Fleet load-crack joint evaluation method based on bayesian update

Abstract

This invention proposes a joint assessment method for fleet mission load and crack state based on Bayesian updates, belonging to the fields of aircraft structural safety, aircraft structural fatigue crack propagation tracking, and aircraft load assessment. The method includes: converting the random load spectrum under different mission histories of the fleet into equivalent constant-amplitude loads using an equivalent load method based on the Paris formula, forming a load level that can be quantified for subsequent analysis; establishing a mission-labeled load library and a fatigue crack damage propagation model; constructing a Bayesian network based on network nodes and interdependencies to build a probabilistic joint assessment architecture for mission load and crack state; using a particle filter algorithm as the inference algorithm for the dynamic Bayesian network to jointly assess the fleet mission load and structural crack state; and testing and verifying the aforementioned network architecture and method in a pre-defined fleet. This invention provides real-time assessment of the load levels of different missions of the fleet, thereby ensuring the safety of the aircraft structure.

Description

Technical Field

[0001] This invention belongs to the fields of aircraft structural safety, aircraft structural fatigue crack propagation tracking, and aircraft load assessment, specifically involving a joint assessment method for fleet mission load and crack state based on Bayesian updates. Background Technology

[0002] Ensuring the safety of aircraft structures is one of the core tasks in the use and maintenance of aviation equipment. Structural fatigue has long been a significant factor contributing to aviation accidents, with many catastrophic incidents closely related to the uncontrolled propagation of fatigue cracks. Therefore, during the service life of an aircraft, effectively tracking and assessing damage to fatigue-sensitive components of critical structures has become a crucial link in ensuring flight safety. The most fundamental and critical task in damage tracking and life assessment is accurately determining the fatigue load levels the structure experiences during actual service. Since loads directly drive crack initiation and propagation, accurate load assessment and design are essential for accurately evaluating aircraft structural fatigue damage, establishing reasonable structural inspection intervals, and thus ensuring the structural integrity and service safety of the aircraft.

[0003] Currently, structural inspection intervals for large transport aircraft and fighter jets are typically determined based on the static load spectrum provided during the design phase, which serves as the design load spectrum. However, the design load spectrum is often derived under idealized assumptions and fails to fully consider the variability in loads experienced by the aircraft during actual missions. Especially in the context of flexible mission formations and changing operational scenarios in modern times, static load assumptions lead to overly conservative or aggressive inspection interval settings, affecting the efficiency of maintenance resource allocation and creating structural safety hazards. Therefore, there is an urgent need for a dynamic load assessment method that can reflect the impact of actual flight missions, thereby accurately assessing fatigue crack conditions and rationally assisting in the determination of structural inspection intervals.

[0004] Although ensuring structural safety urgently requires the support of dynamic load data, in practical engineering applications, directly obtaining the load levels of key structural components still faces many technical and economic challenges, as strain sensors are rarely deployed on current aircraft. Therefore, how to fully utilize available data resources for reverse analysis when structural loads are difficult to measure directly has become a research direction with practical engineering significance. Summary of the Invention

[0005] Compared to structural loads, crack data obtained during regular structural inspections of aircraft is relatively abundant and readily available, and its mission execution history is typically recorded in detail in the flight management system. Crack propagation is a direct result of load application; therefore, a crack propagation model can be used to establish the relationship between mission loads and structural damage, enabling joint inversion and dynamic evaluation of structural load levels and crack states under different mission conditions. Addressing the load differences caused by mission variations in actual operation, this invention provides a method, system, and device for joint evaluation of fleet mission loads and crack states based on Bayesian updates. A dynamic Bayesian inference architecture for joint evaluation of equivalent loads and fatigue crack states is constructed. The Bayesian update method achieves joint evaluation of equivalent loads and fatigue crack states, aiming to realize joint modeling, dynamic inference, and real-time updating of aircraft structural loads and crack propagation states under different mission conditions. The proposed method for joint evaluation of fleet loads and crack states possesses mission-oriented modeling capabilities, an inference mechanism that integrates observation and model information, and application value supporting condition-based structural maintenance, thus supporting the maintenance of aircraft structural integrity.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A joint evaluation method for fleet mission load and crack state based on Bayesian update includes the following steps:

[0008] Step 1: Construct a mission stochastic spectrum equivalent load model to transform the complex stochastic load spectrum of the fleet under different mission trajectories into a constant amplitude load expression with equivalent fatigue effect, and establish a mission load database that can be quantified and analyzed.

