Aircraft component residual life prediction and maintenance decision method based on digital twinning

Abstract

The application relates to the technical field of aviation life prediction, and discloses an aircraft component residual life prediction and maintenance decision method based on digital twinning, which comprises the following steps: step 1, a mechanism model is established, the value range of a key parameter and an initial health state are determined, and a deterministic conversion rule is set; step 2, a virtual-real consistency index is generated, and consistency gating is performed to update the value range of the key parameter; step 3, a flight task is split into a task spectrum scene set, and upper and lower limits of the load are determined; step 4, a mechanism forward calculation is performed based on the task spectrum scene set to determine a residual life interval; step 5, a candidate maintenance scheme is generated; step 6, a virtual drill is used to eliminate schemes that exceed a failure critical standard; and step 7, a final maintenance scheme is selected, and state record data is written back to update the health state and the value range of the key parameter. The application realizes closed-loop optimization of the prediction of the residual life of an aircraft component and maintenance decision.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of aircraft life prediction, and particularly relates to an aircraft component residual life prediction and maintenance decision method based on digital twinning. BACKGROUND

[0002] With the diversification of aircraft operation tasks, components need to experience different intensities and types of loads in actual operation, and their health status continuously evolves with the change of tasks. Traditional maintenance methods usually rely on preset inspection cycles or manual evaluation based on experience, which is difficult to accurately depict the dynamic change characteristics of component state over time and task scenarios, and also difficult to systematically integrate future tasks, component health trends and resource constraints in maintenance planning. In addition, existing life prediction methods based on fixed models or static parameters lack the ability to continuously absorb actual operation data, and cannot real-time correct the model according to the changing state of the component over time, so that the prediction results and maintenance decisions are difficult to maintain consistency and stability.

[0003] With the development of digital twinning technology, mapping the structural characteristics, operating loads, health status and evolution law of physical components to a computable virtual model has become an important path to realize the life cycle management of aircraft components. Digital twinning can express the state change logic of the component under the action of the load through the mechanism model, and continuously update the model parameters and health status based on real-time data, so as to reconstruct the future state evolution trajectory of the component under different tasks in the virtual space, making life prediction and maintenance strategy generation computable and traceable. However, there are still many challenges in the application of current digital twinning in the aviation field, for example, how to correct the model parameters through consistency evaluation to make the model continuously close to the actual component state; how to convert future flight tasks into structured load information that can be used for model deduction; and how to generate executable maintenance schemes based on life prediction results and verify the effectiveness of the schemes through virtual rehearsal. SUMMARY

[0004] The application provides an aircraft component residual life prediction and maintenance decision method based on digital twinning, which solves the technical problems of difficulty in accurately reflecting component state evolution under complex load conditions, insufficient residual life prediction accuracy, and lack of quantitative decision basis for maintenance strategies in related technologies.

[0005] The application provides an aircraft component residual life prediction and maintenance decision method based on digital twinning, which includes the following steps:

[0006] Step 1, establishing a mechanism model, determining the value range of key parameters, the initial health state, and the deterministic conversion rule of load information to the driving quantity of the mechanism model;

[0007] Step 2, collecting component actual data, generating virtual-real consistency index, and performing consistency gating, when the consistency threshold is not reached, the key parameter value range is updated;

[0008] Step 3, splitting the flight task into a task spectrum scene set, setting the load upper limit and load lower limit, and obtaining the mechanism model driving quantity upper limit and lower limit according to the deterministic conversion rule;

[0009] Step 4, based on the task spectrum scene set and the key parameter value range, performing mechanism forward calculation, and determining the residual life interval according to the preset failure critical standard;

[0010] Step 5, taking the shortest residual life combined with the safety margin time as the safety constraint, generating a candidate maintenance scheme according to the maintenance cost, downtime and failure risk, the candidate maintenance scheme including the maintenance time point and the maintenance level;

[0011] Step 6, performing virtual drilling on the candidate maintenance scheme, resetting the initial state after maintenance according to the maintenance level and performing mechanism forward calculation, and eliminating the candidate maintenance scheme that leads to exceeding the preset failure critical standard;

[0012] Step 7, selecting the final maintenance scheme, writing the state record data to the digital twin after maintenance, and updating the initial health state and the key parameter value range.

[0013] Further, a mechanism model is established to determine the key parameter value range, the initial health state, and the deterministic conversion rule for converting load information into mechanism model driving quantity, including:

[0014] Step 11, establishing a mechanism model, and determining the key parameter value range required by the mechanism model based on the allowed range of material constants, component geometric parameters and damage sensitivity coefficients, and recording the initial value of the crack of the current component as the initial health state;

[0015] Step 12, establishing a deterministic conversion rule according to the component geometric information, material information and boundary conditions, wherein the load information is a stress data sequence varying with time, the deterministic conversion rule performs a fixed conversion step for each time of load information, and outputs the mechanism model driving quantity;

[0016] Step 13, storing the mechanism model, the key parameter value range, the initial health state, the load information, the mechanism model driving quantity and the deterministic conversion rule into the basic data object set of the digital twin.

[0017] Further, a virtual-real consistency index is generated, and consistency gating is performed, when the consistency threshold is not reached, the key parameter value range is updated, including:

[0018] Step 21, collect the measured data of the component in a continuous time period to form a measured data sequence, and call the mechanism model, the initial health state, the key parameter value range and the mechanism model driving quantity obtained by the deterministic conversion rule from the load information in the same time period to generate a twin model output sequence, calculate the difference between the measured data at each time and the twin model output at the corresponding time, and obtain the virtual-real consistency index by averaging the absolute value of the difference in a fixed length time window;

[0019] Step 22, compare the virtual-real consistency index with the consistency gating threshold, and when the virtual-real consistency index does not exceed the consistency gating threshold, confirm that the current mechanism model and the key parameter value range meet the consistency requirement;

[0020] Step 23, in the case where the virtual-real consistency index exceeds the consistency gating threshold, determine the key parameter to be adjusted according to the size and sign of the difference between the two, and update the upper and lower limits of the key parameter value range according to the preset interval contraction rule, and write the updated key parameter value range into the basic data object set of the digital twin.

