion i The failure mode is in the first j The comprehensive risk perception matrix is ​​obtained by combining the risk evidence values ​​of each risk factor. .

[0034] Through the above processing, the obtained comprehensive risk perception matrix takes into account the data quality, risk differentiation ability, and consensus support among different evidence sources. It can reduce the impact of missing evidence sources, abnormal deviations, or isolated false alarms on the failure mode risk assessment results, and provide a unified input for subsequent risk factor weights, failure mode handling priority sequence generation, and risk classification.

[0035] S5. Generate risk factor weights based on the comprehensive risk perception matrix.

[0036] Based on the comprehensive risk perception matrix obtained in step S4, the weights of each risk factor are calculated using a risk factor weighting method based on information entropy characteristics, logarithmic bias characteristics, and criterion removal effects.

[0037] First, the comprehensive risk perception matrix X The risk factor columns in the data are normalized, i.e. The normalized matrix is ​​obtained. Normalized Used to characterize the i The failure mode is in the first j The relative strength of risk evidence under each risk factor, with a larger value indicating stronger risk evidence.

[0038] Then, the information entropy feature of each risk factor is calculated. For the first... j First, calculate the proportion of each risk factor in each failure mode. Then calculate the information entropy value of the risk factor. To obtain information difference ,in The larger the value, the higher the value. j The stronger the distinguishing information that a risk factor provides between different failure modes, the better.

[0039] Furthermore, the log-bias characteristics of each risk factor are calculated. For the j Each risk factor was used to calculate its average relative risk evidence strength. And obtain the logarithmic deviation value. ,in The larger the value, the higher the value. j The more significant the deviation of a risk factor between different failure modes, the stronger its ability to characterize the differences in failure mode risk.

[0040] Furthermore, the criterion removal effect for each risk factor is calculated. First, the effect of all risk factors combined is calculated. i Average risk structure value of each failure mode Then calculate the removal of the first j After the first risk factor i Average risk structure value of each failure mode , Indicates the first i The failure mode is in the first The relative strength of risk evidence under each risk factor, where the first... The risk factor belongs to the category excluding the removed one. j Other risk factors besides the first risk factor; thus, the first risk factor is obtained. j Criterion impact value of each risk factor ,in The larger the value, the more likely it is to remove the first element. j The more significant the change in the overall risk structure of the failure mode after a risk factor, the greater the contribution of that risk factor to the information of the comprehensive risk perception matrix.

[0041] Subsequently, the degree of information difference Logarithmic deviation value and the influence value of the criteria Normalization is performed. , , And calculate the first j The combined information contribution value of each risk factor ,in a , b , c The preset adjustment coefficient satisfies a ≥0、 b ≥0、 c ≥0 and a + b + c =1. When more emphasis is placed on the overall discrete information of risk factors, improve... a Value selection; when more emphasis is placed on the local deviation characteristics of risk factors, increase b Value selection; when more emphasis is placed on the impact of risk factors on the overall risk structure, increase c Values.

[0042] Finally, the overall information contribution value of each risk factor is normalized to obtain the risk factor weights. .

[0043] Through the above processing, the distribution dispersion, logarithmic deviation and criterion influence of risk factors can be extracted simultaneously from the comprehensive risk perception matrix. This ensures that the obtained risk factor weights do not depend on subjective or arbitrary assignment, but are determined by the objective contribution of each risk factor to the risk differences of underwater electronic control tree failure modes. This provides risk factor weight input for subsequent failure mode handling priority sequences and risk classifications.

[0044] S6. Generate a fault mode handling priority sequence based on the bidirectional comprehensive risk dominance value.

[0045] Based on the comprehensive risk perception matrix obtained in step S4 and the risk factor weights obtained in step S5, the bidirectional comprehensive risk dominance value among the various failure modes of the underwater electronic control tree is calculated to form a risk dominance matrix, and a failure mode handling priority sequence is generated accordingly.

[0046] First, based on the comprehensive risk perception matrix Perform a bidirectional comparison for any two fault modes. For the first fault mode... j The nth risk factor is calculated. i The failure mode is relative to the first h Local risk dominance of each failure mode Then, combined with risk factor weights. Calculate the first i The failure mode is relative to the first h Overall risk dominance of each failure mode Similarly, calculate the first... h The failure mode is relative to the first i Overall risk dominance of each failure mode Based on the two-way comprehensive risk dominance value, the first... i The failure mode is relative to the first h Risk dominance probability index for each failure mode .in, Indicates the first i The failure mode is relative to the first h Each failure mode has a higher risk dominance relationship. Indicates the first h The failure mode is relative to the first i Each failure mode has a higher risk dominance relationship. This indicates that the risk dominance relationship is the same for both. A risk dominance matrix is ​​formed by pairwise comparisons of all failure modes. diagonal elements Take 0.5.

[0047] Subsequently, a recursive baseline partitioning is performed based on the risk dominance matrix to generate a priority sequence for fault mode handling. Specifically, a fault mode is selected from the set of fault modes to be sorted as the baseline fault mode, and its index is denoted as [index missing]. p The remaining failure modes are then divided into high-risk dominant subsets, approximately equal subsets, and low-risk dominant subsets based on their risk dominance relationship with the baseline failure mode. , will the h Each failure mode is classified into the high-risk dominant subset; when , will the h Each failure mode is classified into the low-risk dominant subset; when , will the h Each failure mode is assigned to an approximately equal subset; among them... t This is the preset tolerance threshold.

[0048] Continue to perform baseline failure mode selection and recursive partitioning on the high-risk dominant subset and the low-risk dominant subset until the number of failure modes in each subset is no greater than 1, and obtain the failure mode handling priority sequence. The failure modes in the failure mode handling priority sequence are arranged from high to low according to the risk handling priority.

[0049] Through the above processing, a fault mode handling priority sequence is output, providing input for subsequent risk classification and differentiated handling strategy generation.

[0050] S7. Implement fault mode risk classification for underwater electrical control trees.