[0009] Step 2: Integrate task information, equivalent load, and crack propagation model to construct a Bayesian network with nodes and dependencies, representing the evolution path and logical relationship between load, state, and observation;

[0010] Step 3: Use particle filtering as the inference mechanism of dynamic Bayesian network to realize joint probability estimation and dynamic update of task load level and structural crack state.

[0011] Step 4: Evaluate the effectiveness of the invention in load level inversion and crack state tracking by conducting verification in typical fleet cases.

[0012] The present invention also provides an electronic system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method for joint evaluation of fleet mission load and crack state based on Bayesian update.

[0013] The present invention also provides a non-transitory computer-readable storage medium device having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for joint evaluation of fleet mission load and crack state based on Bayesian update.

[0014] Beneficial effects:

[0015] This invention employs a task load tagging dynamic read / write method based on Bayesian networks. By predicting and diagnosing the load and crack state of the fleet aircraft based on crack length observation results, it can more accurately estimate the load and crack of the fleet aircraft, thus ensuring the structural safety of the fleet aircraft. Attached Figure Description

[0016] Figure 1 This is a flowchart of a joint evaluation method for fleet mission load and crack state based on Bayesian update according to the present invention;

[0017] Figure 2 A schematic diagram illustrating fleet mission load level updates;

[0018] Figure 3a , Figure 3b , Figure 3c This is a schematic diagram illustrating the changes in payload distribution for three different tasks; among them, Figure 3a For the change in payload distribution of the first mission, Figure 3b For the change in payload distribution of the second mission, Figure 3c The payload distribution changes for the third mission;

[0019] Figure 4 This is a diagram showing the evolution of the mission payload.

[0020] Figure 5 The diagram illustrates the crack propagation tracing of the aircraft fleet; where (a) represents the crack propagation tracing of the first aircraft, (b) represents the crack propagation tracing of the second aircraft, (c) represents the crack propagation tracing of the third aircraft, (d) represents the crack propagation tracing of the fourth aircraft, (e) represents the crack propagation tracing of the fifth aircraft, (f) represents the crack propagation tracing of the sixth aircraft, and (g) represents the crack propagation tracing of the seventh aircraft. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0022] like Figure 1As shown, the joint evaluation method for fleet mission load-crack state based on Bayesian update of the present invention includes the following steps:

[0023] Step 1: Use the equivalent constant amplitude load determined by the generalized root mean square method as the load level.

[0024] Step 2: Establish a database of task load levels and a fatigue crack damage propagation model.

[0025] Step 3: Construct a Bayesian network based on each node and its dependencies.

[0026] Step 4: Considering uncertainties, a particle filtering method is used to jointly evaluate the aircraft mission load and crack state.

[0027] Specifically, step 1 includes:

[0028] First, the task load level is equivalently represented to facilitate subsequent crack propagation prediction and load level updates within the Bayesian network using quantified load levels. To characterize the impact of these random load spectra on crack propagation, it is assumed that the stress values ​​at structural details are linearly related to the loads applied to the structure. The equivalent constant amplitude load value is determined using the root mean square (RMS) method of the m-parameter in the Paris formula, as recommended in British PD6493-1991.

[0029] (1)

[0030] in, It is an equivalent constant amplitude stress. Represents the stress amplitude at each level. Represents the number of stress cycles. Paris Formula The parameters in the formula are: and These are empirical parameters related to materials and the environment. It is the crack length. It is the number of load cycles. This refers to the stress intensity factor amplitude. Based on this method, the random load spectra under different tasks are processed into equivalent constant amplitude loads. Simultaneously, the random load spectra under different tasks are labeled to explore the impact of different tasks on structural crack propagation.

[0031] Specifically, step 2 includes:

[0032] First, the unknown load levels under different mission conditions of the fleet are represented by equivalent constant-amplitude loads, and the mission load levels of the fleet are labeled to establish a mission load level database. Each mission type corresponds to a load level feature. This database not only provides a prior distribution for subsequent inference models, but also provides a data foundation and storage environment for subsequent updates to mission load levels.

