Digital twinborn monitoring method for full-life health state of service aircraft structure

Abstract

A digital twinning monitoring method for the full-life health state of a service aircraft structure belongs to the related technical field of digital twinning, and comprises the following steps: firstly, designing a health state data acquisition scheme to obtain health state data in a real flight state; secondly, performing screening extraction and optimization treatment on the health state data, and constructing a flight data-mechanical response data health state data set of a service aircraft structure; thirdly, establishing a flight data-mechanical response data-life monitoring digital twinborn model, and realizing high-precision prediction of the fatigue residual life of a service aircraft structure; and finally, making an inspection and maintenance task intelligent planning scheme, and dividing inspection and maintenance grades based on the fatigue residual life and the fatigue reliability coefficient. According to the method, fatigue residual life prediction and inspection and maintenance task intelligent planning of the service aircraft structure can be realized, a service aircraft structure full-life health state digital twin monitoring process is formed, and full-life health state monitoring and management can be conveniently carried out for the service aircraft.

Description

Technical Field

[0001] The present invention belongs to the technical field related to digital twins and relates to a digital twin monitoring method for the health status of an in-service aircraft structure throughout its life cycle. Background Art

[0002] Aircraft structural health monitoring technology records aircraft health data during flight, calculates fatigue damage, and estimates the remaining life of the structure, enabling aircraft structural health monitoring and assessment. This technology plays a crucial role in structural design and maintenance planning. Traditional methods rely on deploying sensors on flight or ground test aircraft to monitor structural mechanical response data and determine structural health. For example, Chinese invention patent CN117326090A discloses an online health monitoring device and method for light aircraft landing gear structures. This device, based on ground testing, measures the correlation between different crack forms and the number of load cycles, establishing relationships between different crack forms, ultrasonic waveforms, and the remaining life of the landing gear forks, enabling health monitoring. However, this method is limited by factors such as spatial layout, preventing sensors from fully covering critical areas of the aircraft structure, making it difficult to monitor the full range of the structural mechanical response. Furthermore, this method fails to account for differences in individual aircraft service under real-world flight conditions and can only reflect the mechanical response of the monitored aircraft structure under specific flight conditions. This results in inaccurate and costly health monitoring and analysis.

[0003] Digital twin technology simulates actual motion processes by integrating real data, establishing data connections between physical and digital entities, and achieving real-time monitoring and analysis of the real state of physical entities, thereby realizing simulation, monitoring, and prediction of physical entities. Therefore, using digital twin technology to achieve high-precision simulation of real mechanical responses and then achieve assessment and prediction of remaining life provides a solution for realizing intelligent monitoring of the health status of in-service aircraft structures throughout their life cycle. For example, Chinese invention patent CN117077327A discloses a bearing life prediction method and system based on digital twins, which uses real-time collected bearing state parameters to establish a digital twin model and obtain life data under multiple working conditions. However, in-service aircraft have strict onboard requirements and a huge amount of health status data. How to obtain the health status data of in-service aircraft under real-time flight conditions and construct a high-precision life monitoring digital twin model to achieve accurate remaining life assessment still requires further research.

[0004] Therefore, based on the above problems, it is necessary to design a health status data collection scheme for in-service aircraft structures to obtain health status data under real flight conditions. The health status data includes mechanical response data and flight data; it is necessary to screen, extract and optimize the flight data and mechanical response data to construct a flight data-mechanical response data health status data set for in-service aircraft structures; it is necessary to establish a flight data-mechanical response data-life monitoring digital twin model to achieve high-precision prediction of the fatigue remaining life of in-service aircraft structures; it is necessary to formulate an intelligent planning scheme for inspection and maintenance tasks, and divide the inspection and maintenance levels based on fatigue remaining life and fatigue reliability coefficient. Summary of the Invention

[0005] To address the challenges of existing technologies, the present invention provides a digital twin monitoring method for the lifecycle health status of in-service aircraft structures. This digital twin monitoring method collects health status data of in-service aircraft structures under real-world flight conditions, screens, extracts, and optimizes this data, constructs a flight data-mechanical response data-lifecycle monitoring digital twin model, and conducts fatigue residual life prediction for in-service aircraft structures. This allows for intelligent planning of inspection and maintenance tasks, enabling lifecycle health status monitoring of in-service aircraft structures. This facilitates lifecycle health monitoring and management for in-service aircraft.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A digital twin monitoring method for the lifecycle health status of an in-service aircraft structure comprises the following steps:

[0008] The first step is to design a health status data collection solution for in-service aircraft structures to obtain health status data under real flight conditions. The health status data includes mechanical response data and flight data. Specifically:

[0009] Step 1.1: Select a mechanical response data acquisition device that meets the airworthiness requirements. On the one hand, ensure the continuity and reliability of mechanical response data acquisition during flight. The mechanical response data acquisition device must be able to withstand flight conditions such as low temperature, large temperature difference, turbulence, overload impact, etc. On the other hand, avoid affecting the normal operation of the aircraft during flight. Select a mechanical response data acquisition device that is light in weight, small in size, low in energy consumption, and resistant to electromagnetic interference.