[0051] Based on the comprehensive risk perception matrix obtained in step S4, the risk factor weights obtained in step S5, the risk dominance matrix obtained in step S6, and the failure mode handling priority sequence, the failure modes of the underwater electrical control tree are classified by risk, and the risk level, candidate risk level boundary, level risk index and dominant risk factor corresponding to each failure mode are output.

[0052] First, a risk structure feature vector is constructed for each failure mode. For the th... i One failure mode, based on a comprehensive risk perception matrix and risk factor weights Calculate its overall risk intensity and dominant contribution value The dominant contribution value is then normalized to obtain the contribution vector of key risk factors. Then, based on the risk dominance matrix... Calculate the first i Net dominance of each failure mode The dominant intensity feature is then normalized to obtain the dominant intensity feature. .

[0053] Combining comprehensive risk intensity Dominant intensity characteristics and key risk factor contribution vector , construct the first i Risk structure feature vector of each failure mode .

[0054] Then, the risk structure separation degree between adjacent failure modes is calculated along the failure mode handling priority sequence. For adjacent failure modes in the failure mode handling priority sequence, the order is represented as follows: and Calculate the risk structure separation degree between the two. .

[0055] Furthermore, candidate risk level boundaries are determined based on the separation degree of all adjacent risk structures. The average risk structure separation degree is calculated. and standard deviation When the first The adjacent positions satisfy When this occurs, the adjacent location is determined as the boundary of the candidate risk level, where r This is the preset boundary sensitivity coefficient.

[0056] Based on the candidate risk level boundaries, the candidate classification schemes are obtained as follows: ,in Indicates the first s The set of failure modes in each risk level g This indicates the number of risk levels. Calculate the intra-level compactness of the candidate grading scheme. ,in It belongs to the first s The mean of the risk structure feature vectors of failure modes for each risk level. Calculate the inter-level separation of the candidate grading scheme. ,in It is the mean of the risk structure feature vectors of all failure modes across all risk levels. Indicates the first s The number of failure modes in each risk level. Further calculation of the classification effectiveness. ,in x This is a penalty coefficient for the number of risk levels, used to prevent risk levels from being overly subdivided. e To prevent extremely small positive numbers with a denominator of zero, choose to... The largest grading scheme is used as the target risk grading scheme. .

[0057] Based on the target risk classification scheme, the priority sequence of failure mode handling is divided into several consecutive risk level intervals, and risk levels are assigned in descending order.

[0058] Furthermore, a risk index is calculated for each risk level. For the first... s There are several risk levels, and the risk index for each level is: ,in, Indicates the first i Each failure mode; the risk level index is used to characterize the overall risk intensity and overall dominance of the risk level.

[0059] Further, calculate the first s The dominant risk factor for each risk level This indicates the risk factor that plays a major role in shaping the risk level.

[0060] Based on the target risk classification scheme, risk index, and dominant risk factors, the risk level corresponding to each failure mode is output. The risk level includes two or more of the following: safety and environmental risks, barrier function degradation risks, control and feedback anomaly risks, and general monitoring concerns.

[0061] Through the above processing, S7 not only generates the risk level of each failure mode, but also generates candidate risk level boundaries, level risk indices, and dominant risk factors. The candidate risk level boundaries are used to represent the structural differences between different risk levels, the level risk indices are used to characterize the overall risk level of each risk level, and the dominant risk factors are used to explain the main reasons for the formation of the corresponding risk level and provide a basis for subsequent differentiated treatment recommendations.

[0062] S8. Output risk classification results and differentiated verification and handling signals.

[0063] Based on the failure mode handling priority sequence obtained in step S6, and the risk level, level risk index and dominant risk factor corresponding to the failure mode obtained in step S7, the underwater electrical control tree failure mode risk classification result is generated. Combined with the failure mode set with mapping attributes obtained in step S2, the verification handling signal corresponding to the risk level is output.

[0064] For any fault mode, generate its output result. ,in Indicates the first i One failure mode, This indicates the order of the failure mode in the failure mode handling priority sequence. This indicates the risk level of the failure mode. This indicates the risk level index corresponding to the risk level of the failure mode. This indicates the dominant risk factor for this failure mode. This indicates the corresponding verification and handling signal.

[0065] Furthermore, the verification processing signal It consists of signal type, key points of verification, and target object. .in For the verification and handling signal type determined by the risk level, The key areas of verification are determined by the dominant risk factors. The object to which the fault mode mapping property is defined.

[0066] Furthermore, the type of verification and handling signal is determined according to the risk level of the failure mode: when the failure mode belongs to the safety and environmental protection risk, This is a significant risk verification signal, used to trigger a review of wellhead isolation capability, emergency shutdown function, or production channel isolation status; when the failure mode belongs to the risk of barrier function degradation, This is a barrier function verification signal used to trigger verification of pressure holding capability, valve sealing status, or dual-valve isolation capability; when the failure mode falls under the control and feedback anomaly risk category, The control feedback verification signal is used to trigger the verification of control command response, valve position feedback, pressure feedback, communication link, or hydraulic control loop; when the fault mode belongs to the general monitoring concern risk, To continuously monitor and track signals, it is used to trigger status trend tracking, periodic functional testing, or historical record review.

[0067] Furthermore, the focus of the investigation is determined according to the dominant risk factor of the failure mode: when the dominant risk factor is abnormal exposure, For the review of evidence of anomalies; when the dominant risk factor is the degree of barrier influence, For verification of barrier effectiveness; when the dominant risk factor is detection difficulty, To ensure comprehensive coverage; when the dominant risk factor is recovery difficulty, For remote recovery capability review; when the dominant risk factor is the severity of the consequences, For the purpose of boundary verification.

[0068] The output includes risk classification results such as failure mode, response priority, risk level, risk index, dominant risk factor, verification response signal type, key points of verification, and target of the failure. This transforms the underwater electrical control tree failure risk assessment results into structured technical outputs for barrier function verification, control feedback verification, remote recovery review, and continuous monitoring and tracking.

