Flight load tracking methods, equipment, media, and products that integrate flight test data with load inversion

Abstract

The present application discloses a flight load tracking method, equipment, medium and product that integrates test flight data and load inversion, and relates to the field of flight load tracking. The method includes obtaining flight parameters and measured strain data of an aircraft test flight during the test flight phase; constructing a flight parameter-strain prediction model; training the flight parameter-strain prediction model based on the flight parameters and measured strain data of the aircraft test flight; obtaining the actual flight parameters of the aircraft in the service phase in real time; inputting the actual flight parameters of the aircraft in the service phase into the trained flight parameter-strain prediction model to obtain the strain prediction value of the corresponding sensor position; and using the load inversion algorithm of the reduced-order model based on the strain prediction value of the corresponding sensor position to obtain the load distribution of the entire field and the target area. The present application can track flight loads more accurately and achieve better prediction accuracy, and can be extended to various types of aircraft structures.

Description

Technical Field

[0001] The present application relates to the field of flight load tracking, and in particular to a flight load tracking method, equipment, medium and product that integrates test flight data and load inversion. Background Art

[0002] At present, all countries in the world are facing the problem of aging aircraft fleets. The main reason why aging poses a major threat to aircraft flight safety is structural fatigue, which affects the combat capability of military aircraft and the operational safety of civil aircraft. With the development of engineering practice, the concept of aircraft structural safety assurance is also evolving in stages. Since the loads of aircraft are complex and changeable during service, and there is a difference between the load history and the design load spectrum in the design stage, in order to accurately perceive the damage and health status of a single aircraft structure and accurately predict the evolution of the future damage state of the aircraft, the main aviation structure fatigue life tracking method based on the digital twin of the aircraft structure has been carried out at home and abroad.

[0003] Aircraft fatigue life tracking based on digital twins is a systematic work, including load and usage monitoring, order reduction and proxy efficient modeling, data fusion and multi-fidelity modeling, as well as structural damage diagnosis and remaining life prediction. During the flight, the aircraft is continuously subjected to aerodynamic loads, which causes structural deformation and may cause structural damage. In order to estimate the consumption of aircraft fatigue life, it is necessary to obtain the actual load spectrum and structural deformation distribution of the aircraft in operation, which is a key step in realizing structural fatigue life tracking based on digital twins.

[0004] At present, there are three methods for obtaining load data from aircraft: simulation prediction method based on flight parameters, load inversion based on strain measurement, and data-driven flight parameter-load prediction. The simulation prediction method based on flight parameters uses flight parameter record data and a series of simulation methods such as flight mechanics, computational fluid dynamics, and structural finite elements to infer the loads and stresses of key parts of the aircraft. This method can reproduce the real loading sequence experienced by the structure and provide more detailed load distribution information, improve the insight of the structural digital twin, and reduce operating costs. However, the accuracy of this method that relies entirely on simulation prediction depends on the fidelity of the model, and the credibility of the results is difficult to guarantee.

[0005] The load inversion method based on strain measurement is to directly measure the stress and strain information of the structure by arranging sensors on the structure to obtain the time-varying information of the load at the key parts. The load inversion algorithm is used to obtain the load distribution of the entire structure or the area of ​​interest. The load inversion method based on strain measurement can obtain the measured strain in real time, and obtain the load data of the entire structure or the area of ​​interest with high time resolution according to the inversion algorithm. However, since the aircraft of various countries rarely install sensors on the fuselage during the service stage, the load inversion method based on strain measurement is limited.

[0006] The load acquisition method of data-driven flight parameter-load prediction mainly records the time history of flight parameters, and predicts the loads affecting the key parts of the structure through flight parameters based on the correlation between flight parameters and structural loads. In the current practice of single-aircraft load tracking, strain sensors are generally installed on a few test aircraft during the test flight stage. Flight parameters and strain data are collected simultaneously through test flight subjects to establish a proxy model of flight parameters and strain data. During the service stage, the flight data collected in real time from each aircraft can be used to predict the strain of key parts. This method has been widely used at present. Its advantages are high reliability of flight parameter recording system, long service life, and low cost of sensor installation and modification. However, there are two problems with the existing methods: first, the existing methods use multivariate linear regression or artificial neural network to establish flight parameter-load prediction models. These methods cannot consider the time series characteristics of flight parameters, strain, and data, so the accuracy of strain prediction is difficult to guarantee; second, the existing methods focus on the strain values ​​of a few key parts, but it is difficult to track the deformation distribution of specific areas or the entire aircraft. When it is difficult to install strain sensors on key parts, it is also impossible to accurately obtain strain.