[0021] Further, when the key parameter value range is updated, the direction of the upper limit and the lower limit of the key parameter value range is determined according to the sign of the difference between the virtual-real consistency index and the consistency gating threshold, and the upper limit and the lower limit are offset inward by a preset offset amount; and after the offset, the difference between the upper limit and the lower limit is compared, and when the difference is less than the preset minimum interval width, the difference is reset to the minimum interval width to form the updated key parameter value range.

[0022] Further, the flight task is divided into a task spectrum scene set, the load upper limit and the load lower limit are set, and the mechanism model driving quantity upper limit and lower limit are obtained according to the deterministic conversion rule, including:

[0023] Step 31, obtain the task information during the future flight to generate a flight task information sequence, analyze according to the flight phase, attitude parameter, speed parameter and external load parameter, and split the flight task information sequence into multiple task spectrum scene sets with independent load behavior characteristics according to the task duration; for each task spectrum scene, a structured description of the time interval and the load action mode is established, and the load action mode includes the load change trend and the load action type in the time interval;

[0024] Step 32, for each task spectrum scene in the task spectrum scene set, determine the load upper limit and the load lower limit according to the flight attitude, the aerodynamic load and the environmental conditions;

[0025] Step 33, input the upper load limit and the lower load limit into the deterministic conversion rule respectively, obtain the upper limit of the mechanism model driving quantity and the lower limit of the mechanism model driving quantity through the fixed conversion step, and construct the driving quantity boundary set of the mission spectrum scene based on the upper limit of the driving quantity and the lower limit of the driving quantity of the entire mission spectrum scene.

[0026] Further, based on the mission spectrum scene set and the key parameter value range, the mechanism forward calculation is performed, and the residual life interval is determined according to the preset failure critical standard, including:

[0027] Step 41, for each mission spectrum scene in the mission spectrum scene set, the mechanism model, the initial health state and the key parameter value range are called, and the upper limit of the mechanism model driving quantity and the lower limit of the mechanism model driving quantity corresponding to the mission spectrum scene are respectively input, and the mechanism forward calculation is performed at a fixed time step; at each time step, the parameter combination in the mechanism model driving quantity and the key parameter value range is substituted into the state update rule of the mechanism model to obtain a damage evolution time sequence;

[0028] Step 42, compare the damage evolution time sequence with the failure critical value at each time point, and use the time point at which the failure critical value is first reached as the mission spectrum scene failure time; the upper limit of the mission spectrum scene failure time is obtained under the condition of the upper limit of the mechanism model driving quantity, the lower limit of the mission spectrum scene failure time is obtained under the condition of the lower limit of the mechanism model driving quantity, and the mission spectrum scene failure time interval is formed;

[0029] Step 43, all mission spectrum scene failure time intervals are summarized, the minimum value of the upper limit of the mission spectrum scene failure time is selected as the lower limit of the residual life, the maximum value of the lower limit of the mission spectrum scene failure time is selected as the upper limit of the residual life, and the residual life interval is formed by the lower limit of the residual life and the upper limit of the residual life.

[0030] Further, when calculating the residual life interval, the lower limit of the failure time and the upper limit of the failure time corresponding to all mission spectrum scenes are recorded, and in the next two rounds of mechanism forward calculation, the absolute difference between the lower limit of the residual life of the last round and the lower limit of the residual life of the current round and the absolute difference between the upper limit of the residual life of the last round and the upper limit of the residual life of the current round are compared with the preset convergence threshold value respectively; when the absolute difference of the two is not more than the preset convergence threshold value, it is determined that the residual life interval meets the convergence condition.

[0031] Further, the candidate maintenance scheme is generated according to the maintenance cost, the downtime and the failure risk, including:

[0032] Step 51, call the lower limit of the remaining life, and calculate the safety margin time with the difference between the lower limit of the remaining life and the preset safety margin coefficient, determine the upper limit of the maintenance time point by subtracting the safety margin time from the lower limit of the remaining life; take the current time as the lower limit of the maintenance time point, and generate candidate maintenance time points between the lower limit and the upper limit of the maintenance time point at fixed time intervals to form a feasible maintenance time point interval;

[0033] Step 52, in the feasible maintenance time point interval, combine each candidate maintenance time point with the maintenance level set one by one to form an initial candidate maintenance scheme set; and determine the downtime based on the work hour table corresponding to the maintenance level, determine the maintenance cost based on the resource consumption corresponding to the maintenance level, and take the reciprocal of the difference between the lower limit of the remaining life and the maintenance time point as the failure risk quantity, obtain the maintenance cost, downtime and failure risk quantity of each item of the initial candidate maintenance scheme set respectively; wherein the maintenance level includes: inspection level, repair level, partial replacement level and whole replacement level;

[0034] Step 53, screen the initial candidate maintenance scheme set, and include the candidate scheme whose maintenance time point falls into the feasible maintenance time point interval and whose failure risk quantity is lower than the preset risk threshold into the candidate maintenance scheme set.