[0010] Furthermore, the mechanical response data includes strain, stress, displacement, temperature, and acceleration; the mechanical response data acquisition equipment includes a collector, a sensor, an acquisition line, a power line, and a fixed support; and the sensor includes a strain sensor, a displacement sensor, a temperature sensor, and an acceleration sensor.

[0011] Step 1.2: Design a sensor layout and determine the locations for monitoring mechanical response data. Based on mechanical mechanisms, simulation results, and maintenance data, ensure coverage of areas of aircraft structural stress concentration and significant strain, providing accurate input for the subsequent construction of a digital twin model combining flight data and mechanical response data.

[0012] Furthermore, the aircraft structure stress concentration areas include holes, corners, and notches, and the strain-obvious areas include variable thickness areas and connection areas.

[0013] Step 1.3: Design the installation plan for the mechanical response data acquisition device based on the mechanical response data acquisition device selected in Step 1.1 and the sensor layout plan determined in Step 1.2. The data acquisition device should be installed close to the mechanical response data monitoring location to reduce redundancy in the acquisition line, and fixed with a fixed support to ensure that the data acquisition device remains stable during flight. When installing the sensor, locate and mark the sensor installation location, polish and clean it, and then install it with flame-retardant glue to ensure that it is not damaged and is firmly installed. Apply flame-retardant protective glue to the sensor surface. Wrap the sensor leads, acquisition lines, and power lines with flame-retardant tape at the joints for insulation and fire protection. Determine the wiring arrangement based on the mechanical response data monitoring location. After installation, debug the data acquisition device to ensure that the readings are stable and the equipment is in good working condition.

[0014] In step 1.4, develop a health status data collection plan and begin health status data collection. Use the mechanical response data acquisition equipment installed in step 1.3 to collect mechanical response data, and use the flight parameter recording system on the in-service aircraft to collect flight data. Health status data collection is divided into three phases: In the first phase of health status data collection, collect health status data m times to ensure collection stability and data accuracy. The second phase of health status data collection is the period of stable sensor operation. During this second phase, collect data t times per week to ensure that the amount of data collected by the end of the second phase meets the requirements for subsequent health status monitoring. In the third phase of health status data collection, collect data p times per month until a large number of sensor data anomalies are detected.

[0015] Furthermore, the flight data includes weight, altitude, speed, acceleration, attitude, control, engine, and environmental data; the first stage is within two weeks of starting health status data collection, and the number of collections is 3≤m≤14; the second stage is within three months of starting health status data collection, and the number of collections per week is 2≤t≤7; the third stage is within ten months of starting health status data collection, and the number of collections per month is 2≤p≤30.

[0016] The second step is to screen, extract, optimize and manage the flight data and mechanical response data to build a flight data-mechanical response data health status dataset for in-service aircraft structures.

[0017] In step 2.1, for the flight data collected in step 1.4, check and process the flight data anomalies caused by sensor failure and erroneous flight data during the data collection process, and fill in the missing values; use the correlation analysis method to reduce the correlation of the flight data, set the correlation threshold of the flight data to θ, calculate the correlation between each two types of flight data, and delete the flight data with a higher correlation with other flight data in the two types of flight data with a correlation greater than θ to reduce redundancy.

[0018] Furthermore, the correlation analysis method includes Pearson correlation analysis or Spearman correlation analysis; the correlation threshold θ ranges from 0 to 1.

[0019] In step 2.2, for the mechanical response data collected in step 1.4, the drift correction method is used to correct the drift, the noise reduction method is used to perform noise reduction processing, and the resampling method is used to align the acquisition time and unify the acquisition frequency of the processed flight data and mechanical response data to obtain flight data and mechanical response data with consistent acquisition time and acquisition frequency, and obtain the flight data-mechanical response data health status data set, which is divided into training set and test set.

[0020] Furthermore, the drift correction method includes linear regression method, polynomial fitting method or piecewise function method, the noise reduction method includes wavelet noise reduction or wavelet packet decomposition, and the resampling method includes interpolation method, averaging method, decimation method, Fourier transform or wavelet transform.

[0021] The third step is to establish a digital twin model of flight data, mechanical response data, and life monitoring to achieve high-precision fatigue remaining life prediction of in-service aircraft structures.