[0033] In this invention, any applicable crack propagation model can be used to perform step-by-step calculations of crack propagation, wherein the time step is set to... The state vector at is As the time step progresses (load cycle increment is...), The state transition equation of the crack propagation model for:

[0034] (2)

[0035] in, , They represent in Time and The crack length value at time [time]. Represented as Time's up The number of load cycles experienced at any given time. , Indicates respectively in Time and The load value at time, and , They represent in Time and The parameters in the crack propagation model at time t. For time step Process errors introduced during state transitions at different times.

[0036] Considering observation error, i.e., the uncertainty introduced by damage detection, and assuming that the measured values ​​follow a normal distribution around the true values ​​of the damage state parameters, the corresponding measurement equation can be obtained as follows:

[0037] (3)

[0038] in, yes Crack observation values ​​at time [time] It is a random variable that follows a normal distribution. It is the standard deviation of the measurement result, and its value is determined by the accuracy of the measuring tool.

[0039] Specifically, step 3 includes:

[0040] After completing the representation of equivalent load levels, database construction, and crack propagation modeling, this invention further constructs a causal relationship structure between "task-load-state-observation" based on Bayesian networks, enabling joint modeling and dynamic reasoning of structural states. A Bayesian network is a directed acyclic graph structure where each node represents a system variable, such as task category, load level, crack state, or observation result, and edges represent causal or conditional dependencies between variables. The dependencies between nodes are described through probability transition functions and conditional probabilities.

[0041] The Bayesian network structure in this invention mainly includes the following types of nodes:

[0042] Task node: Indicates the task label or task type currently being performed by the aircraft, which determines the selection of the labeled payload level;

[0043] Load node: The equivalent constant amplitude load level under the corresponding task. Here, the load distribution corresponding to the task label will be assigned to the node from the current task load database based on the prior information of the task node.

[0044] Crack state node: Represents the length or state of fatigue cracks in the current structure, which evolves according to the Paris model, load level, and material parameters;

[0045] Observation node: Represents the actual crack inspection data obtained, including observation errors.

[0046] In this process, the task nodes determine the prior distribution of the load nodes, the load nodes drive the crack propagation model to determine the crack state, and the observation results of the observation nodes diagnose and update the crack state nodes and load nodes. The joint assessment process of load level and crack propagation for the fleet aircraft is as follows: Figure 2 As shown:

[0047] In the initial stage, all aircraft (Aircraft 1, Aircraft 2, ..., Aircraft n) have the same initial load level distribution. After the crack inspection data of a certain aircraft is entered into the Bayesian network, the crack state of that aircraft is updated, and the task load distribution (Task 1 load, Task 2 load, ..., Task n load) under the corresponding flight mission is also updated and stored in the task load estimation database. When an aircraft in the fleet undergoes a mission switch, the load distribution is extracted from the task load estimation database based on the task label, and the current load node is updated. Simultaneously, when a task label is missing, the load distribution is temporarily stored and retained. For the same task label within the same time period, a load distribution fusion method is adopted.

[0048] Specifically, step 4 includes:

[0049] In this invention, a particle filtering method based on Sequential Importance Sampling (SIS) is used as the inference algorithm for dynamic Bayesian networks to achieve model diagnosis and updating based on measured data.

[0050] In this problem, it was used as the state variable in the simulation model. The evolution of the sequence is used to track the damage state and observe the sequence. Estimate the posterior distribution of the state This enables the prediction and diagnosis of damage propagation behavior based on prior distribution. ,sampling There are 3 weighted particles, each with an equal initial weight:

[0051] (4)

[0052] in, This represents the state variable vector of the system at the initial moment. Represents the distribution from the initial state The first one drawn from One sample, For the first Each particle at time step The state vector at that point, For the initial first The weight values ​​of each particle. As time progresses, the prediction process is based on the state transition equation. Generate predicted particles:

[0053] (5)

[0054] in, It is a known time step Under the condition of the particle state at that time step State vector at time step Distribution; For time step The first The state vector of each particle This refers to noise during the state transition process.