[0035] Further, the candidate maintenance scheme is virtually practiced, the initial state after maintenance is reset according to the maintenance level, and the mechanism forward calculation is performed, and the candidate maintenance scheme leading to exceeding the preset failure critical standard is eliminated, including:

[0036] Step 61, for each candidate maintenance scheme in the candidate maintenance scheme set, call its maintenance level, and perform a deterministic reset operation on the initial health state before maintenance according to the health state reset rule corresponding to the maintenance level; the health state reset rule includes resetting the initial value of the crack; write the initial health state after maintenance into the basic data object set as the starting state of the mechanism model forward calculation;

[0037] Step 62, call the maintenance time point in the candidate maintenance scheme, and call the task spectrum scene set in the time interval after the maintenance time point in turn; for each task spectrum scene, replace the upper limit and lower limit of the mechanism model driving quantity into the state update rule of the mechanism model respectively, and perform iterative calculation with the time resolution of the task spectrum scene as the fixed time step to obtain the damage evolution time sequence after maintenance;

[0038] Step 63, compare the damage evolution time sequence after maintenance with the preset failure critical standard at fixed time steps, judge whether the damage evolution value exceeds the preset failure critical standard, if any point in the sequence exceeds, it is judged as damage overrun, and the corresponding candidate maintenance scheme is marked as unfeasible; otherwise, the maintenance time point, the maintenance level and the corresponding damage evolution time sequence after maintenance are included in the feasible maintenance scheme set.

[0039] Further, the final maintenance scheme is selected, and the state record data is written back to the digital twin after maintenance execution, and the initial health state and the key parameter value range are updated, including:

[0040] Step 71, after the final maintenance scheme is executed, the state record of the component is performed, the initial value of the crack after maintenance, the initial value of the cumulative damage after maintenance, the local geometric feature quantity and the material detection parameter are collected, and the state record data object is generated according to the preset field structure;

[0041] Step 72, the state record data object is called, the crack initial value, cumulative damage initial value and material detection parameter in the state record data object are written into the digital twin basic data object set, the initial health state before maintenance is executed to determine the replacement operation, the initial health state after maintenance is formed, and the initial health state after maintenance is used as the starting state of the next round of forward calculation of the mechanism model;

[0042] Step 73, the key parameters to be corrected are determined according to the material detection parameter and the local geometric feature quantity in the state record data object, the interval adjustment is performed on the key parameter value range according to the preset interval update rule, and the corrected key parameter value range is written into the digital twin basic data object set.

[0043] The beneficial effects of the present application are that the present application constructs a digital twin model closed loop system for aircraft components, realizes the continuous synchronization between the physical component and the virtual model through the unified expression of the mechanism model, the parameter interval, the health state and the task load. By introducing the virtual-real consistency index and the interval contraction mechanism based on the difference, the present application can dynamically correct the model parameters during operation, so that the digital twin can long-term maintain the fitting of the actual component state. In addition, the present application structures the future flight task into a task spectrum scene set, and generates a mechanism model driving quantity boundary based on the load boundary, so that the life prediction can cover the state evolution interval under different task conditions, so as to obtain a residual life interval with boundary significance.

[0044] On the basis of the life prediction result, the present application generates candidate maintenance schemes in combination with the safety margin, the maintenance level, the parking time, the resource consumption and the failure risk, and verifies the feasibility of each scheme by using the virtual training mechanism, so that the maintenance decision is changed from experience dependence to model driving. The state record and parameter update after maintenance execution further ensure that the digital twin always maintains consistency in the life cycle. Overall, the present application realizes the digital twin closed loop of prediction, decision, execution and update, and improves the determinacy, traceability and intelligent level of the aircraft component maintenance planning. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1is a flowchart of an aircraft component residual life prediction and maintenance decision method based on digital twinning of the present application. DETAILED DESCRIPTION

[0046] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can include changes, modifications, additions or omissions of the functions and arrangements of the elements discussed without departing from the scope of the present disclosure. Various examples can omit, substitute, or add various procedures or components as appropriate. Also, it should be understood that instead of a single arrangement, several arrangements of the elements discussed can be used, also in combination with each other.

[0047] As shown in Figure 1 the aircraft component residual life prediction and maintenance decision method based on digital twinning includes the following steps:

[0048] Step 1, establishing a mechanism model, determining the key parameter value range, the initial health state, and the deterministic conversion rule for converting the load information into the mechanism model driving quantity;

[0049] Step 2, collecting component measured data, generating virtual-real consistency indicators, and performing consistency gating, when the consistency threshold is not reached, the key parameter value range is updated;

[0050] Step 3, splitting the flight task into a task spectrum scene set, setting the load upper limit and the load lower limit, and obtaining the mechanism model driving quantity upper limit and lower limit according to the deterministic conversion rule;

[0051] Step 4, performing mechanism forward calculation based on the task spectrum scene set and the key parameter value range, and determining the residual life interval according to the preset failure critical standard;

[0052] Step 5, taking the shortest residual life combined with the safety margin time as a safety constraint, generating a candidate maintenance scheme according to the maintenance cost, the parking time and the failure risk, the candidate maintenance scheme including the maintenance time point and the maintenance level;

[0053] Step 6, performing virtual training on the candidate maintenance scheme, resetting the initial state after maintenance according to the maintenance level and performing mechanism forward calculation, and eliminating the candidate maintenance scheme that leads to exceeding the preset failure critical standard;

[0054] Step 7, selecting the final maintenance scheme, collecting state record data after maintenance execution and writing back to the digital twin, and updating the initial health state and the key parameter value range.