[0022] Step 3.1: Based on the flight data-mechanical response data health status data obtained in step 2.2, a flight data-mechanical response data digital twin model is established. The flight data and mechanical response data of the training set in the flight data-mechanical response data health status dataset are imported into the machine learning model. The flight data of the training set is used as the input of the machine learning model, and the mechanical response data of the training set is used as the output of the machine learning model. The machine learning model is used to train the mapping relationship between flight data and mechanical response data to obtain the flight data-mechanical response data digital twin model. By inputting flight data into the flight data-mechanical response data digital twin model, mechanical response data prediction is achieved. To evaluate the accuracy of the flight data-mechanical response data digital twin model, the mechanical response data of the test set in the flight data-mechanical response data health status dataset is used as the true value. The flight data of the test set is input into the flight data-mechanical response data digital twin model to obtain the predicted value of the mechanical response data of the test set. The error between the predicted value of the mechanical response data of the test set and the true value of the mechanical response data of the test set is used as the evaluation index to evaluate the accuracy of the flight data-mechanical response data digital twin model.

[0023] Furthermore, the machine learning models include Gaussian process regression, random forest, support vector machine, deep neural network, and extreme gradient boosting tree, all of which are known in the art. The error range between the predicted value of the mechanical response data of the test set and the true value of the mechanical response data of the test set is 0-1. When the error range is between [0-0.1] (i.e., 0 ≤ error range < 0.1), the accuracy of the flight data-mechanical response data digital twin model meets the requirements and subsequent analysis can be performed. When the error range is between [0.1-1] (i.e., 0.1 ≤ error range ≤ 1), the accuracy of the flight data-mechanical response data digital twin model does not meet the requirements and the machine learning model needs to be adjusted to reduce the error. Step 3.2: Establish a flight data-mechanical response data-life monitoring digital twin model. The flight data of the test set is input into the flight data-mechanical response data digital twin model established in step 3.1 to obtain the predicted value of the mechanical response data of the test set. The peak-valley pair extraction method is used to obtain the peak-valley values ​​of the mechanical response data prediction value, and the peak-valley values ​​are used as the input of the fatigue remaining life prediction method. The fatigue remaining life is calculated using the fatigue remaining life prediction method to obtain the flight data-mechanical response data-life monitoring digital twin model. By inputting the flight data into the flight data-mechanical response data-life monitoring digital twin model, the mechanical response data prediction is realized, and then the fatigue remaining life is calculated as N flights.

[0024] Furthermore, the peak-valley pair extraction method includes a rain flow counting method, a time domain extraction method, a principal component analysis method or a wavelet transform method; the fatigue remaining life prediction method includes a nominal stress method, a local stress-strain method, a continuous damage mechanics method or a detailed fatigue rating (DFR) method.

[0025] The fourth step is to develop an intelligent planning scheme for inspection and maintenance tasks, and divide the inspection and maintenance levels based on fatigue residual life and fatigue reliability coefficient.

[0026] In step 4.1, according to the damage tolerance requirements of the maintenance party for the in-service aircraft structure, the safety threshold is set to R flights. Based on the fatigue residual life N obtained in step 3.2, when N>R, the in-service aircraft structure is in a healthy and safe period; when N≤R, the in-service aircraft structure is in a healthy and dangerous period.

[0027] Step 4.2, based on the health status determined in step 4.1, mainly carry out inspection tasks for the situation in the safe period of health status, including light inspection tasks, medium inspection tasks, and heavy inspection tasks. According to the fatigue reliability coefficient of the in-service aircraft structure, the inspection reliability level is divided into levels 1 to n:

[0028] When the fatigue reliability factor of the aircraft structure in service is the maximum value FRF max , then the inspection reliability level is n, that is, the highest inspection reliability level, then when the actual number of flights of the service aircraft structure is N R =N / FRF max When conducting light inspection tasks;

[0029] When the fatigue reliability factor of the service aircraft structure is the middle value FRF, then level 1 < inspection reliability level < level n, then when the actual number of flights of the service aircraft structure N R =N / FRF, carry out medium-weight inspection tasks;

[0030] When the fatigue reliability factor of the aircraft structure in service is the minimum value FRF min , then the inspection reliability level is level 1, which is the lowest inspection reliability level. Then when the actual number of flights of the in-service aircraft structure is N R =N / FRF min At the same time, carry out heavyweight inspection tasks.

[0031] Furthermore, the lightweight inspection tasks include visual inspection and surface penetration inspection, the medium-weight inspection tasks include magnetic particle inspection and eddy current inspection, and the heavy-weight inspection tasks include ultrasonic inspection, radiographic inspection, and infrared thermal imaging inspection; the fatigue reliability coefficient is a coefficient introduced in this field for the reliability target requirements of fatigue design, and its value range is 1.0≤FRF min <FRF<FRFmax ≤4.0; the inspection reliability level n ranges from 3 to 10.