[0007] Therefore, how to track flight loads more accurately and achieve better prediction accuracy has become an urgent problem to be solved in this field. Summary of the invention

[0008] The purpose of this application is to provide a flight load tracking method, equipment, medium and product that integrates test flight data and load inversion, which can track flight loads more accurately and achieve better prediction accuracy, and can be extended to various types of aircraft structures.

[0009] To achieve the above objectives, this application provides the following solutions:

[0010] In a first aspect, the present application provides a method for tracking flight loads by integrating test flight data and load inversion, comprising:

[0011] Obtain flight parameters and measured strain data of aircraft test flights during the flight test phase;

[0012] Constructing a flight parameter-strain prediction model;

[0013] Training the flight parameter-strain prediction model based on the flight parameters of the aircraft test flight and the measured strain data;

[0014] Obtain the actual flight parameters of the aircraft in service in real time;

[0015] Inputting the actual flight parameters of the aircraft in the service phase into the trained flight parameter-strain prediction model to obtain the strain prediction value corresponding to the sensor position;

[0016] Based on the strain prediction value of the corresponding sensor position, a load inversion algorithm of a reduced-order model is used to obtain the load distribution of the entire field and the target area.

[0017] Optionally, between the step of acquiring the flight parameters and measured strain data of the aircraft test flight in the test flight phase and the step of building the flight parameter-strain prediction model, the step further includes: preprocessing the flight parameters and measured strain data of the aircraft test flight in the test flight phase; the preprocessing specifically includes the following steps:

[0018] Scaling the flight parameters and measured strain data of the aircraft test flight to a preset range;

[0019] Dividing the flight parameters of the scaled aircraft test flight and the measured strain data into sequences of equal length according to a preset step length;

[0020] The data after step size division is divided into training set and validation set.

[0021] Optionally, the strain data is strain data after temperature compensation is removed.

[0022] Optionally, the flight parameter-strain prediction model specifically includes:

[0023] The input layer, the first convolution module, the second convolution module, the first LSTM module, the second LSTM module, the first fully connected layer, the second fully connected layer, the third fully connected layer and the output layer are connected in sequence.

[0024] Optionally, obtaining the load distribution of the entire field and the target area by using a load inversion algorithm of a reduced-order model based on the strain prediction value of the corresponding sensor position specifically includes the following steps:

[0025] Based on the aerodynamic and structural simulation model of the aircraft, a corresponding simulation database is constructed; the simulation database includes: aerodynamic load distribution under different flight parameters , deformation field distribution and the virtual sensor value of the strain sensor ;

[0026] The aerodynamic load distribution Perform principal component analysis to obtain a low-dimensional representation of the corresponding aerodynamic load distribution ;

[0027] The deformation field distribution Perform principal component analysis to obtain a low-dimensional representation of the deformation field distribution ;

[0028] Establishing a load inversion proxy model ;

[0029] Based on the low-dimensional representation of the aerodynamic load distribution Virtual sensing value with fiber Bragg grating strain sensor The load inversion proxy model Conduct training;

[0030] Building a deformation prediction proxy model ;

[0031] Based on the low-dimensional representation of the aerodynamic load distribution With the low-dimensional representation of the deformation field distribution Deformation prediction agent model Conduct training;

[0032] The fiber Bragg grating strain sensing value obtained during the flight Input into the trained load inversion proxy model to obtain the predicted value of the low-dimensional representation of the load distribution ;

[0033] The predicted value of the low-dimensional representation of the load distribution Input into the trained deformation prediction agent model to obtain the deformation field distribution prediction value of the entire structure and the target position or the low-dimensional representation prediction value of the deformation field distribution;

[0034] When the predicted value is a low-dimensional representation of the deformation field distribution, the actual distribution field is restored through PCA inverse mapping.