[0069] This invention also provides a fault mode risk classification system for marine oil subsea electrical control trees, such as... Figure 2 As shown, it includes a data access layer 100, a model organization layer 200, a risk calculation layer 300, and a result output layer 400.

[0070] The data access layer 100 includes an operation monitoring evidence access module 110, a function verification evidence access module 120, and a fault handling evidence access module 130. The operation monitoring evidence access module 110, the function verification evidence access module 120, and the fault handling evidence access module 130 respectively access the operation monitoring evidence, function verification evidence, and fault handling evidence of the underwater electrical control tree, and convert the data from each evidence source into a unified data record to form an evidence data package, which is then sent to the model organization layer 200.

[0071] The model organization layer 200 includes an underwater electronic control tree control-feedback-barrier mapping model library 210 and a fault mode library 220. The underwater electronic control tree control-feedback-barrier mapping model library 210 is used to store the structured mapping relationship between control behavior, execution object, safety constraints, status feedback, barrier function, and failure consequences; the fault mode library 220 is used to store a set of fault modes formed by control execution class, control timing class, feedback anomaly class, and barrier function degradation class fault modes.

[0072] The risk calculation layer 300 is used to calculate and process fault mode data with mapping attributes, that is, to execute steps S3-S7 in the above-mentioned fault mode risk classification driven by underwater electrical control tree control-feedback-barrier mapping.

[0073] The result output layer 400 is used to output the risk classification results and differentiated verification and handling signals, that is, to execute step S8 in the above-mentioned underwater electrical control tree control-feedback-barrier mapping driven fault mode risk classification.

[0074] Compared with the prior art, the present invention has at least the following beneficial effects: 1. This invention constructs an underwater electrical control tree control-feedback-barrier mapping model, which structurally associates control behavior, execution objects, safety constraints, status feedback, barrier functions, and failure consequences. This allows fault modes to be identified not only by the name of the equipment anomaly, but also by their association with the corresponding control link, feedback evidence, barrier functions, and consequence boundaries. Consequently, it can identify hidden fault modes such as control commands being executed but barrier functions not being effectively verified, and status feedback showing normal operation but pressure holding capacity deteriorating. This improves the completeness and technical relevance of identifying underwater electrical control tree barrier function degradation faults.

[0075] 2. This invention constructs a homogeneous risk perception matrix using operational monitoring evidence, functional verification evidence, and fault handling evidence, respectively. Evidence source weights are generated based on the characteristics of the evidence source data and the degree of consensus support among the evidence sources. The multi-source risk perception matrices are then fused. This reduces the impact of missing evidence sources, abnormal deviations, or isolated false alarms on fault risk assessment results, enabling the risk assessment results to simultaneously reflect actual operational anomalies, functional verification anomalies, and the difficulty of fault handling, thereby improving the stability and objectivity of the underwater electronic control tree fault mode risk perception results.

[0076] 3. This invention generates risk factor weights based on a comprehensive risk perception matrix, and generates failure mode handling priorities, risk levels, risk indexes, and dominant risk factors through bidirectional risk dominance relationships and risk classification. Simultaneously, it outputs verification handling signal types and key points for inspection based on risk levels and dominant risk factors. This avoids insufficient handling basis caused by outputting only a single risk score or simple ranking, transforming failure risk assessment results into structured technical outputs for barrier function verification, control feedback verification, remote recovery review, and continuous monitoring and tracking, thereby improving the interpretability and engineering feasibility of underwater electrical control tree failure risk classification results. Attached Figure Description

[0077] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0078] Figure 1 This is a flowchart of a fault mode risk classification method for an underwater electrical control tree in marine oil according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the fault mode risk classification system for an underwater electrical control tree in marine oil according to an embodiment of the present invention. Detailed Implementation

[0079] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0080] Example 1: This embodiment uses a subsea electrical control tree (DCS) of a deepwater oil and gas field as an example. This DCS includes a main production valve, a production wing valve, an annular valve, a chemical injection valve, a hydraulic control loop, a pressure monitoring unit, a valve position feedback unit, a communication unit, and emergency shutdown logic. The system collects data from three sources: operational monitoring evidence, functional verification evidence, and fault handling evidence.

[0081] I. Constructing an underwater electronic control tree-control-feedback-barrier mapping model: Based on the control logic and safety barrier configuration of the underwater electronic control tree, a mapping relationship is established between control behavior, execution objects, safety constraints, status feedback, barrier functions, and failure consequences. Table 1 shows some of the control-feedback-barrier mapping results.

[0082] Table 1: Examples of Control-Feedback-Barrier Mapping Models

[0083] II. Identifying Fault Modes Based on the Control-Feedback-Barrier Mapping Model: Based on the mapping relationships in Table 1, failure modes are identified from four aspects: control behavior execution status, control behavior timing, status feedback effectiveness, and barrier function verification results. This embodiment identifies six typical failure modes, as shown in Table 2. Each failure mode is not an isolated failure name, but rather corresponds to mapping attributes in Table 1 such as control behavior, execution object, status feedback, barrier function, and failure consequences.

[0084] Table 2: Failure Mode Set of Underwater Electrical Control Tree

[0085] III. Constructing a multi-source risk perception matrix for underwater electrical control trees: This embodiment employs five risk factors: anomalous exposure, barrier impact, detection difficulty, recovery difficulty, and consequence severity. Evidence levels are converted into risk evidence values ​​between 0 and 1, where no evidence is 0, weak evidence is 0.25, moderate evidence is 0.50, strong evidence is 0.75, and invalid evidence is 1.00. For continuous data, risk evidence values ​​are obtained by interpolation between adjacent levels based on the duration of the anomaly, the magnitude of the deviation, test bias, or recovery time.

[0086] Based on the various failure modes and their mapping relationships in Tables 1 and 2, risk perception matrices are generated from operational monitoring evidence, functional verification evidence, and failure handling evidence, as shown in Tables 3 to 5.