[0032] In step 4.3, based on the health status determined in step 4.1, maintenance tasks are primarily carried out for aircraft in critical health situations, with aircraft grounded for inspection and repair as soon as possible. Initially, inspection tasks are gradually performed, progressing from light to heavy, to determine the presence, location, and severity of structural damage on in-service aircraft, thereby determining the structural damage status of the in-service aircraft. Furthermore, maintenance tasks are assigned to a maintenance level based on the damage status of the in-service aircraft, including surface maintenance, localized repairs, component-level repairs, regional-level repairs, and structural-level repairs.

[0033] The beneficial effects of the present invention are:

[0034] (1) The present invention designs a health status data collection scheme for in-service aircraft structures, which can meet the airworthiness requirements of in-service aircraft and obtain health status data under real flight conditions;

[0035] (2) The present invention establishes a method for screening, extracting and optimizing flight data and mechanical response data, which can reduce the redundancy of flight data and mechanical response data and obtain a high-quality flight data-mechanical response data health status dataset;

[0036] (3) The present invention establishes a digital twin model of flight data-mechanical response data-life monitoring. By inputting flight data into the digital twin model, the mechanical response data can be predicted, and then the fatigue remaining life can be calculated, forming a digital twin monitoring process for the health status of the entire life cycle of in-service aircraft structures.

[0037] (4) The present invention has developed an intelligent planning scheme for inspection and maintenance tasks, which can divide the inspection and maintenance levels based on fatigue residual life and fatigue reliability coefficient, facilitating the full life cycle health monitoring and management of in-service aircraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flowchart for implementing a digital twin monitoring method for the lifecycle health status of in-service aircraft structures;

[0039] Figure 2 A schematic diagram of a digital twin model of flight data-mechanical response data-life monitoring;

[0040] Figure 3 A schematic diagram of an intelligent planning scheme for inspection and maintenance tasks. DETAILED DESCRIPTION

[0041] This specification will describe the embodiments of the present invention in detail with reference to the accompanying drawings. It should be noted that the specific implementation cases described below are only used to help understand the technical solution of the present invention and do not constitute a limitation on the scope of protection. In order to clearly demonstrate the technical points, the accompanying drawings only present the technical features directly related to the content of the invention and omit other non-essential details. On the basis of the technical solution of the present invention, any alternative solutions that can be obtained by any person skilled in the art through conventional technical means should be regarded as the scope of protection of the claims of the present invention.

[0042] Figure 1 The present invention provides a flow chart for the implementation of a digital twin monitoring method for the life cycle health status of an in-service aircraft structure. Figure 1 As shown, the present invention provides a digital twin monitoring method for the lifecycle health status of an in-service aircraft structure, including:

[0043] The first step is to design a health status data collection solution for in-service aircraft structures to obtain health status data under real flight conditions. The health status data includes mechanical response data and flight data. Specifically:

[0044] Step 1.1: Select airworthy mechanical response data acquisition equipment for the fuselage panel structure, including a data acquisition instrument, sensors, data acquisition cables, power cables, and fixed supports. To ensure the continuity and reliability of mechanical response data acquisition during flight, the equipment must withstand low temperatures, large temperature fluctuations, turbulence, and overload shocks. Furthermore, to avoid disrupting the aircraft's normal operations during flight, the equipment selected was lightweight, compact, energy-efficient, and resistant to electromagnetic interference. The data acquisition instrument selected was the compact and lightweight DH5916 rugged micro-dynamic data acquisition and analysis system, which can withstand low temperatures, large temperature fluctuations, turbulence, and overload shocks, and accurately measures mechanical response data such as force, pressure, displacement, velocity, and acceleration. The sensors used were BA120 series (single-chip) sensors with an operating temperature range of -80°C to +150°C and self-temperature compensation for aluminum alloy materials. The data acquisition cables and power cables were flame-retardant, and the fixed supports were constructed from composite honeycomb sandwich panels.

[0045] In step 1.2, based on the sensors selected in step 1.1, a sensor layout was designed to determine the locations for monitoring mechanical response data. Based on mechanical mechanisms, simulation results, and maintenance data, the sensor layout ensured coverage of areas of aircraft structural stress concentration and significant strain, providing accurate input for the subsequent construction of a digital twin model combining flight data and mechanical response data. Based on mechanical mechanisms, strain gauges were placed away from bolt holes and connections, such as in the center of the skin milling area, and in critical stress areas, such as the edges of holes. Based on simulation results, strain gauges were placed in areas where significant strain would occur under different operating conditions, such as the bulkhead flange, bulkhead web, and the center of the skin milling area. Based on maintenance data, strain gauges were placed in areas prone to fatigue crack initiation, such as the edges of bolt holes. A total of 32 sensors were placed on the bulkheads, connectors, and fuselage skin: 7 sensors were placed on the bulkhead flanges, 7 on the bulkhead webs, 8 on connectors, 9 in the center of the fuselage skin milling area, and 1 on the floor.