[0055] At this point, the prior distribution is approximately:

[0056] (6)

[0057] in, For time step First The weight of each particle, For the delta function, Indicates the time from the initial moment to The measurement sequence of time points.

[0058] At this point, based on state and observation sequence Based on the Bayesian update principle, the particle weights are updated:

[0059] (7)

[0060] in, This represents the observational likelihood, i.e., the likelihood of a particle in state 1. At that time, observations were generated. The probability of; Represents the state transition probability, describing the system's transition from the previous state. Evolved to the current state The probability of; As a suggested density (importance density), in a typical standard particle filtering method, we take... That is, it depends only on the state transition model.

[0061] Based on this, the updated weights are normalized to obtain:

[0062] (8)

[0063] in, These are the normalized particle weights. At this point, the posterior distribution of the model's state variables is the result of a weighted correction of the prior distribution using Bayes' theorem, and can be approximated by a weighted particle set:

[0064] (9)

[0065] Finally, according to the normalized weights Resampling is performed to generate a new set of equally weighted particles:

[0066] (10)

[0067] It can be seen that the core of the particle filtering inference process lies in the transition of particle states and the updating of particle weights. Based on this method and the aforementioned Bayesian network architecture, this invention performs collaborative estimation and updating of task load and crack state values. First, in the update step of particle filtering, the weights of the particles are updated according to the particle distribution and observations, and particle resampling is performed based on the weights to update the distribution of load parameters in the particles. The corresponding load distribution is then updated in the task load database according to the current aircraft's task label. Throughout the aircraft's lifecycle, this value will be continuously updated based on observations. Second, when the aircraft's task changes, the load value is extracted from the task load database according to the task label and replaced with the value in the current particles based on the original particle order. The variables are then further evolved within the Bayesian network. Finally, during fleet operation, the distribution of mission loads in the mission load database and the crack propagation history of the fleet's aircraft will be displayed in real time as visual images.

[0068] Example:

[0069] In this embodiment, the process and method of this invention are used to conduct application tests on a preset fleet to verify the feasibility of the process and the accuracy of prediction, diagnosis, and updating. The preset fleet has a full lifecycle of 70,000 cycles, with an inspection performed every 5,000 cycles. The fleet consists of 7 aircraft, and the material parameters of each aircraft are as follows. Both are 2.97. The logC value, the aircraft's mission history, and the change times are as follows:

[0070] Table 1 Fleet Mission History

[0071] Based on the calculation of the equivalent constant amplitude load, and using the overload exceedance number curve for each task and the Monte Carlo rejection sampling method, the actual equivalent constant amplitude load levels are as follows: Task 1 load: 20 MPa; Task 2 load: 25 MPa; Task 3 load: 30 MPa.

[0072] Therefore, based on the particle filtering method and the established Bayesian network model, the mission load values ​​of the fleet are evaluated, and the crack propagation of each aircraft in the fleet is tracked. The crack propagation model adopted here is the Paris model.

[0073] (11)

[0074] in, It is the crack length. and These are empirical parameters related to material crack propagation in an environment-dependent manner. Representing the stress intensity amplitude, the state vector at time step t is: As the time step progresses (load cycle increment is...), The state transition equation of the crack propagation model for:

[0075] (12)

[0076] Among them, the number of particles in particle filtering The initial load distribution is Gaussian, with a mean μ=25 and a standard deviation σ=4.

[0077] Based on the established Bayesian network and Bayesian update method, the Gaussian distribution of the task load is obtained as follows: Figure 3a , Figure 3b , Figure 3c As shown.

[0078] like Figure 4 As shown, the Gaussian distribution of the task load level at 70,000 load cycles is as follows:

[0079] Task 1 load distribution: mean μ = 19.96, standard deviation σ = 2.30;

[0080] Task 2 load distribution: mean μ = 23.94, standard deviation σ = 2.14;

[0081] Task 3 load distribution: mean μ = 30.12, standard deviation σ = 2.89;

[0082] Similarly, crack propagation tracking of aircraft in this fleet is as follows: Figure 5 As shown, where, Figure 5 (a) shows the crack propagation tracing of the first aircraft. Figure 5 (b) shows the crack propagation tracing of the second aircraft. Figure 5 (c) shows the crack propagation tracing of the third aircraft. Figure 5 (d) represents the crack propagation tracking of the fourth aircraft. Figure 5 (e) is the crack propagation tracking of the fifth aircraft. Figure 5 (f) represents the crack propagation tracking of the 6th aircraft. Figure 5 (g) is the crack propagation tracking of the 7th aircraft.