[0055] In an embodiment of the present application, establishing a mechanism model, determining the key parameter value range, the initial health state, and the deterministic conversion rule for converting the load information into the mechanism model driving quantity includes:

[0056] Step 11, for the target aircraft component, based on its material properties, geometric structure and damage sensitivity constraints formed in the design stage, a mechanism model corresponding to the physical component is established, which is used to describe the state evolution logic of the aircraft component under different load conditions and serves as the core execution unit of the digital twin; and based on the allowed ranges of material constants, component geometric parameters and damage sensitivity coefficients, the value range of the key parameters required by the mechanism model is determined, and the initial value of the crack of the current component is recorded as the initial health state; the material constant is used to describe the inherent properties of the aircraft component material in the model, including but not limited to: elastic modulus, Poisson's ratio, material strength coefficient, etc., the component geometric parameter is the geometric feature data describing the structural form of the aircraft component, including but not limited to: component thickness, cross-sectional shape parameter, key position size, etc., the damage sensitivity coefficient is used to describe the parameterized response degree of the component to the damage evolution behavior in the digital twin; the key parameter is a set of core adjustable parameters that constitute the internal mechanism model of the component-level digital twin, including but not limited to: material constant, component geometric parameter, damage sensitivity coefficient and other adjustable input variables related to mechanism model calculation; the crack initial value is used to represent the initial damage state of the mechanism model in the digital twin, which comes from the termination state of the last prediction period or the baseline maintenance record on the digital twin;

[0057] Step 12, according to the component geometric information, material information and boundary conditions, a deterministic conversion rule is established, wherein the load information is a stress data sequence varying with time, in order to avoid structural differences between data from different sources, the deterministic conversion rule performs fixed conversion steps on the load information at each time, and outputs the mechanism model driving quantity, so that the mechanism model can perform state update under unified input format; the conversion rule is a deterministic rule, which ensures that the same input generates the same mechanism model driving quantity, and ensures that the digital twin still maintains consistent calculation logic when facing data from different sources;

[0058] Step 13, the mechanism model, key parameter value range, initial health state, load information, mechanism model driving quantity and deterministic conversion rule are uniformly stored in the basic data object set of the digital twin. The basic data object set, as the core storage structure of the digital twin, is used to record all the data required for forward calculation of the mechanism, update of the health state, prediction of the remaining life and generation of the maintenance strategy.

[0059] By constructing the component-level digital twin, the embodiment can realize continuous mapping of the physical component state in a unified data space, enable the digital twin to dynamically update the model state from real-time data, and further support data continuity and model controllability in the remaining life prediction and maintenance decision-making process, realize a digital twin-driven life prediction and maintenance decision-making mechanism for the whole life cycle of the aircraft, and make the prediction calculation and decision-making process not only rely on physical data, but also rely on the parameter space evolution and model consistency inside the digital twin, thereby improving the intelligence and self-consistency of the whole process.

[0060] In an embodiment of the application, a virtual-real consistency index is generated, and consistency gating is performed, and when the consistency threshold is not reached, the key parameter value range is updated, including:

[0061] Step 21, collecting aircraft component measured data in a continuous time period to form a measured data sequence, and calling a mechanism model, an initial health state, a key parameter value range and a mechanism model driving quantity obtained by a deterministic conversion rule from load information in the same time period, generating a twin model output sequence strictly corresponding to the measured data time axis, and calculating the difference between the measured data at each time and the twin model output at the corresponding time, and obtaining the virtual-real consistency index by averaging the absolute value of the difference in a fixed length time window; the virtual-real consistency index is used to quantify the deviation between the current model output of the digital twin and the actual running state of the physical component;

[0062] Step 22, comparing the virtual-real consistency index with a consistency gating threshold, when the virtual-real consistency index does not exceed the consistency gating threshold, it means that the current mechanism model and the key parameter value range of the digital twin can accurately reflect the evolution trend of the component state in the time period, and it is confirmed that the current mechanism model and the key parameter value range meet the consistency requirement, and the existing parameters can be maintained to continue the subsequent calculation;

[0063] Step 23, when the virtual-real consistency index exceeds the consistency gating threshold, it indicates that the output trend of the mechanism model inside the digital twin deviates from the actual state of the aircraft component in this time period, and the value range of the key parameters used by the digital twin needs to be quantitatively corrected; according to the size and sign of the difference between the virtual-real consistency index and the consistency gating threshold, the key parameters to be adjusted and their corresponding adjustment direction are determined. The sign of the difference is used to determine whether the response trend of the current mechanism model is higher or lower than the actual component operating state, so as to determine the direction in which the upper and lower limits of the key parameter value range should be shifted inward. After determining the shift direction, the embodiment performs an inward convergence update operation on the upper and lower limits of the key parameter value range according to the preset interval contraction rule. Specifically, the upper and lower limits are respectively moved inward by a preset offset, so that the key parameter value range shrinks in the direction closer to the actual state while maintaining physical reasonableness. To avoid excessive contraction of the key parameter value range, the present embodiment compares the difference between the updated upper and lower limits after completing the offset operation, and when the difference is less than the preset minimum interval width, it is reset to the minimum interval width, thereby ensuring the stability of the key parameter value range in numerical sense. After completing the above interval contraction, the updated key parameter value range is written into the basic data object set of the digital twin, and is used for the forward calculation of the mechanism model in the next round. In this way, the digital twin can adaptively converge the parameter space based on real-time data, so that the model behavior gradually approaches the true state of the physical component over time, forming a dynamic consistency maintenance mechanism for remaining life prediction and maintenance decision-making.

[0064] By executing the above virtual-real consistency calibration link, the present embodiment can form a dynamic parameter self-calibration mechanism based on real-time data feedback inside the digital twin, so that the digital twin can continuously maintain an approximate reflection of the true state of the component, providing an accurate and stable model basis for subsequent task spectrum deduction, remaining life prediction and maintenance scheme generation, so that the present invention has higher computational reliability and adaptability in the state reasoning and decision-making process of the aircraft component throughout its life cycle.