[0046] Step 1.3: Design a plan for installing the mechanical response data acquisition device based on the mechanical response data acquisition device selected in step 1.1 and the sensor layout plan determined in step 1.2. Two data collectors are installed in the luggage rack near the monitoring fuselage wall panel. Machine No. 1 is connected to sensors No. 1001-1016, and machine No. 2 is connected to sensors No. 2017-2032. The data collectors are fixed with high-strength bolts and fixed with fixed supports to ensure that the data collectors remain stable during flight. When installing the sensors, the sensor installation positions are located and marked. After polishing and cleaning, the sensors are installed with flame-retardant glue to ensure that there is no damage and the installation is firm. Flame-retardant protective glue is applied to the sensor surface. The data collection lines are tied and fixed along the outer cloth surface of the aircraft cabin. Holes are opened on the outer cloth surface for routing, and an opening is opened at the rear of the luggage rack to connect the data collection lines and power lines. The sensor leads and the data collection lines are welded with special aviation tools, and flame-retardant protective glue is applied to the sensors. The leads are fixed with tape to prevent the sensors from being damaged due to aircraft vibration, etc. After installation, the data collection equipment is debugged to ensure that the readings are stable and the equipment is in good working condition.

[0047] Step 1.4: Develop a health status data collection plan and begin health status data collection. Mechanical response data were collected using the mechanical response data acquisition equipment installed in Step 1.3, and flight data were collected using the in-service aircraft's built-in flight parameter recording system. Health status data collection was divided into three phases: In the first phase of health status data collection, within the first two weeks of data collection, health status data was collected five times to ensure stability and data accuracy. In the second phase of health status data collection, from the third week to the end of the third month, during which the sensors were operating stably, data was collected twice per week. By the end of the second phase, a total of 60 health status data were collected, meeting the data volume required for subsequent health status monitoring. In the third phase of health status data collection, after the fourth month, data was collected twice per month until a large number of sensor data anomalies were discovered by the end of the sixth month.

[0048] The second step is to screen, extract, optimize and manage the flight data and mechanical response data to build a flight data-mechanical response data health status dataset for in-service aircraft structures.

[0049] In step 2.1, for the flight data collected in step 1.4, 120 categories related to the mechanical response of the fuselage panels were initially selected from the 2,000 flight data types for analysis. Outliers caused by sensor failures and erroneous flight data during data collection were examined and processed, and missing values ​​were filled. Twenty-six categories of flight data due to invalid sensors or indirect sensor acquisition were removed, and 20 categories of flight data required preprocessing such as removing sudden changes and filling in missing values. The Spearman correlation analysis method was used to reduce the correlation of the flight data. The correlation threshold for the flight data was set at θ = 0.90. The correlation between each pair of flight data was calculated. Among the two flight data types with correlations greater than θ, the flight data with a higher correlation than the other flight data was removed. A total of 46 flight data types were removed, ultimately retaining 48 categories for subsequent analysis and training, reducing data redundancy and improving subsequent training efficiency.

[0050] In step 2.2, for the mechanical response data collected in step 1.4, the drift is corrected using the linear regression method, and the noise reduction is performed using wavelet packet decomposition. For the health status data of 60 real flights collected, the acquisition time of the processed flight data and mechanical response data is aligned and the acquisition frequency is unified to 8 Hz using the interpolation method to resample, so as to obtain flight data and mechanical response data with consistent acquisition time and acquisition frequency. The data are then divided into training sets and test sets, and 60 groups of flight data-mechanical response data health status data sets are obtained, of which 56 groups are used as training sets and 4 groups are used as test sets. Each group contains 48 columns of flight data and 32 columns of mechanical response data.