[0035] Optionally, the expression of the load inversion proxy model is as follows:

[0036]

[0037] in,

[0038] , represents the strain sensing value, represents the load inversion proxy model, A low-dimensional representation of the aerodynamic load distribution, represents the deviation vector caused by the PCA mean, is the number of strain sensors, is the dimension of the reduced aerodynamic load.

[0039] Optionally, the deformation prediction agent model is expressed as follows:

[0040]

[0041] in, represents the deformation prediction agent model, is the dimension reduction load Deformation under full load The proportionality coefficient between is the bias term.

[0042] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the flight load tracking method of integrating test flight data and load inversion as described above.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the flight load tracking method of integrating test flight data and load inversion as described above.

[0044] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the flight load tracking method of integrating test flight data and load inversion as described above.

[0045] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0046] The present application provides a flight load tracking method, device, medium and product that integrates test flight data and load inversion. According to the correlation between external load and structural response, a flight parameter-strain prediction model is established based on flight parameter and strain measurement data during the test flight phase, and an inverse-forward load inversion model is trained based on a simulation data set. During the service phase, the strain is predicted by real-time flight parameters and the complete deformation field distribution is obtained from the load inversion model. The present application uses a deep learning method to train a flight parameter-strain prediction model that takes into account time series dependencies, combined with an inverse-forward load inversion method based on a reduced-order proxy model, to propose a flight load tracking method that integrates test flight data and load inversion. A load tracking framework is proposed and deployed in the flight load tracking of the structural digital twin for the service stage. This application is based on the flight parameter-strain data training during the aircraft test flight stage. During the service stage, the flight parameters can be used to predict the full-field load, saving the cost of installing strain sensors without affecting the flight performance of the aircraft. Because the acceleration data is considered in the flight parameter data, and there is a mapping relationship between the flight parameters and the real strain, it can achieve better results than the aerodynamic / structural reduced-order prediction based on flight parameters, balance the model fidelity and computational cost, can more accurately track the flight load and achieve better prediction accuracy, and can be extended to various types of aircraft structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A schematic diagram of a flow chart of a flight load tracking method that integrates test flight data and load inversion in an embodiment of the present application;

[0049] Figure 2 A flowchart of a flight load tracking process that integrates test flight data and load inversion is provided in one embodiment of the present application;

[0050] Figure 3 A schematic diagram of the basic structure and component modules of a flight parameter-strain prediction neural network provided in an embodiment of the present application;

[0051] Figure 4 A schematic diagram of a process for constructing a load inversion model based on strain measurement provided in an embodiment of the present application;

[0052] Figure 5 A schematic diagram comparing flight parameter-strain prediction results of different network structures provided in one embodiment of the present application;

[0053] Figure 6 A schematic diagram showing a comparison of load inversion results based on flight parameter-strain prediction values ​​provided in an embodiment of the present application;

[0054] Figure 7 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0056] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0057] In an exemplary embodiment, Figure 1 and Figure 2 As shown, a flight load tracking method integrating test flight data and load inversion is provided, which is executed by a computer device, specifically, it can be executed by a computer device such as a terminal or a server alone, or it can be executed by a terminal and a server together, including the following steps 101 to 106. Among them:

[0058] Step 101: Acquire flight parameters and measured strain data of an aircraft test flight during a test flight phase.

[0059] Step 102: Construct a flight parameter-strain prediction model.

[0060] Specifically, using the flight parameters and measured strain data of the aircraft test flight, based on the LSTM network and the 1DCNN network, a basic structure of a flight parameter-strain prediction model is designed, which includes an input layer, a first convolution module, a second convolution module, a first LSTM module, a second LSTM module, a first fully connected layer, a second fully connected layer, a third fully connected layer and an output layer connected in sequence. The basic structure and components of the model are as follows: Figure 3 shown.

[0061] It should be noted that the strain data used in this application is the data after removing the temperature compensation strain. Therefore, this method can directly output the prediction results after removing the thermal strain, thereby avoiding the influence of temperature changes on load prediction.

[0062] Step 103: training the flight parameter-strain prediction model based on the flight parameters of the aircraft test flight and the measured strain data.