[0087] Table 3: Operational Monitoring Risk Perception Matrix

[0088] Table 4: Functional Validation Risk Perception Matrix

[0089] Table 5: Fault Handling Risk Perception Matrix

[0090] IV. Generating evidence source weights and fusing them into a risk perception matrix: Data completeness rate, effective evidence coverage rate, and risk discrimination are calculated for the three types of risk perception matrices, and the evidence source weights are determined by combining the consensus support between the matrices. The results obtained in this embodiment are shown in Table 6.

[0091] Table 6: Results of Evidence Source Weight Calculation

[0092] Based on the evidence source weights in Table 6, Tables 3 to 5 are weighted and fused to obtain the comprehensive risk perception matrix, as shown in Table 7.

[0093] Table 7: Comprehensive Risk Perception Matrix

[0094] V. Generating risk factor weights based on the comprehensive risk perception matrix: Based on the comprehensive risk perception matrix in Table 7, the information difference, logarithmic deviation characteristics and criterion removal effects of each risk factor are extracted and normalized. The resulting risk factor weights are shown in Table 8.

[0095] Table 8: Risk Factor Weights

[0096] As shown in Table 8, in this embodiment, the detection difficulty has the highest weight, indicating that the differences in detectability among different failure modes are the most significant and contribute the most to the differentiation of the failure mode risk structure. The weights of barrier impact, recovery difficulty, and consequence severity are similar, indicating that the failure risk of the underwater electrical control tree is simultaneously affected by barrier function, remote recovery capability, and consequence boundary.

[0097] VI. Generating a failure mode handling priority sequence based on the two-way risk dominance relationship: Based on the comprehensive risk perception matrix in Table 7 and the risk factor weights in Table 8, a two-way risk dominance comparison is performed on the failure modes to form a risk dominance matrix, and a failure mode handling priority sequence is generated accordingly. Table 9 shows the ranking results of this embodiment.

[0098] Table 9: Results of Failure Mode Handling Priority

[0099] As shown in Table 9, FM1, FM6, and FM4 are of higher priority. Combining Tables 1 and 2, FM1 directly affects the production isolation barrier, FM6 affects the emergency shutdown barrier, and FM4 indicates that the pressure maintenance barrier verification failed.

[0100] VII. Failure Mode Risk Classification for Underwater Electrical Control Tree (EDB): Based on the failure mode handling priority sequence, a risk structure feature vector is constructed, and the risk structure separation degree between adjacent failure modes is calculated along the priority sequence. Based on the candidate risk level boundary and the grading effectiveness, the risk grading results of this embodiment are obtained, as shown in Table 10.

[0101] Table 10: Risk Levels and Dominant Risk Factors

[0102] As shown in Table 10, FM1 and FM6 are classified as safety and environmental protection risk levels, indicating that they are at a high level in terms of risk intensity and priority of treatment; FM4 is classified as barrier function degradation risk level, indicating that its risk source is mainly the failure of pressure holding function verification; FM3 is classified as control and feedback anomaly risk level, indicating that its main manifestation is hydraulic control anomaly; FM2 and FM5 are classified as general monitoring concern risk level, but since their dominant risk factor is the difficulty of detection, the detection coverage still needs to be monitored in the future.

[0103] 8. Output risk classification results and differentiated verification and handling signals: Based on the failure mode handling priority, risk level, level risk index and dominant risk factor, as well as the failure mode mapping attribute, the failure mode risk classification results and verification handling signals are generated, as shown in Table 11.

[0104] Table 11: Failure Mode Risk Classification Results and Verification Handling Signals

[0105] As shown in Table 11, this embodiment not only outputs the priority of fault mode handling, but also the risk level, risk index of the level, dominant risk factor, verification and handling signal type, key points of verification, and target objects. For example, FM1 corresponds to the main production valve and production isolation function, so its verification and handling signal points to major risk review and remote recovery capability review; FM4 corresponds to the pressure holding function, so its verification and handling signal points to barrier effectiveness review; although FM2 is ranked lower, its detection difficulty is the dominant risk factor, so the key points of verification for detection coverage review are output.

[0106] This embodiment demonstrates that the present invention can utilize the control-feedback-barrier mapping relationship to bind underwater electronic control tree failure modes with execution objects, barrier functions, and failure consequences. Furthermore, it generates a risk perception matrix, evidence source weights, risk factor weights, failure mode handling priorities, risk levels, and verification and handling signals based on three types of objective evidence sources. Compared to methods that only output failure ranking, the present invention can transform failure mode risk classification results into structured technical outputs oriented towards barrier function verification, control feedback verification, remote recovery review, and continuous monitoring and tracking.

[0107] The above calculation results can be automatically completed by the system described in Example 2.

[0108] Example 2: This embodiment provides an engineering implementation method for a fault mode risk classification system for an underwater electrical control tree in marine oil. This embodiment focuses on the system deployment method, data flow process, and result output method.

[0109] In this embodiment, the system is deployed in a joint computing environment consisting of an offshore platform monitoring server and a shore-based operation and maintenance server. The offshore platform monitoring server is used to access underwater electrical control tree (EDS) operation monitoring evidence, functional verification evidence, and fault handling evidence. The shore-based operation and maintenance server is used to store the control-feedback-barrier mapping model, fault mode library, and historical risk classification results. Data synchronization between the offshore platform monitoring server and the shore-based operation and maintenance server is achieved through an industrial data interface or a secure communication link.

[0110] In a risk classification task, the system operates according to the following data flow.

[0111] First, the data access layer 100 receives three types of evidence source data. Specifically, the operation monitoring evidence access module 110 receives control commands, valve position status, pressure parameters, hydraulic parameters, communication status, and alarm status; the function verification evidence access module 120 receives valve action test, pressure holding test, hydraulic circuit test, and emergency shutdown test results; and the fault handling evidence access module 130 receives alarm handling, remote reset, valve retest, maintenance and replacement, shutdown recovery, and external intervention records. The data access layer 100 converts the above data into unified data records and forms an evidence data package.