[0065] In an embodiment of the present application, the flight task is divided into a task spectrum scene set, the load upper limit and the load lower limit are set, and the upper and lower limits of the mechanism model driving quantity are obtained according to the deterministic conversion rule, including:

[0066] Step 31, obtaining the task information of the aircraft during future flight generates a flight task information sequence, parses according to flight phase, attitude parameter, speed parameter and external load parameter, and splits the flight task information sequence into a plurality of task spectrum scene sets with independent load behavior characteristics according to the task duration; a structured description of the time interval and the load action mode is established for each task spectrum scene, the independent load behavior characteristics refer to that after the flight task information sequence is divided, each divided segment forms a relatively stable load change mode in its corresponding time interval; the load action mode is used to describe the change trend of the load with time and the action characteristics of the load in the time interval, including the load change trend and the load action type in the time interval, the load action type includes stretching, compression, bending, torsion, etc.;

[0067] Step 32, for each task spectrum scene in the task spectrum scene set, the load upper limit and the load lower limit are determined according to the flight attitude, the aerodynamic load and the environmental conditions, reflecting the maximum load and the minimum load that may occur during task execution;

[0068] Step 33, inputting the load upper limit and the load lower limit into the deterministic conversion rule respectively, obtaining the mechanism model driving quantity upper limit and the mechanism model driving quantity lower limit through the fixed conversion step, and constructing a task spectrum scene driving quantity boundary set based on the driving quantity upper limit and the driving quantity lower limit of all task spectrum scenes, the task spectrum scene driving quantity boundary set is used to trigger the forward calculation of the mechanism model inside the digital twin, and provides boundary driving data for the solution of the remaining life interval and the virtual rehearsal of the maintenance scheme.

[0069] Through the above steps, the future flight task can be converted into structured load driving data that can be directly used in the digital twin, so that the task input is converted from the original, unstructured flight state description to the task spectrum scene set with clear time interval, load action mode and load boundary. By independently determining the load upper limit and the load lower limit for each task spectrum scene, and further obtaining the mechanism model driving quantity upper limit and the mechanism model driving quantity lower limit through the deterministic conversion rule, a task spectrum scene driving quantity boundary set for forward calculation is constructed, so that the remaining life prediction and the maintenance scheme derivation remain calculation consistency during execution.

[0070] In an embodiment of the present application, the mechanism forward calculation is performed based on the task spectrum scene set and the key parameter value range, and the remaining life interval is determined according to the preset failure critical standard, including:

[0071] Step 41, for each mission spectrum scenario in the mission spectrum scenario set, call the mechanism model, the initial health state and the key parameter value range, and respectively input the upper limit and the lower limit of the mechanism model driving quantity corresponding to the mission spectrum scenario to perform the mechanism forward calculation with a fixed time step; at each time step, the mechanism model driving quantity and the parameter combination in the key parameter value range are substituted into the state update rule of the mechanism model to obtain a damage evolution time sequence, which is used to describe the state evolution process of the component under the mission spectrum scenario over time;

[0072] Step 42, compare the damage evolution time sequence with the failure threshold at each time point, and use the time point at which the failure threshold is first reached as the mission spectrum scenario failure time; obtain the upper limit of the mission spectrum scenario failure time under the upper limit of the mechanism model driving quantity, and obtain the lower limit of the mission spectrum scenario failure time under the lower limit of the mechanism model driving quantity, and form a mission spectrum scenario failure time interval; since the upper limit of the mechanism model driving quantity may lead to faster damage accumulation, and the lower limit of the mechanism model driving quantity may lead to slower damage change, the present application independently performs the above forward deduction under the upper limit of the mechanism model driving quantity and the lower limit of the mechanism model driving quantity to obtain the upper limit of the mission spectrum scenario failure time and the lower limit of the mission spectrum scenario failure time. Both of them constitute the mission spectrum scenario failure time interval corresponding to the mission spectrum scenario.

[0073] Step 43, aggregate all mission spectrum scenario failure time intervals, select the minimum value of the upper limit of the mission spectrum scenario failure time as the lower limit of the remaining life, select the maximum value of the lower limit of the mission spectrum scenario failure time as the upper limit of the remaining life, and form a remaining life interval with the lower limit of the remaining life and the upper limit of the remaining life. This step enables the digital twin to uniformly and structurally predict the life evolution trend under complex and multi-scenario task conditions, and provides quantifiable life boundary data for subsequent maintenance time point determination and maintenance strategy generation.

[0074] The present embodiment enables the digital twin to have life prediction capability under multi-task and multi-load conditions through mission spectrum scenario-oriented model deduction, forms a sustainable iteration and scenario-scheduled state evolution link, and enables maintenance decisions to be based on complete task-driven life evolution, thereby improving the determinability and consistency of maintenance decisions.

[0075] In an embodiment of the present application, when calculating the residual life interval, the lower limit and the upper limit of the failure time corresponding to the entire task spectrum scenario are recorded, and in the continuous two rounds of mechanism forward calculation, the absolute difference between the lower limit of the residual life of the last round and the lower limit of the residual life of the current round and the absolute difference between the upper limit of the residual life of the last round and the upper limit of the residual life of the current round are compared with the preset convergence threshold value respectively; when the absolute difference of the two is not more than the preset convergence threshold value, it means that the life deduction result of the mechanism model in continuous iteration tends to be stable in value, and the digital twin determines that the residual life interval meets the convergence condition accordingly.

[0076] Through the convergence judgment mechanism of the present embodiment, the digital twin can perform numerical stability verification on the residual life interval in multiple rounds of mechanism model forward calculation, so that the life prediction is no longer dependent on a single deduction result, but forms a verifiable convergence state in continuous iteration, ensuring that the residual life interval has consistency and stability before entering the maintenance decision link, and improving the certainty of the decision process and the credibility of the model output.