[0063] For the training of the flight parameter-strain prediction model, a convolutional neural network is proposed because convolutional neural networks are usually more suitable for capturing high-frequency features. Since CNN shares weights in the spatial dimension, it can effectively capture local patterns and details. For the convolutional neural network, the convolution kernel size is set to 5, the padding length is set to 2, and the stride is set to 1. This ensures that the size before and after the convolution remains unchanged, that is, the output of the convolution module is the same as the input length, which is convenient for the superposition of subsequent convolution modules. It should be noted that when the number of convolutional modules and LSTM modules is 0, the constructed fully connected neural network will decompose the sequence data into discrete sample data for processing, which is consistent with conventional artificial neural networks. The number of hidden neurons in each layer of all networks is 64.

[0064] Step 104: Acquire the actual flight parameters of the aircraft in the service phase in real time.

[0065] Step 105: inputting the actual flight parameters of the aircraft in the service phase into the trained flight parameter-strain prediction model to obtain the strain prediction value corresponding to the sensor position.

[0066] Step 106: Based on the strain prediction value of the corresponding sensor position, a load inversion algorithm of a reduced-order model is used to obtain the load distribution of the entire field and the target area.

[0067] like Figure 4 As shown, Figure 4 A schematic diagram of the process for constructing a load inversion model based on strain measurements, which includes the following steps:

[0068] Step 1061: Based on the aerodynamic and structural simulation model of the aircraft, a corresponding simulation database is constructed; the simulation database includes: aerodynamic load distribution under different flight parameters , deformation field distribution and the virtual sensor value of the strain sensor .

[0069] Step 1062: Distribute the aerodynamic load Perform principal component analysis to obtain a low-dimensional representation of the corresponding aerodynamic load distribution .

[0070] If the deformation field distribution has more degrees of freedom, the low-dimensional representation of the deformation field distribution can also be obtained based on principal component analysis. .

[0071] Step 1063: Distribute the deformation field Perform principal component analysis to obtain a low-dimensional representation of the deformation field distribution .

[0072] Step 1064: Establish load inversion proxy model .

[0073] Specifically, the expression of the load inversion proxy model is as follows:

[0074] (1)

[0075] in, It is called the calibration matrix, which is a linear operator that maps the reduced-dimensional load representation to the strain sensor value, and represents the proportionality coefficient between the strain sensor value and the reduced-dimensional load: , represents the strain sensing value, represents the load inversion proxy model, A low-dimensional representation of the aerodynamic load distribution, represents the deviation vector caused by the PCA mean, is the number of strain sensors, is the dimension of the reduced aerodynamic load. and The low-dimensional representation of the aerodynamic load distribution in the simulation database can be used , Virtual Strain Sensing Obtained using linear regression fitting.

[0076] Step 1065: Based on the low-dimensional representation of the aerodynamic load distribution The load inversion proxy model is trained.

[0077] Step 1066: Convert the fiber Bragg grating (FBG) strain virtual sensor value The input is fed into the trained load inversion proxy model to obtain the predicted value of the low-dimensional representation of the load distribution.

[0078] Step 1067: Establish deformation prediction agent model .

[0079] The specific formula is as follows:

[0080] (2)

[0081] in, is the dimension reduction load Deformation under full load The proportionality coefficient between is the bias term. The low-dimensional representation of the aerodynamic load distribution in the simulation database can be , deformation field distribution Obtained using linear regression fitting.

[0082] Step 1068: Low-dimensional representation based on the deformation field distribution The deformation prediction agent model is trained.

[0083] Step 1069: Low-dimensional representation of the load distribution The data is input into the trained deformation prediction agent model to obtain the deformation field distribution prediction value of the entire structure and the target position or the low-dimensional representation prediction value of the deformation field distribution.

[0084] Step 10610: When the predicted value is , and the actual distribution field is restored through PCA inverse mapping.

[0085] During online use, after obtaining the measured strain sensing value, the inverted reduced-dimensional load and temperature values ​​can be obtained using formula (4).

[0086] The measurement error caused by thermal strain is taken into account in the load inversion model, and the formula is:

[0087] (3)

[0088] in, is the augmented vector of the inverse load, is the augmented matrix of the calibration matrix.

[0089] When measured by strain sensors Inversion When , the inverse value of the aerodynamic load distribution can be obtained by formula (4) and the inversion value of temperature change .