[0083] Preferably, the applied models include, but are not limited to, models that differentiate between mission loads, such as those for aircraft fatigue-sensitive structures.

[0084] Preferably, the crack propagation model used in this invention includes, but is not limited to, the Paris model, the NASGRO model, and other crack propagation models. The methods used for equivalent load levels include, but are not limited to, the root mean square method and the generalized root mean square method based on Paris law parameters.

[0085] The present invention also provides an electronic system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method for joint evaluation of fleet mission load and crack state based on Bayesian update.

[0086] The present invention also provides a non-transitory computer-readable storage medium device having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for joint evaluation of fleet mission load and crack state based on Bayesian update.

[0087] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0088] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0092] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A joint evaluation method for fleet mission load-crack state based on Bayesian update, characterized in that, Includes the following steps: Step 1: Construct a mission stochastic spectrum equivalent load model to transform the complex stochastic load spectrum of the fleet under different mission trajectories into a constant amplitude load expression with equivalent fatigue effect, and establish a mission load database that can be quantified and analyzed. Step 2: Integrate task information, equivalent load, and crack propagation model to construct a Bayesian network with nodes and dependencies, representing the evolution path and logical relationship between load, state, and observation; Step 3: Use particle filtering as the inference mechanism of dynamic Bayesian network to realize joint probability estimation and dynamic update of task load level and structural crack state. Step 4: Evaluate the effectiveness of the invention in load level inversion and crack state tracking by conducting verification in typical fleet cases.

2. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, Step 1 includes: extracting the entire mission history of each aircraft based on the fleet's mission plan and flight management system records, and uniquely identifying the random load spectrum using mission tags, so that all subsequent assessments of load levels and crack states are carried out using mission tags as an index.

3. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, Step 1 includes: calculating the equivalent constant amplitude load for all random load spectra under the same task label, and storing the obtained equivalent values ​​in the task load level database; the database uses the task label as the key and the equivalent constant amplitude load as the value to form a prior distribution of task loads that can be directly called by subsequent Bayesian network nodes.

4. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, In step 2, when constructing the Bayesian network, the nodes and dependencies are established in the following order: first, the task node points to the load node, then the load node points to the crack state node, and finally the crack state node points to the observation node, forming a directed connection sequence that limits the task.

5. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 4, characterized in that, The directed connection sequence defines the causal process of task-determined load, load-driven crack propagation, and crack state generating observation data.

6. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, Step 3 includes: using a particle filtering algorithm, at each observation time, first predicting the crack state particles based on the state transition model, then updating the particle weights using the observation data, and then resampling to obtain a new set of equally weighted particles; this order of prediction, updating, and resampling constitutes a dynamic loop, realizing the synchronous joint estimation of load level and crack state.

7. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, Step 3 further includes: writing the updated load level distribution back to the mission load level database and replacing the old distribution with the mission tag as the index; when the aircraft mission is switched, immediately reading the latest load distribution with the corresponding mission tag from the database and injecting it into the particle filter to form a closed-loop feedback of "database update-mission switch-distribution injection".

8. The method for joint evaluation of fleet mission load and crack state based on Bayesian update as described in claim 1, characterized in that, In step 4, the verification process is executed in the following order: first, a known mission history and crack inspection plan are set in a typical fleet, then steps 1-3 are run to obtain the prediction results, and finally the prediction results are compared with the measured crack length. This order of setting, running and comparing is used to confirm the consistency between load level inversion and crack state tracking.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the joint evaluation method for fleet mission load-crack state based on Bayesian update as described in any one of claims 1 to 8.

10. A non-transitory computer-readable storage medium device having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a joint evaluation method for fleet mission load-crack status based on Bayesian update as described in any one of claims 1 to 8.

Images