[0077] In an embodiment of the present application, the shortest residual life is combined with the safety margin time as a safety constraint, and candidate maintenance schemes are generated according to maintenance cost, downtime and failure risk, which contain maintenance time point and maintenance level, including:

[0078] Step 51, call the lower limit of the residual life, and calculate the safety margin time by the difference between the preset safety margin coefficient and the lower limit of the residual life, determine the upper limit of the maintenance time point by subtracting the safety margin time from the lower limit of the residual life; take the current time as the lower limit of the maintenance time point, and generate candidate maintenance time points at fixed time intervals between the lower limit and the upper limit of the maintenance time point to form a feasible maintenance time point interval; the safety margin time is used to ensure that the maintenance time point will not approach the failure boundary;

[0079] Step 52, in the feasible maintenance time point interval, combine each candidate maintenance time point with the maintenance level set to form an initial set of candidate maintenance schemes; and determine the downtime based on the man-hour table corresponding to the maintenance level, determine the maintenance cost based on the resource consumption corresponding to the maintenance level, and take the reciprocal of the difference between the lower limit of the residual life and the maintenance time point as the failure risk, so that the risk measurement can reflect the characteristics that the closer the maintenance time point is to the lower limit of the life, the higher the potential risk is, and obtain the maintenance cost, downtime and failure risk of each item in the initial set of candidate maintenance schemes respectively; wherein the maintenance level is determined according to the set of maintenance processes executable by the component, each maintenance process corresponds to different health state reset rules, maintenance man-hour and cost parameters; the health state reset rule includes a deterministic update operation on the initial value of the crack, and the maintenance man-hour and cost parameters are obtained based on the fixed process resource table. The maintenance level includes: inspection level, repair level, local replacement level and whole replacement level;

[0080] Step 53, screening the initial set of candidate maintenance schemes, and including the candidate schemes whose maintenance time falls within the feasible maintenance time interval and whose failure risk is lower than the preset risk threshold into the candidate maintenance scheme set.

[0081] The embodiment can convert the life prediction result into an executable maintenance candidate scheme directly by joint reasoning of the residual life interval, the feasible maintenance time interval and the maintenance level set, and provide a model-driven decision basis for the generation of the maintenance strategy.

[0082] In an embodiment of the present application, the candidate maintenance schemes are virtually simulated, the initial state after maintenance is reset according to the maintenance level, and the mechanism forward calculation is performed, and the candidate maintenance schemes that cause the exceeding of the preset failure threshold are eliminated, including:

[0083] Step 61, for each candidate maintenance scheme in the candidate maintenance scheme set, calling the maintenance level thereof, and performing a deterministic reset operation on the initial health state before maintenance according to the health state reset rule corresponding to the maintenance level; the health state reset rule includes the reset of the initial value of the crack; and writing the initial health state after maintenance into the basic data object set as the starting state of the mechanism model forward calculation;

[0084] Step 62, calling the maintenance time point in the candidate maintenance scheme, and sequentially calling the task spectrum scene set in the time interval after the maintenance time point; for each task spectrum scene, substituting the upper limit and the lower limit of the mechanism model driving quantity into the state update rule of the mechanism model respectively, and performing iterative calculation with the time resolution of the task spectrum scene as the fixed time step to obtain the post-maintenance damage evolution time sequence, which is used to describe the damage development trend of the component in future tasks under different maintenance conditions and different task load boundaries;

[0085] Step 63, comparing the post-maintenance damage evolution time sequence with the preset failure threshold at each time point according to the fixed time step, and judging whether the damage evolution value exceeds the preset failure threshold; if any point in the sequence exceeds, it is considered that the candidate maintenance scheme cannot guarantee the component to run in a safe state under future task conditions, and is determined as damage out-of-limit, and the corresponding candidate maintenance scheme is marked as unfeasible; otherwise, the maintenance time point, the maintenance level and the corresponding post-maintenance damage evolution time sequence are included in the feasible maintenance scheme set, and the feasible maintenance scheme set is used as the input of the next stage of maintenance decision optimization and final maintenance scheme screening of the digital twin.

[0086] Through the above maintenance virtual rehearsal process, the digital twin has the ability to reconstruct the post-maintenance state evolution path in a completely virtual environment, integrates the differences in maintenance levels, the uncertainty of task loads, and the evolution trend of component states into the deducible model, enables the effectiveness of the maintenance scheme to be directly quantified in the digital space, and enables subsequent maintenance decisions to be made based on the model reasoning results of the digital twin rather than relying on single experience, thereby improving the certainty and reliability of maintenance decisions.

[0087] In an embodiment of the application, the final maintenance scheme is selected, and the state record data is collected after maintenance execution and written back to the digital twin, and the initial health state and the key parameter value range are updated, including:

[0088] Step 71, after the execution of the final maintenance scheme, the state of the aircraft component is recorded, the initial value of the crack after maintenance, the initial value of the cumulative damage after maintenance, the local geometric feature quantity and the material detection parameter are collected, and the state record data object is generated according to the preset field structure;

[0089] Step 72, calling the state record data object, writing the crack initial value, cumulative damage initial value and material detection parameter in the state record data object into the digital twin basic data object set, performing a deterministic replacement operation on the initial health state before maintenance to form the initial health state after maintenance, and taking the initial health state after maintenance as the starting state of the next round of forward calculation of the mechanism model, so that the digital twin can restart the remaining life deduction and task spectrum load analysis with the consistent health baseline as the physical component, realize the natural transition and continuity of the model state after maintenance; wherein the crack initial value after maintenance and the cumulative damage initial value after maintenance are used to represent the virtual health starting point of the component after the execution of the maintenance process, are generated based on the health state reset rules corresponding to the maintenance level, and are written into the state record data object according to the preset field structure, so that the digital twin can start the forward calculation of the mechanism model with a new health baseline after maintenance, and is not used for material detection or damage measurement, but is used as a model state initialization parameter;

[0090] Step 73, determining the key parameters that need to be corrected according to the material detection parameter and the local geometric feature quantity in the state record data object, and performing interval adjustment on the key parameter value range according to the preset interval update rule, so that the key parameters reflect the change trend of the component material properties and structural characteristics after maintenance, and the corrected key parameter value range is written into the digital twin basic data object set for consistent calibration of virtual and real in the next calculation period, task spectrum model deduction, and maintenance scheme generation.