[0090] (4)

[0091] After the reduced-dimensional aerodynamic loads are obtained through the above inversion process, the full-field load deformation of the aircraft can be obtained through the forward prediction process. :

[0092] (5)

[0093] In addition, since the amplitudes of flight parameters and strain measurements in different dimensions are different, directly using them as network inputs may cause mismatches in the calculation of loss functions in each dimension, affecting the prediction performance of the model. The present application further includes between step 101 and step 102: preprocessing the flight parameters and measured strain data of the aircraft test flight in the test flight phase, specifically including the following steps:

[0094] The flight parameters of the aircraft test flight and the measured strain data are scaled to a preset range; the preset range in this embodiment is .

[0095] Dividing the flight parameters of the scaled aircraft test flight and the measured strain data into sequences of equal length according to a preset step length;

[0096] The data after step division is divided into training set and validation set, specifically in a ratio of 8:2.

[0097] The above-mentioned aircraft in this embodiment is a drone.

[0098] Below, the technical solution of this application is further described in detail with a specific example:

[0099] In this embodiment, the method flow in this application is used to perform application testing on a certain type of drone instance to verify the prediction accuracy of this application.

[0100] The 13 flight parameters and their ranges used by the example UAV are shown in Table 1:

[0101] Table 1 Flight parameters used in the load prediction model

[0102]

[0103] The flight parameters and strain data measured in the flight test are used to train the deep learning network introduced. 800 seconds of flight test data are used, the sampling rate of the test data is 50Hz, and the total number of samples is 40,000. The first 20,000 samples are used for training, and the remaining 20,000 samples are used for testing. The sequence length is set to 800 and the step size is 20, and 961 time series can be obtained. The flight parameters and measured strain data of the UAV test flight are scaled to a suitable range and divided into a training set and a validation set at a ratio of 80% and 20%. The information of the preprocessed data is shown in Table 2.

[0104] Table 2 Preprocessed data for flight parameter-strain prediction network

[0105]

[0106] Table 3 shows the performance of deep learning models with typical network structures. The training loss and validation loss are the loss function values ​​after normalization of the training set and validation set, and the test set strain error is the RMSE error obtained from the original strain data. It can be seen that for all network structures, the validation loss is about twice the training loss.

[0107] Table 3 Parameters and model prediction performance of different network structures

[0108]

[0109] Three neural network structures with roughly the same number of parameters, namely, convolutional neural network (9CNN+2FNN), LSTM neural network (6LSTM+2FNN), and CNN-LSTM combined neural network (4CNN-4LSTM+2FNN), were selected for comparison, and 2FNN was used as the benchmark. Figure 5 As shown. Figure 5 The convolutional neural network in part a of Figure 1 recovers the high-frequency changes of strain very well, and the predicted results are very close to the measured values, with only oscillations in the predicted results in some areas. Figure 5 The result of the LSTM neural network in part b is relatively smooth, and the prediction in some areas is biased. This shows that the error of the LSTM neural network mainly comes from the second item. Figure 5 The result of the combined neural network in part c is similar to Figure 5 The fully connected neural network in the d part of the figure is better than the LSTM network result. This shows that adding a CNN module in front can improve the extraction of high-frequency features, but the effect is still not as good as the pure CNN network, and the result obtained by 2FNN is also relatively smooth.

[0110] Based on the comparison of the flight parameter-strain prediction results of the above different network structures, the 9CNN+2FNN network is used as the final flight parameter-strain prediction model, and then the measured flight parameters are input into the flight parameter-strain prediction model based on deep learning during the service phase to obtain the predicted strain value. After that, the FBG strain sensor is used to carry out strain measurement, and the reduced-dimensional aerodynamic load is obtained using the principal component analysis method. The load inversion model based on strain measurement can be obtained using equations 3 and 4 in step 106. Finally, the predicted strain value is input into the load inversion model to obtain the full-field load distribution. Figure 6 Parts a-f in the figure show the flight load tracking results obtained by combining the predicted strain based on the convolutional neural network with the load inversion algorithm. The load tracking results of the second half of the flight test are shown here. It can be seen that the method of integrating test flight data and load inversion used in this application has achieved similar results as the load inversion method based on strain measurement, which confirms the effectiveness of the proposed method.

[0111] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store flight load tracking data that integrates test flight data and load inversion. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a flight load tracking method that integrates test flight data and load inversion is implemented.