[0051] The third step is to establish a digital twin model of flight data-mechanical response data-life monitoring, such as Figure 2 As shown, high-precision fatigue residual life prediction of in-service aircraft structures is achieved. Specifically:

[0052] Step 3.1, based on the flight data-mechanical response data health status data obtained in step 2.2, establish a flight data-mechanical response data digital twin model. The flight data and mechanical response data of the training set in the flight data-mechanical response data health status data set are selected, and the extreme gradient boosting tree model is selected as the machine learning model for this example. The 48 columns of flight data from the 56 training sets are used as the input of the machine learning model, and the 32 columns of mechanical response data from the 56 training sets are used as the output of the extreme gradient boosting tree model. The extreme gradient boosting tree model is used to train the mapping relationship between flight data and mechanical response data to obtain the flight data-mechanical response data digital twin model. By inputting flight data into the flight data-mechanical response data digital twin model, the mechanical response data can be predicted. In order to evaluate the accuracy of the flight data-mechanical response data digital twin model, the three mechanical response columns 1001, 1011, and 2007 of the four test sets in the flight data-mechanical response data health status dataset were used as the true values. The flight data of the four test sets were input into the flight data-mechanical response data digital twin model. The prediction error of the 1001 mechanical response of the test set was 0.049, and the training time was 797.38 seconds; the prediction error of the 1011 mechanical response of the test set was 0.114, and the training time was 669.07 seconds; the prediction error of the 2007 mechanical response of the test set was 0.057, and the training time was 547.90 seconds.

[0053] Step 3.2: Establish a digital twin model for flight data, mechanical response data, and life monitoring. Input the 48 columns of flight data from the four test sets into the digital twin model for flight data and mechanical response data established in step 3.1, and obtain the predicted values ​​for the 32 columns of mechanical response data from the four test sets. The peak and valley values ​​of the predicted mechanical response data are obtained using the rainflow counting method. These peak and valley values ​​are used as input for the fatigue remaining life prediction method, yielding a first principal stress peak of 44.295 MPa and a valley value of -6.898 MPa. The fatigue remaining life is calculated using the detailed fatigue rating (DFR) method, resulting in a digital twin model for flight data, mechanical response data, and life monitoring. By inputting flight data into the digital twin model, mechanical response data can be predicted, and then the fatigue remaining life (N = 32,216 flights) can be calculated.

[0054] The fourth step is to develop an intelligent planning scheme for inspection and maintenance tasks, and divide the inspection and maintenance levels based on fatigue residual life and fatigue reliability coefficient, such as Figure 3 Specifically:

[0055] In step 4.1, according to the damage tolerance requirements of the maintenance party for the in-service aircraft structure, the safety threshold is set to R = 10,000 flights. Based on the fatigue residual life N = 32,216 obtained in step 3.2, N>R, and the structure is in a healthy and safe period.

[0056] Step 4.2, based on the health status determined in step 4.1, mainly carry out inspection tasks for the situation in the safe period of health status, including light inspection tasks, medium inspection tasks, and heavy inspection tasks. According to the fatigue reliability coefficient of the in-service aircraft structure, the inspection reliability level is divided into 1 to 3 levels:

[0057] Let the fatigue reliability factor of the in-service aircraft structure be FRF max =3.0, the inspection reliability level is level 3, which is the highest inspection reliability level. Then the actual number of flights of the aircraft structure in service N R =N / FRF max =10738: Carry out light inspection mission and conduct visual inspection;

[0058] If the fatigue reliability factor of the service aircraft structure is FRF=2.0, the inspection reliability level is level 2. Then when the actual number of flights of the service aircraft structure N R =N / FRF=16108, carry out medium-level inspection tasks and perform eddy current testing;

[0059] Let the fatigue reliability factor of the in-service aircraft structure be FRF min =1.5, the inspection reliability level is level 1, which is the lowest inspection reliability level. Then the actual number of flights of the aircraft structure in service N R =N / FRF min =21477, carry out heavyweight inspection task and conduct ultrasonic testing.

[0060] Finally, it is particularly noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit its scope of protection. Those skilled in the art, based on a full understanding of the above embodiments, may reasonably adjust the technical solutions or replace some of the technical features with equivalents. Such modifications or replacements shall still fall within the scope of protection defined by the claims of the present invention.