[0091] Through the above steps, the embodiment realizes model resynchronization of the digital twin after the maintenance is completed, enables the digital twin to construct a new health state and model parameter space based on actual maintenance results, forms a closed-loop prediction structure facing the entire life cycle, and ensures that the next round of remaining life prediction and maintenance decision is executed based on the latest and reliable state input.

[0092] It should be noted that the interval and threshold size are set for ease of comparison, wherein the size of the threshold depends on the amount of sample data and the base number set by the person skilled in the art for each group of sample data, as long as the proportional relationship of the parameters and the quantized values is not affected. And the above formula is a dimensionless calculation of the value, and the formula is obtained by software simulation of a large amount of data to obtain a formula of the nearest true situation, and the preset parameters in the formula are set by the person skilled in the art according to the actual situation.

[0093] The embodiments of the application are described above, but the application is not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative but not limiting, and a person skilled in the art can make many forms under the inspiration of the embodiments, which all belong to the protection of the embodiments.

Claims

1. A method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins, characterized in that, Includes the following steps: Step 1: Establish a mechanism model, determine the range of key parameter values, the initial health state, and the deterministic conversion rules for converting load information into mechanism model driving quantities; Step 2: Collect actual measurement data of components, generate virtual-physical consistency indicators, and perform consistency gating. When the consistency threshold is not reached, the range of key parameter values ​​is narrowed and updated, including: Step 21: Collect measured data of components within a continuous time period to form a measured data sequence. Within the same time period, call the mechanism model, initial health state, key parameter value range, and mechanism model driving quantity obtained from load information through deterministic conversion rules to generate a twin model output sequence. Calculate the difference between the measured data at each moment and the twin model output at the corresponding moment. Average the absolute value of the difference within a fixed-length time window to obtain the virtual-real consistency index. Step 22: Compare the virtual-real consistency index with the consistency threshold. When the virtual-real consistency index does not exceed the consistency threshold, confirm that the current mechanism model and the range of key parameter values ​​meet the consistency requirements. Step 23: When the virtual-real consistency index exceeds the consistency gate threshold, determine the key parameters that need to be adjusted based on the magnitude and sign of the difference between the two, and tighten and update the upper and lower limits of the key parameter value range according to the preset interval shrinkage rule, and write the updated key parameter value range into the basic data object set of the digital twin. Step 3: Decompose the flight mission into a mission spectrum scenario set, set the upper and lower limits of the load, and obtain the upper and lower limits of the driving quantity of the mechanism model according to the deterministic conversion rules. Step 4: Perform forward mechanistic calculations based on the task spectrum scenario set and the value range of key parameters, and determine the remaining lifetime range according to the preset failure critical criteria. Step 5: Using the shortest remaining lifespan combined with the safety margin time as a safety constraint, generate candidate maintenance plans based on maintenance costs, downtime, and failure risks. The candidate maintenance plans include the maintenance timing and maintenance level. Step 6: Conduct virtual drills for candidate maintenance solutions, reset the initial state after maintenance according to maintenance level and perform forward mechanism calculations to eliminate candidate maintenance solutions that cause the failure threshold to be exceeded. Step 7: Select the final maintenance plan. After the maintenance is performed, collect the status record data and write it back to the digital twin, and update the initial health status and key parameter value range.

2. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, Establish a mechanistic model, determine the range of key parameter values, the initial health state, and a deterministic conversion rule for converting load information into mechanistic model driving quantities, including: Step 11: Establish a mechanism model, and based on the allowable range of material constants, component geometric parameters and damage sensitivity coefficients, determine the value range of key parameters required for the mechanism model, and record the initial crack value of the current component as the initial healthy state; Step 12: Establish deterministic conversion rules based on component geometry, material information and boundary conditions. The load information is a stress data sequence that changes over time. The deterministic conversion rules perform fixed conversion steps on the load information at each moment and output the mechanism model driving quantity. Step 13: Store the mechanism model, key parameter value range, initial health state, load information, mechanism model driving quantity, and deterministic conversion rules into the basic data object set of the digital twin.

3. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, In step 23, when tightening and updating the range of key parameter values, the offset direction of the upper and lower limits of the key parameter value range is determined according to the sign of the difference between the virtual-real consistency index and the consistency gate threshold, and the upper and lower limits are offset inward by a preset offset amount. After offsetting, the difference between the upper and lower limits is compared. When the difference is less than the preset minimum interval width, the difference is reset to the minimum interval width to form the updated range of key parameter values.

4. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, Flight missions are broken down into a set of mission spectrum scenarios, upper and lower load limits are set, and upper and lower limits of the driving quantities of the mechanistic model are obtained according to deterministic conversion rules, including: Step 31: Obtain mission information during future flights to generate a flight mission information sequence. Analyze the sequence according to flight phase, attitude parameters, velocity parameters, and external load parameters. Based on the mission duration, split the flight mission information sequence into multiple mission spectrum scenario sets with independent load behavior characteristics. Establish a structured description of the time interval and load action mode for each mission spectrum scenario. The load action mode includes the load change trend and load action type within the time interval. Step 32: For each mission spectrum scenario in the mission spectrum scenario set, determine the upper and lower limits of the load based on the flight attitude, aerodynamic load, and environmental conditions. Step 33: Input the upper and lower limits of the load into the deterministic conversion rules respectively, and obtain the upper and lower limits of the driving quantity of the mechanism model through fixed conversion steps. Construct the boundary set of the driving quantity of the task spectrum scenario based on the upper and lower limits of the driving quantity of the driving quantity of the entire task spectrum scenario.

5. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, Mechanism forward calculation is performed based on the task spectrum scenario set and the value range of key parameters. The remaining lifetime range is determined according to the preset failure criticality criteria, including: Step 41: For each task spectrum scenario in the task spectrum scenario set, call the mechanism model, initial health state and key parameter value range, and take the upper limit and lower limit of the mechanism model driving quantity corresponding to the task spectrum scenario as input, and perform mechanism forward calculation with a fixed time step; at each time step, substitute the mechanism model driving quantity and the parameter combination within the key parameter value range into the state update rule of the mechanism model to obtain the damage evolution time series; Step 42: Compare the damage evolution time series with the failure threshold value time by time, and take the time point when the failure threshold value is first reached as the failure time of the task spectrum scenario; under the condition of the upper limit of the driving quantity of the mechanism model, the upper limit of the failure time of the task spectrum scenario is obtained, and under the condition of the lower limit of the driving quantity of the mechanism model, the lower limit of the failure time of the task spectrum scenario is obtained, and the failure time interval of the task spectrum scenario is formed. Step 43: Summarize the failure time intervals of all task spectrum scenarios, select the minimum value of the upper limit of the failure time of the task spectrum scenarios as the lower limit of the remaining lifetime, select the maximum value of the lower limit of the failure time of the task spectrum scenarios as the upper limit of the remaining lifetime, and form the remaining lifetime interval by combining the lower limit of the remaining lifetime and the upper limit of the remaining lifetime.

6. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 5, characterized in that, When calculating the remaining lifetime interval, the lower and upper limits of failure time corresponding to all task spectrum scenarios are recorded. In two consecutive rounds of forward mechanism calculations, the absolute difference between the lower and current remaining lifetime limits of the previous round and the absolute difference between the upper and current remaining lifetime limits of the previous round are compared with preset convergence thresholds. When the absolute difference between the two does not exceed the preset convergence threshold, the remaining lifetime interval is determined to meet the convergence condition.

7. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, Candidate maintenance solutions are generated based on maintenance costs, downtime, and failure risks, including: Step 51: Call the remaining life lower limit, and calculate the safety margin time based on the difference between the preset safety margin coefficient and the remaining life lower limit. Subtract the safety margin time from the remaining life lower limit to determine the upper limit of the maintenance time point. Use the current time as the lower limit of the maintenance time point, and generate candidate maintenance time points between the lower limit and the upper limit of the maintenance time point at fixed time intervals to form a feasible maintenance time point interval. Step 52: Within the feasible maintenance timeframe, combine each candidate maintenance timeframe with the maintenance level set to form an initial set of candidate maintenance solutions; determine the downtime based on the timetable corresponding to the maintenance level, determine the maintenance cost based on the resource consumption corresponding to the maintenance level, and use the reciprocal of the difference between the remaining lifespan lower limit and the maintenance timeframe as the failure risk quantity. Obtain the maintenance cost, downtime, and failure risk quantity for each item in the initial set of candidate maintenance solutions; wherein, the maintenance levels include: inspection level, repair level, partial replacement level, and overall replacement level; Step 53: Filter the initial set of candidate maintenance solutions and include the candidate maintenance solutions whose maintenance time falls within the feasible maintenance time interval and whose failure risk is lower than the preset risk threshold into the candidate maintenance solution set.

8. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, Virtual simulations were conducted on candidate maintenance plans. The initial post-maintenance state was reset according to maintenance level, and forward mechanistic calculations were performed to eliminate candidate maintenance plans that would cause them to exceed the preset failure threshold. These included: Step 61: For each candidate maintenance scheme in the candidate maintenance scheme set, call its maintenance level, and perform a deterministic reset operation on the initial health state before maintenance according to the health state reset rule corresponding to the maintenance level; the health state reset rule includes resetting the initial value of the crack; write the initial health state after maintenance into the basic data object set as the starting state for the forward calculation of the mechanism model. Step 62: Call the maintenance time point in the candidate maintenance scheme, and call the task spectrum scenario set in sequence within the time interval after the maintenance time point; for each task spectrum scenario, substitute the upper limit and lower limit of the mechanism model driving quantity into the state update rule of the mechanism model respectively, and perform iterative calculation with the time resolution of the task spectrum scenario as a fixed time step to obtain the damage evolution time series after maintenance. Step 63: Compare the post-repair damage evolution time series with the preset failure threshold at fixed time steps to determine whether the damage evolution value exceeds the preset failure threshold. If any point in the series exceeds the threshold, it is determined that the damage exceeds the limit, and the corresponding candidate repair scheme is marked as infeasible; otherwise, the repair time point, repair level and its corresponding post-repair damage evolution time series are included in the set of feasible repair schemes.

9. The method for predicting the remaining life of aircraft components and making maintenance decisions based on digital twins according to claim 1, characterized in that, After selecting the final maintenance plan and executing the maintenance, the collected status data is written back to the digital twin, and the initial health status and key parameter value ranges are updated, including: Step 71: After the final repair plan is executed, the status of the component is recorded, and the initial value of the crack after repair, the initial value of the cumulative damage after repair, the local geometric feature quantity and the material detection parameters are collected. The status record data object is generated according to the preset field structure. Step 72: Call the state record data object, write the initial value of crack, the initial value of cumulative damage and the material detection parameters into the digital twin basic data object set, perform a deterministic replacement operation on the initial health state before repair to form the initial health state after repair, and use the initial health state after repair as the starting state for the next round of forward calculation of the mechanism model. Step 73: Determine the key parameters that need to be corrected based on the material detection parameters and local geometric features in the status record data object, and perform interval adjustment on the value range of the key parameters according to the preset interval update rules, and write the corrected value range of the key parameters into the digital twin basic data object set.

Images