[0112] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0113] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0114] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0115] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0116] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0117] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0118] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0119] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A flight load tracking method integrating flight test data and load inversion, characterized in that: The flight load tracking method integrating the test flight data and the load inversion comprises: Obtain flight parameters and measured strain data of aircraft test flights during the flight test phase; Constructing a flight parameter-strain prediction model; Training the flight parameter-strain prediction model based on the flight parameters of the aircraft test flight and the measured strain data; Obtain the actual flight parameters of the aircraft in service in real time; Inputting the actual flight parameters of the aircraft in the service phase into the trained flight parameter-strain prediction model to obtain the strain prediction value corresponding to the sensor position; Based on the strain prediction value of the corresponding sensor position, a load inversion algorithm of a reduced-order model is used to obtain the load distribution of the entire field and the target area; The strain data is the strain data after temperature compensation is removed; The load distribution of the entire field and the target area is obtained by using a load inversion algorithm of a reduced-order model based on the strain prediction value of the corresponding sensor position, which specifically includes the following steps: Based on the aerodynamic and structural simulation model of the aircraft, a corresponding simulation database is constructed; the simulation database includes: aerodynamic load distribution under different flight parameters , deformation field distribution And the virtual sensing value of the strain sensor ; The aerodynamic load distribution Perform principal component analysis to obtain a low-dimensional representation of the corresponding aerodynamic load distribution ; The deformation field distribution Perform principal component analysis to obtain a low-dimensional representation of the deformation field distribution ; Establishing a load inversion proxy model ; Based on the low-dimensional representation of the aerodynamic load distribution Virtual sensing value with fiber Bragg grating strain sensor The load inversion proxy model Conduct training; Building a deformation prediction proxy model ; Based on the low-dimensional representation of the aerodynamic load distribution With the low-dimensional representation of the deformation field distribution Deformation prediction agent model Conduct training; The fiber Bragg grating strain sensing value obtained during the flight Input into the trained load inversion proxy model to obtain the predicted value of the low-dimensional representation of the load distribution ; The predicted value of the low-dimensional representation of the load distribution Input into the trained deformation prediction agent model to obtain the deformation field distribution prediction value of the entire structure and the target position or the low-dimensional representation prediction value of the deformation field distribution; When the predicted value is a low-dimensional representation of the deformation field distribution, the actual distribution field is restored through PCA inverse mapping.

2. The flight load tracking method integrating flight test data and load inversion according to claim 1, characterized in that: The step of obtaining the flight parameters and measured strain data of the aircraft test flight in the test flight phase and constructing the flight parameter-strain prediction model also includes: preprocessing the flight parameters and measured strain data of the aircraft test flight in the test flight phase; the preprocessing specifically includes the following steps: Scaling the flight parameters and measured strain data of the aircraft test flight to a preset range; Dividing the flight parameters of the scaled aircraft test flight and the measured strain data into sequences of equal length according to a preset step length; The data after step size division is divided into training set and validation set.

3. The flight load tracking method integrating flight test data and load inversion according to claim 1, characterized in that: The flight parameter-strain prediction model specifically includes: The input layer, the first convolution module, the second convolution module, the first LSTM module, the second LSTM module, the first fully connected layer, the second fully connected layer, the third fully connected layer and the output layer are connected in sequence.

4. The flight load tracking method integrating flight test data and load inversion according to claim 1, characterized in that: The expression of the load inversion proxy model is as follows: ; in, , , represents the strain sensing value, represents the load inversion proxy model, A low-dimensional representation of the aerodynamic load distribution, represents the deviation vector caused by the PCA mean, is the number of strain sensors, is the dimension of the reduced aerodynamic loads.

5. The flight load tracking method integrating flight test data and load inversion according to claim 1, characterized in that: The expression of the deformation prediction agent model is as follows: ； in, represents the deformation prediction agent model, is the dimension reduction load Deformation under full load The proportionality coefficient between is the bias term.

6. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the flight load tracking method for fusing test flight data and load inversion as described in any one of claims 1-5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the flight load tracking method for fusing test flight data and load inversion described in any one of claims 1 to 5 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the flight load tracking method for fusing test flight data and load inversion described in any one of claims 1 to 5 are implemented.

Images