Claims

1. A digital twin monitoring method for the lifecycle health status of an in-service aircraft structure, characterized by: The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure comprises the following steps: The first step is to design a health data collection scheme for in-service aircraft structures to obtain health data under real flight conditions. The health data includes mechanical response data and flight data. The second step is to screen, extract, and optimize the flight data and mechanical response data to construct a flight data-mechanical response data health status dataset for in-service aircraft structures, and divide it into a training set and a test set. The third step is to establish a digital twin model of flight data, mechanical response data, and life monitoring to achieve high-precision prediction of the fatigue remaining life of in-service aircraft structures. Specifically: Step 3.1: Based on the flight data and mechanical response data in the training set, train the mapping relationship between flight data and mechanical response data, establish a flight data-mechanical response data digital twin model, and realize mechanical response data prediction; Step 3.2: Establish a digital twin model of flight data, mechanical response data, and life monitoring; The flight data of the test set is input into the flight data-mechanical response data digital twin model to obtain the predicted value of the mechanical response data of the test set; the peak and valley values ​​of the predicted value of the mechanical response data are used as input to obtain the flight data-mechanical response data-life monitoring digital twin model, and the fatigue remaining life is calculated as N flights; The fourth step is to develop an intelligent planning scheme for inspection and maintenance tasks, and classify inspection and maintenance levels based on fatigue residual life and fatigue reliability coefficient; specifically: Step 4.1: Based on the damage tolerance requirements of the in-service aircraft structure by the maintenance party, a safety threshold of R flights is set and compared with the fatigue residual life N obtained in step 3.2 to determine the health status of the in-service aircraft structure, including the safe health status period and the critical health status period. Step 4.2: Carry out inspection tasks for aircraft in a healthy and safe period; classify the inspection reliability level into levels 1 to n based on the fatigue reliability coefficient of the in-service aircraft structure; Step 4.3: For aircraft in critical health conditions, carry out maintenance tasks and arrange for aircraft to be parked for inspection and repair.

2. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 1 is characterized in that: The first step is specifically: Step 1.1, selecting a mechanical response data acquisition device; the mechanical response data acquisition device includes a data acquisition instrument, a sensor, a data acquisition line, a power line, and a fixed support, wherein the sensor includes a strain sensor, a displacement sensor, a temperature sensor, and an acceleration sensor; Step 1.2: Design a sensor layout plan and determine the locations for monitoring mechanical response data, including strain, stress, displacement, temperature, and acceleration. Based on mechanical mechanisms, simulation results, and maintenance data, ensure coverage of areas of concentrated stress and significant strain in the aircraft structure, providing accurate input for the subsequent construction of a digital twin model combining flight data and mechanical response data. Step 1.3: Design a plan for installing the mechanical response data acquisition device based on the mechanical response data acquisition device selected in step 1.1 and the sensor layout plan determined in step 1.

2. Step 1.4: Develop a health status data collection plan and begin health status data collection. Use the mechanical response data collection equipment installed in step 1.3 to collect mechanical response data, and use the flight parameter recording system on the in-service aircraft to collect flight data. The collection of health status data is divided into three stages: in the first stage of health status data collection, health status data is collected m times, and the collection stability and data accuracy are checked; the second stage of health status data collection is the stable working period of the sensor. In this second stage, data is collected t times a week to ensure that the amount of data collected before the end of the second stage meets the needs of subsequent health status monitoring; in the third stage of health status data collection, data is collected p times a month until a large number of sensor data anomalies are found.

3. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 2 is characterized in that: In the first step: In step 1.2, the aircraft structure stress concentration areas include holes, corners, and notches, and the strain-significant areas include variable thickness areas and connection areas; In step 1.4, the flight data includes weight, altitude, speed, acceleration, attitude, control, engine, and environmental data; the first stage is within two weeks of starting health status data collection, and the number of collections is 3≤m≤14; the second stage is within three months of starting health status data collection, and the number of collections per week is 2≤t≤7; the third stage is within ten months of starting health status data collection, and the number of collections per month is 2≤p≤30.

4. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 2 is characterized in that: The second step is specifically as follows: Step 2.1: For the flight data collected in step 1.4, check and process outliers caused by sensor failures and erroneous flight data during data collection, and fill in missing values. Use correlation analysis to filter the flight data for correlation reduction. Set the correlation threshold for the flight data to θ, calculate the correlation between each pair of flight data, and delete any flight data with a correlation greater than θ that has a higher correlation with the other flight data. In step 2.2, the mechanical response data collected in step 1.4 is drift corrected and noise reduced, and the acquisition time and acquisition frequency of the processed flight data and mechanical response data are aligned using the resampling method to obtain flight data and mechanical response data with consistent acquisition time and frequency, and a flight data-mechanical response data health status dataset is obtained, which is divided into a training set and a test set.

5. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 4 is characterized in that: In the second step: In step 2.1, the correlation analysis method adopts Pearson correlation analysis or Spearman correlation analysis; the correlation threshold θ ranges from 0 to 1; In the step 2.2, a drift correction method is used to correct the drift, and the drift correction method includes a linear regression method, a polynomial fitting method or a piecewise function method; a noise reduction method is used to perform noise reduction processing, and the noise reduction method includes wavelet noise reduction or wavelet packet decomposition; the resampling method includes an interpolation method, an averaging method, a decimation method, a Fourier transform or a wavelet transform method.

6. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 4 is characterized in that: The third step is specifically as follows: The specific process of realizing the mechanical response data prediction in the said step 3.1 is as follows: based on the flight data-mechanical response data health status data obtained in step 2.2, a flight data-mechanical response data digital twin model is established; the flight data and mechanical response data of the training set are imported into the machine learning model, the flight data in the training set is used as the input of the machine learning model, and the mechanical response data in the training set is used as the output of the machine learning model; the machine learning model is used to train the mapping relationship between the flight data and the mechanical response data, and the flight data-mechanical response data digital twin model is obtained; and the mechanical response data prediction is realized by inputting the flight data into the flight data-mechanical response data digital twin model; In step 3.2, the specific process of establishing the flight data-mechanical response data-life monitoring digital twin model is as follows: The peak-valley pair extraction method is used to obtain the peak-valley values ​​of the mechanical response data prediction value, and the peak-valley values ​​are used as the input of the fatigue remaining life prediction method. The fatigue remaining life is calculated using the fatigue remaining life prediction method to obtain the flight data-mechanical response data-life monitoring digital twin model. By inputting the flight data into the flight data-mechanical response data-life monitoring digital twin model, the mechanical response data prediction is realized, and then the fatigue remaining life is calculated as N flights.

7. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 6 is characterized in that: The third step is specifically as follows: In step 3.1, in order to evaluate the accuracy of the flight data-mechanical response data digital twin model, the mechanical response data in the test set is used as the true value, the flight data of the test set is input into the flight data-mechanical response data digital twin model, and the predicted value of the mechanical response data of the test set is obtained. The error between the predicted value of the mechanical response data of the test set and the true value of the mechanical response data of the test set is used as an evaluation index to evaluate the accuracy of the flight data-mechanical response data digital twin model; In step 3.1, the machine learning model includes Gaussian process regression, random forest, support vector machine, deep neural network or extreme gradient boosting tree; In step 3.2, the peak-valley pair extraction method includes a rain flow counting method, a time domain extraction method, a principal component analysis method or a wavelet transform method; the fatigue remaining life prediction method includes a nominal stress method, a local stress-strain method, a continuous damage mechanics method or a detailed fatigue rating method.

8. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 6 is characterized in that: In step 3.1, the error range between the predicted value of the mechanical response data of the test set and the true value of the mechanical response data of the test set is 0~1. When 0≤error range<0.1, the accuracy of the flight data-mechanical response data digital twin model meets the requirements. When 0.1≤error range≤1, the accuracy of the flight data-mechanical response data digital twin model does not meet the requirements, and the machine learning model needs to be adjusted to reduce the error.

9. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 6 is characterized in that: The fourth step is specifically as follows: The health status of the in-service aircraft structure in step 4.1 is determined as follows: when N>R, the in-service aircraft structure is in a safe health period; when N≤R, the in-service aircraft structure is in a dangerous health period; In step 4.2, based on the health status determined in step 4.1, inspection tasks are carried out for those in the safe health period, including lightweight inspection tasks, medium-weight inspection tasks, and heavyweight inspection tasks. The inspection reliability levels are divided into levels 1 to n, specifically: When the fatigue reliability factor of the aircraft structure in service is the maximum value FRF max , then the inspection reliability level is n, that is, the highest inspection reliability level, then when the actual number of flights of the service aircraft structure is N R =N / FRF max When conducting light inspection tasks; When the fatigue reliability factor of the service aircraft structure is the middle value FRF, then level 1 < inspection reliability level < level n, then when the actual number of flights of the service aircraft structure N R =N / FRF, carry out medium-weight inspection tasks; When the fatigue reliability factor of the aircraft structure in service is the minimum value FRF min , then the inspection reliability level is level 1, which is the lowest inspection reliability level. Then when the actual number of flights of the in-service aircraft structure is N R =N / FRF min At the same time, carry out heavyweight inspection tasks.

10. The digital twin monitoring method for the lifecycle health status of an in-service aircraft structure according to claim 9 is characterized in that: In the fourth step: In step 4.2, the lightweight inspection tasks include visual inspection and surface penetration inspection, the medium-weight inspection tasks include magnetic particle inspection and eddy current inspection, and the heavy-weight inspection tasks include ultrasonic inspection, radiographic inspection, and infrared thermal imaging inspection; the fatigue reliability coefficient has a value range of 1.0 ≤ FRF min <FRF<FRF max ≤4.0 The inspection reliability level n ranges from 3 to 10; The specific steps of step 4.3 are as follows: first, carry out inspection tasks from light weight to heavy weight step by step to confirm whether the structure of the service aircraft is damaged, the location of the damage, and the damage level, and obtain the damage status of the structure of the service aircraft; then, divide the maintenance level according to the damage status of the structure of the service aircraft, and carry out maintenance tasks, including surface maintenance, local repair, component-level repair, regional-level repair, and structural-level repair.

Images