[0112] Then, the control-feedback-barrier mapping model library 210 of the model organization layer 200 matches the corresponding security constraints, state feedback, barrier functions and failure consequences according to the execution object and control behavior in the access evidence data packet; the fault mode library 220 receives the matching results output by the control-feedback-barrier mapping model library 210, and determines the corresponding fault mode according to the identification rules of control execution abnormality, control timing abnormality, feedback abnormality and barrier function degradation, forming a fault mode data packet with mapping attributes.

[0113] Subsequently, the risk calculation layer 300 processes the failure mode data with mapping attributes. Specifically, the risk perception matrix generation module generates operational monitoring risk perception matrices, functional verification risk perception matrices, and failure handling risk perception matrices; the evidence source weight generation module generates evidence source weights based on the data completeness rate, effective evidence coverage rate, risk discrimination, and consensus support between the three types of risk perception matrices; the matrix fusion module generates a comprehensive risk perception matrix based on the evidence source weights; the risk factor weighting module calculates the risk factor weights for abnormal exposure, barrier impact, detection difficulty, recovery difficulty, and consequence severity from the comprehensive risk perception matrix; the risk dominance analysis module generates a failure mode handling priority sequence; and the risk grading module generates the failure mode risk level, level risk index, and dominant risk factor.

[0114] Finally, the output of the risk calculation layer 300 is fed into the result output layer 400. The risk grading result output module generates a fault risk grading result, including the fault mode name, handling priority, risk level, level risk index, and dominant risk factor. The verification handling signal output module receives the fault risk grading result and simultaneously calls the fault mode data packet with mapping attributes output by the model organization layer 200. Based on the risk level, it determines the verification handling signal type, based on the dominant risk factor, it determines the key points of verification, and based on the fault mode mapping attributes, it determines the target.

[0115] In this embodiment, after the system processes the six fault modes in Embodiment 1, the output results are shown in Table 11.

[0116] In another implementation, the data access layer 100, model organization layer 200, risk calculation layer 300, and result output layer 400 can be deployed in the same industrial computer; alternatively, the data access layer 100 can be deployed on an offshore platform, while the model organization layer 200, risk calculation layer 300, and result output layer 400 can be deployed in a shore-based operation and maintenance center. The above deployment methods do not affect the data flow between the modules of this invention or the method of generating risk classification results.

[0117] Example 3: This embodiment provides an electronic device and a computer-readable storage medium for implementing the above-described underwater electrical control tree fault mode risk classification method.

[0118] The electronic device includes a processor, a memory, a communication interface, and a display interface. The communication interface receives operational monitoring evidence, functional verification evidence, and fault handling evidence from the underwater electronic control tree (ECMA). The memory stores a control-feedback-barrier mapping model, a fault mode library, a risk perception matrix, evidence source weights, a comprehensive risk perception matrix, risk factor weights, a risk dominance matrix, and risk classification results. The processor executes program instructions stored in the memory to perform the following processes: constructing a control-feedback-barrier mapping model; identifying fault modes in the underwater ECMA; generating three types of risk perception matrices; calculating evidence source weights and fusing the risk perception matrices; generating risk factor weights; generating a fault mode handling priority sequence; completing fault mode risk classification; and outputting verification and handling signals.

[0119] The display interface is used to output the fault mode risk classification results, which include fault mode name, handling priority, risk level, level risk index, dominant risk factor, verification handling signal type, and key points for verification.

[0120] The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the underwater electrical control tree control-feedback-barrier mapping-driven fault mode risk classification method described in this invention. The computer-readable storage medium includes, but is not limited to, hard disks, solid-state drives, memory cards, read-only memory, random access memory, or server storage media.

[0121] The method of the present invention can be deployed as an independent functional unit in an underwater electrical control tree safety operation and maintenance system using the aforementioned electronic device or computer-readable storage medium, or it can be integrated into an offshore platform monitoring system, a shore-based operation and maintenance system, or an underwater production system safety analysis platform.

[0122] Therefore, those skilled in the art should recognize that although numerous exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Thus, the scope of the present invention should be understood and construed as covering all such other variations or modifications.

Claims

1. A method for fault mode risk classification of subsea electrical control trees in marine oilfields, characterized in that, Includes the following steps: S1: Construct an underwater electronic control tree control-feedback-barrier mapping model; Identify the structured relationships between control behaviors, execution objects, safety constraints, status feedback, barrier functions, and failure consequences involved in the operation of the underwater electronic control tree, and form an underwater electronic control tree control-feedback-barrier mapping model; S2. Identify fault modes based on the underwater electronic control tree control-feedback-barrier mapping model; The fault modes that can lead to abnormal barrier function or failure during the operation of the underwater electrical control tree are identified, and a set of fault modes with mapping attributes is formed. S3. Construct a multi-source risk perception matrix for underwater electrical control trees; The operation monitoring risk perception matrix, the function verification risk perception matrix, and the fault handling risk perception matrix are generated from three evidence sources: operation monitoring evidence, function verification evidence, and fault handling evidence, respectively, forming a multi-source risk perception matrix; S4. Generate evidence source weights and fuse them into a risk perception matrix; The weights of each evidence source are determined by combining data feature dimensionality reduction with consensus support, and the three risk perception matrices are fused to obtain a comprehensive risk perception matrix. S5. Generate risk factor weights based on the comprehensive risk perception matrix; The weights of each risk factor are calculated using a risk factor weighting method based on information entropy characteristics, logarithmic bias characteristics, and criterion removal effects. S6. Generate a fault mode handling priority sequence based on bidirectional comprehensive risk dominance value; Calculate the bidirectional comprehensive risk dominance value among the various failure modes of the underwater electronic control tree, form a risk dominance matrix, and generate a failure mode handling priority sequence based on it. S7. Implement fault mode risk classification for underwater electrical control trees; Risk classification is performed on the failure modes of the underwater electronic control tree, and the risk level, candidate risk level boundary, level risk index and dominant risk factor corresponding to each failure mode are output. S8. Output risk classification results and differentiated verification and handling signals; Generate the risk classification results of underwater electrical control tree failure modes, and combine them with the failure mode set with mapping attributes obtained in step S2 to output the verification and handling signals corresponding to the risk level.

2. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 1, characterized in that, The specific steps of S1 are as follows: The control behavior refers to the operational behavior issued by the control system during the operation of the underwater electrical control tree, including one or more of the following: production control, annular control, chemical injection control, hydraulic control, and emergency shutdown control; the execution object refers to the equipment or circuit directly affected by the control behavior, including one or more of the following: production main valve, production wing valve, annular valve, chemical injection valve, and hydraulic control circuit. The safety constraints refer to the control requirements, response requirements, or status requirements that must be met under the corresponding operating conditions to ensure the effectiveness of the underwater electrically controlled tree barrier function; the safety constraints include one or more of the following: the valve should act according to the preset control command; the valve should generate corresponding status feedback after acting; the pressure should be maintained after shutting down or isolating; and the barrier should not be confirmed as effective based on a single feedback signal when feedback is abnormal. The status feedback refers to the operating status information returned by the underwater electronic control tree after the control behavior is executed, including one or more of the following: valve position status, pressure parameters, temperature parameters, flow parameters, hydraulic parameters, communication status, and alarm information; the status feedback is used to determine whether the control behavior has been executed correctly and to form an objective basis for verifying the effectiveness of the barrier function. The barrier function refers to the function of the subsea electrical control tree in maintaining wellhead safety, production isolation, annular isolation, pressure maintenance, injection isolation, flow regulation, or emergency shutdown. The failure consequences refer to the consequences caused by the failure of the barrier function or its inability to be effectively verified, including one or more of the following: unintended release of wellhead fluid, oil and gas leakage, uncontrolled production pressure, uncontrolled annular pressure, abnormal discharge of chemical agents, inability to remotely shut down, equipment damage, threats to personnel safety, and marine environmental pollution.

3. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 2, characterized in that, The specific steps of S2 are as follows: For each control action, fault modes are identified from four aspects: control action execution status, control action timing, status feedback effectiveness, and barrier function verification results, forming a fault mode set of the underwater electronic control tree. Furthermore, when a control action is not executed, the execution object is incorrect, or the execution result does not meet safety constraints, it is identified as a control execution type failure mode; when a control action is executed too early, too late, in the wrong order, or the duration does not meet the preset requirements, it is identified as a control timing type failure mode; when status feedback is missing, delayed, frozen, drifting, abrupt, or inconsistent with the expected response, it is identified as a feedback anomaly type failure mode; when a control action has been executed and the status feedback shows normal, but cannot prove that the corresponding barrier function is effective, it is identified as a barrier function degradation type failure mode. Furthermore, for each identified fault mode, its corresponding control behavior, execution object, safety constraints, status feedback, barrier function, and failure consequences are retained to form a set of fault modes with mapping attributes.

4. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 3, characterized in that, The specific steps of S3 are as follows: The operational monitoring evidence is used to characterize the abnormal behavior of the failure mode during actual operation; the functional verification evidence is used to characterize the abnormal behavior of the failure mode during functional verification. The fault handling evidence is used to characterize the handling performance of fault modes in alarm handling, remote reset, valve retesting, maintenance and replacement, production shutdown recovery or external intervention processes; For each source of evidence, a risk evidence value is generated according to the same risk factor dimensions; the risk factors include abnormal exposure, barrier impact, detection difficulty, recovery difficulty, and consequence severity. Furthermore, the original records from each source of evidence are converted into risk evidence values ​​of a uniform dimension according to a preset evidence level; the preset evidence levels include no evidence, weak evidence, moderate evidence, strong evidence, and invalid evidence, and the higher the evidence level, the greater the corresponding risk evidence value. A risk perception matrix is ​​constructed with fault modes as rows and risk factors as columns for each evidence source: operation monitoring risk perception matrix, functional verification risk perception matrix, and fault handling risk perception matrix. These three risk perception matrices constitute a multi-source risk perception matrix.

5. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 4, characterized in that, The specific steps of S4 are as follows: Assume the number of fault modes of the underwater electrical control tree is . m The number of risk factors is n The number of evidence sources is ,in ;No. k The risk perception matrix corresponding to each source of evidence is represented as follows: ,in, Indicates the first k The first source of evidence i The failure mode is in the first j Risk evidence values ​​for each risk factor, and After processing in step S3, it is converted into a numerical value with uniform dimensions; First, a data feature vector is constructed for each source of evidence. The data feature vector includes one or more of the following: data completeness rate, effective evidence coverage rate, and risk discrimination rate. Among them, the data completeness rate is used to characterize the proportion of non-missing risk evidence values ​​in the source of evidence; the effective evidence coverage rate is used to characterize the proportion of units in the source of evidence that can provide effective risk evidence; and the risk discrimination rate is used to characterize the ability of the source of evidence to distinguish the risk differences of different failure modes. For the k Each source of evidence, its data integrity rate Valid evidence coverage and risk differentiation They are respectively , , ;in, For the first k The number of non-missing units in each risk perception matrix For the first k The number of units in a risk perception matrix whose risk evidence value is greater than a preset valid evidence threshold. For the first k The first source of evidence j The average risk evidence value of each risk factor; according to , and Constructing the feature matrix of evidence sources , No. k The line represents the first k Data feature vectors corresponding to each source of evidence; feature matrix of the source of evidence. F Centralized processing, i.e. F Subtracting the average of the data in each column from the average of the data in the matrix yields the centered matrix. For the centralized matrix Standardization processing, i.e. Divide each data point by the standard deviation of the data in its corresponding column to obtain the standardized matrix. Based on standardized matrix Constructing the covariance matrix For the covariance matrix Perform eigenvalue decomposition to obtain eigenvalues. and corresponding feature vectors ,in t Indicate the principal component index; sort by eigenvalue from largest to smallest, and select the cumulative contribution rate. Before reaching the preset threshold q Principal components, For the first k For each source of evidence, calculate its projection value along the directions of each principal component. ,in Represents the normalized matrix The first in k Row vectors; further, the first row vector is calculated based on the cumulative contribution rate of each principal component. k Data feature scores of each source of evidence ;right Perform nonnegation processing. Feature scores of the data after nonnegation processing Normalization is performed to obtain the data feature weights. ; Then, the consensus support of the evidence sources is calculated based on the similarity among the three risk perception matrices; for the first... k The first source of evidence and the first r Given several sources of evidence, calculate the matrix distance between them. Convert matrix distance into consensus support. ,in The maximum value among all pairwise matrix distances of all evidence sources; for the th k Each source of evidence is used to calculate its total consensus support. and will Normalization, gaining consensus support for weights ; Then, a fusion coefficient was introduced. λ Weighting data features and consensus support weights By merging, we obtain the first... k Weight of each source of evidence ;in, When greater emphasis is placed on the data integrity, coverage, and discriminative power of the evidence sources themselves, improving... λ Value; when greater emphasis is placed on consistency among sources of evidence, reduce λ Values; Finally, based on the weight of the evidence source The three risk perception matrices are weighted and fused together, and then... Fusion i The failure mode is in the first j The comprehensive risk perception matrix is ​​obtained by combining the risk evidence values ​​of each risk factor. .

6. The method for fault mode risk classification of an underwater electrical control tree for offshore oil and gas as described in claim 5, characterized in that, The specific steps of S5 are as follows: First, the comprehensive risk perception matrix X The risk factor columns in the data are normalized, i.e. The normalized matrix is ​​obtained. Normalized Used to characterize the i The failure mode is in the first j The relative strength of risk evidence under each risk factor, with a larger value indicating stronger risk evidence; Then, the information entropy feature of each risk factor is calculated; For the j First, calculate the proportion of each risk factor in each failure mode. Then calculate the information entropy value of the risk factor. To obtain information difference ,in The larger the value, the higher the value. j The stronger the distinguishing information that a risk factor provides between different failure modes; Furthermore, the logarithmic bias characteristics of each risk factor are calculated; For the j Each risk factor was used to calculate its average relative risk evidence strength. And obtain the logarithmic deviation value. ,in The larger the value, the higher the value. j The more significant the deviation of a risk factor between different failure modes, the stronger its ability to characterize the differences in failure mode risk. Furthermore, the criterion removal effect for each risk factor is calculated; First calculate the effect of all risk factors combined. i Average risk structure value of each failure mode Then calculate the removal of the first j After the first risk factor i Average risk structure value of each failure mode , Indicates the first i The failure mode is in the first The relative strength of risk evidence under each risk factor, where the first... The risk factor belongs to the category excluding the removed one. j Other risk factors besides the individual risk factors; Therefore, we obtain the first... j Criterion impact value of each risk factor ,in The larger the value, the more likely it is to remove the first element. j The more significant the change in the overall risk structure of the failure mode after a risk factor, the greater the contribution of that risk factor to the information of the comprehensive risk perception matrix. Subsequently, the degree of information difference Logarithmic deviation value and the influence value of the criteria Normalization is performed. , , And calculate the first j The combined information contribution value of each risk factor ,in a , b , c The preset adjustment coefficient satisfies a ≥0、 b ≥0、 c ≥0 and a + b + c =1; When more emphasis is placed on the overall discrete information of risk factors, improve a Value selection; when more emphasis is placed on the local deviation characteristics of risk factors, increase b Value selection; when more emphasis is placed on the impact of risk factors on the overall risk structure, increase c Values; Finally, the overall information contribution value of each risk factor is normalized to obtain the risk factor weights. .

7. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 6, characterized in that, The specific steps of S6 are as follows: First, based on the comprehensive risk perception matrix Perform a bidirectional comparison for any two fault modes; for the first fault mode... j The nth risk factor is calculated. i The failure mode is relative to the first h Local risk dominance of each failure mode ; Then, combined with risk factor weights Calculate the first i The failure mode is relative to the first h Overall risk dominance of each failure mode Similarly, calculate the first... h The failure mode is relative to the first i Overall risk dominance of each failure mode ; Based on the bidirectional comprehensive risk dominance value, construct the first i The failure mode is relative to the first h Risk dominance probability index for each failure mode ;in, Indicates the first i The failure mode is relative to the first h Each failure mode has a higher risk dominance relationship. Indicates the first h The failure mode is relative to the first i Each failure mode has a higher risk dominance relationship. This indicates that the risk dominance relationship is the same for both; a risk dominance matrix is ​​formed by pairwise comparison of all failure modes. diagonal elements Take 0.5; Subsequently, a recursive baseline partitioning is performed based on the risk dominance matrix to generate a priority sequence for fault mode handling. Specifically, a fault mode is selected from the set of fault modes to be sorted as the baseline fault mode, and its index is denoted as [index missing]. p The remaining failure modes are then divided into high-risk dominant subsets, approximately equal subsets, and low-risk dominant subsets based on their risk dominance relationship with the baseline failure mode; when , will the h Each failure mode is classified into the high-risk dominant subset; when , will the h Each failure mode is classified into the low-risk dominant subset; when , will the h Each failure mode is assigned to an approximately equal subset; among them... τ The preset tolerance threshold; Continue to perform baseline failure mode selection and recursive partitioning on the high-risk dominant subset and the low-risk dominant subset until the number of failure modes in each subset is no greater than 1, and obtain the failure mode handling priority sequence. The failure modes in the failure mode handling priority sequence are arranged from high to low according to the risk handling priority.

8. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 7, characterized in that, The specific steps of S7 are as follows: First, a risk structure feature vector is constructed for each failure mode; for the th i One failure mode, based on a comprehensive risk perception matrix and risk factor weights Calculate its overall risk intensity and dominant contribution value The dominant contribution value is then normalized to obtain the contribution vector of key risk factors. Then, based on the risk dominance matrix Calculate the first i Net dominance of each failure mode The dominant intensity feature is then normalized to obtain the dominant intensity feature. ; Combining comprehensive risk intensity Dominant intensity characteristics and key risk factor contribution vector , construct the first i Risk structure feature vector of each failure mode ; Then, the risk structure separation degree between adjacent failure modes is calculated along the failure mode handling priority sequence; For adjacent failure modes in the failure mode handling priority sequence, the order is represented as follows: and Calculate the risk structure separation degree between the two. ; Furthermore, the candidate risk level boundary is determined based on the separation degree of all adjacent risk structures; Calculate the average value of risk structure separation. and standard deviation When the first The adjacent positions satisfy When this occurs, the adjacent location is determined as the boundary of the candidate risk level, where ρ The preset boundary sensitivity coefficient; Based on the candidate risk level boundaries, the candidate classification schemes are obtained as follows: ,in Indicates the first s The set of failure modes in each risk level g Indicates the number of risk levels; Calculate the intra-level compactness of the candidate grading scheme. ,in It belongs to the first s The mean of the risk structure feature vectors of failure modes for each risk level; Calculate the inter-level separation of the candidate grading scheme. ,in It is the mean of the risk structure feature vectors of all failure modes across all risk levels. Indicates the first s The number of failure modes in each risk level; further calculate the effectiveness of the classification. ,in χ This is a penalty coefficient for the number of risk levels, used to prevent risk levels from being overly subdivided. ε To prevent extremely small positive numbers with a denominator of zero; choose to make The largest grading scheme is used as the target risk grading scheme. ; Based on the target risk classification scheme, the failure mode handling priority sequence is divided into several consecutive risk level intervals, and risk levels are assigned in descending order. Furthermore, a risk index is calculated for each risk level; for the first... s There are several risk levels, and the risk index for each level is: ,in, Indicates the first i Each failure mode; the risk level index is used to characterize the overall risk intensity and overall dominance of the risk level; Further, calculate the first s The dominant risk factor for each risk level This indicates the risk factor that plays a major role in the formation of the risk level; Based on the target risk classification scheme, level risk index and dominant risk factors, output the risk level corresponding to each failure mode; the risk level includes two or more of the following: safety and environmental protection risk, barrier function degradation risk, control and feedback anomaly risk, and general monitoring concern risk.

9. The method for fault mode risk classification of an underwater electrical control tree for marine oilfields according to claim 8, characterized in that, The specific steps of S8 are as follows: For any fault mode, generate its output result. ,in Indicates the first i One failure mode, This indicates the order of the failure mode in the failure mode handling priority sequence. This indicates the risk level of the failure mode. This indicates the risk level index corresponding to the risk level of the failure mode. This indicates the dominant risk factor for this failure mode. This indicates the corresponding verification and handling signal; Furthermore, the verification processing signal It consists of signal type, key points of verification, and target object. ;in For the verification and handling signal type determined by the risk level, The key areas of verification are determined by the dominant risk factors. The object to which the failure mode mapping property points; Furthermore, the type of verification and handling signal is determined according to the risk level of the failure mode: when the failure mode belongs to the safety and environmental protection risk, This is a major risk verification signal, used to trigger the verification of wellhead isolation capability, emergency shutdown function, or production channel isolation status; When the failure mode is at risk of barrier function degradation This is a barrier function verification signal used to trigger verification of pressure holding capability, valve sealing status, or dual-valve isolation capability; when the failure mode falls under the control and feedback anomaly risk category, The control feedback verification signal is used to trigger the verification of control command response, valve position feedback, pressure feedback, communication link, or hydraulic control loop; when the fault mode belongs to the general monitoring concern risk, To continuously monitor and track signals, used to trigger status trend tracking, periodic functional testing, or historical record review; Furthermore, the focus of the investigation is determined according to the dominant risk factor of the failure mode: when the dominant risk factor is abnormal exposure, For the review of evidence of anomalies; when the dominant risk factor is the degree of barrier influence, For the purpose of verifying the effectiveness of the barrier; When the dominant risk factor is the difficulty of detection To ensure comprehensive coverage; when the dominant risk factor is recovery difficulty, For remote recovery capability review; when the dominant risk factor is the severity of the consequences, For the purpose of boundary verification; The output includes fault modes, handling priorities, risk levels, risk indexes, dominant risk factors, verification and handling signal types, key points of verification, and risk classification results of the target objects. This transforms the underwater electrical control tree fault risk assessment results into structured technical outputs for barrier function verification, control feedback verification, remote recovery review, and continuous monitoring and tracking.

10. A fault mode risk classification system for an underwater electrical control tree in marine oilfields, characterized in that, The method for fault mode risk classification of an underwater electrical control tree for marine oil, as described in any one of claims 1-9, includes a data access layer (100), a model organization layer (200), a risk calculation layer (300), and a result output layer (400). The data access layer (100) includes an operation monitoring evidence access module (110), a function verification evidence access module (120), and a fault handling evidence access module (130). The operation monitoring evidence access module (110), the function verification evidence access module (120), and the fault handling evidence access module (130) respectively access the operation monitoring evidence, function verification evidence, and fault handling evidence of the underwater electrical control tree, and convert the data from each evidence source into a unified data record to form an evidence data package, which is then sent to the model organization layer (200). The model organization layer (200) includes an underwater electronic control tree control-feedback-barrier mapping model library (210) and a fault mode library (220); the underwater electronic control tree control-feedback-barrier mapping model library (210) is used to store the structured mapping relationship between control behavior, execution object, safety constraints, state feedback, barrier function and failure consequences; the fault mode library (220) is used to store a set of fault modes formed by control execution class, control timing class, feedback anomaly class and barrier function degradation class fault modes; The risk calculation layer (300) is used to perform calculations on fault mode data with mapping attributes; The result output layer (400) is used to output risk classification results and differentiated verification and handling signals.

Images