Aero-engine state space model matrix online identification method and device based on digital twinning and performance parameter sensing method

Abstract

The invention discloses an aero-engine state space model matrix online identification method based on digital twinning, and belongs to the technical field of aero-engine control. The invention discloses an aero-engine state space model matrix online identification method. The method comprises the following steps: S1, constructing past and future line full-rank Hankel matrixes according to a finite input-output data set of an auxiliary airborne model in a current finite time period; s2, constructing an optimal prediction output estimator; and S3, obtaining a future f-step prediction output equation according to the optimal prediction output estimator, and then obtaining an aero-engine state space model matrix by solving a simultaneous equation of estimated output values and actual output values of future k + 1 and k + 2 steps. The invention further discloses an aero-engine state space model matrix online identification device based on digital twinning and an aero-engine performance parameter sensing method. According to the invention, online real-time high-precision sensing of performance parameters of the aero-engine can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of aircraft engine control, and in particular relates to an aircraft engine state space model matrix online identification method, device and performance parameter perception method. Background Art

[0002] The airborne adaptive model is an important technical means for sensing the performance parameters of modern aircraft engines. It is indispensable in the intelligent control and health management system of aircraft engines. In order to achieve the performance parameter estimation of the entire life cycle of aircraft engines, the current airborne adaptive model mainly consists of an airborne model + a Kalman filter: the Kalman filter first estimates the health parameter degradation amount based on the deviation between the airborne model and the actual engine measurement, and then uses this health parameter degradation amount to update the airborne model, so as to adapt to the performance degradation of the engine components throughout the life cycle and achieve accurate estimation of operating parameters and key performance parameters.

[0003] At present, from the perspective of real-time performance, designing an airborne adaptive model based on a linear Kalman filter is a preferred choice. However, the linear Kalman filter needs to know the linear state space model matrix of the current steady-state operating point of the engine at all times. The existing method mainly uses small perturbation method, fitting method and other methods to solve the state space model matrix of the standard engine operating point offline, and then calls it online. This method has low envelope coverage and poor accuracy. A Chinese invention patent (Sheng Hanlin et al., A model-based aviation engine performance recovery control method and device, CN202110958916.8) discloses an airborne model online linearization method based on frequency analysis, which uses low-frequency dynamic state and high-frequency dynamic state to represent the nonlinear system, and then performs Taylor series expansion on the high-frequency state dynamics equation in the form of deviation to obtain a linearized state space model. The state matrix can be updated in real time online, but the entire calculation process has many high-dimensional inverse matrix calculations, high complexity, and needs to be recalculated every time it flies, and software / hardware computing resources are seriously wasted. Therefore, there are still many technical deficiencies in the current engine performance parameter estimation method based on the airborne adaptive model. Summary of the invention

[0004] The purpose of the present invention is to overcome the technical problem of insufficient performance perception accuracy of the existing airborne adaptive model based on linear Kalman filter, and to provide an online identification method for the state space model matrix of an aircraft engine based on digital twin, which can perform online high-precision identification of the state space model matrix of the aircraft engine, and further realize online real-time high-precision perception of the performance parameters of the aircraft engine.

[0005] The technical solution proposed by the present invention is as follows:

[0006] In the online identification method of the state space model matrix of an aircraft engine based on digital twin, the control input u and the health parameter excitation signal h are input online and in real time into an auxiliary airborne model that is the same as the main airborne model to obtain the corresponding output data of the auxiliary airborne model; and according to the input-output data of the auxiliary airborne model, the online identification of the state space model matrix of the aircraft engine is performed by the following method:

[0007] S1. Construct past and future row full-rank Hankel matrices based on the finite input-output data set of the auxiliary airborne model in the current finite time:

[0008]

[0009]

[0010] Among them, X f is the Hankel matrix of future state variables; U p and U f are the Hankel matrices of past and future control variables; Y p and Y f are the Hankel matrices of past and future output variables respectively; p represents the past time domain; f represents the future time domain; n x Represents the dimension of the state variable x, n u represents the total dimension of the control variables u and health parameters h, n y Indicates the dimension of the output variable y;

[0011] S2. Construct an optimal prediction output estimator of the following form:

[0012]

[0013] in, represents the future estimated output of the optimal predicted output estimator; the coefficient matrix and By solving for the future estimated output and the future actual output Y f The least squares problem of : get; is the Frobenius 2 norm;

[0014] S3, according to the optimal prediction output estimator, the prediction output equation for the next f steps is obtained:

[0015]

[0016] in, m is the control time domain;

[0017] Then, by solving the following estimated output values ​​for the next k+1 and k+2 steps: The actual output value y k+1 ,y k+2 The simultaneous equations of the state matrix A, input matrix B, output matrix C, and transfer matrix D of the aircraft engine state space model are obtained:

[0018]

[0019] Preferably, the least squares problem is solved using the following method:

[0020] Step 1: Convert the solution to the least squares problem into matrix Y f The row space of the matrix [W p U f ] T The orthogonal projection is:

[0021]

[0022] Where, “ / ” indicates orthographic projection;

[0023] Step 2: Use QR decomposition to decompose the following combined Hankel matrix:

[0024]

[0025] in, is an upper triangular matrix, is an orthogonal matrix;

[0026] Step 3: Get the coefficient matrix L by the following formula: w and L u Values:

[0027]

[0028] The superscript represents the generalized inverse.

[0029] Further preferably, the matrix R is updated in real time using the input-output data updated in real time by the auxiliary airborne model, and the coefficient matrix L is directly calculated according to the formula in step 3 using the updated matrix R w and L u The value of .

[0030] Based on the same inventive concept, the following technical solutions can also be obtained:

[0031] An online identification device for the state space model matrix of an aircraft engine based on digital twins includes:

[0032] The auxiliary onboard model is the same as the main onboard model and is used to obtain the corresponding output data according to the control input u and health parameter excitation signal h input online in real time;

[0033] The online finite data driven identification model is used to perform online identification of the aircraft engine state space model matrix according to the input-output data of the auxiliary airborne model by the following method:

[0034] S1. Construct past and future row full-rank Hankel matrices based on the finite input-output data set of the auxiliary airborne model in the current finite time:

[0035]

[0036]

[0037]

[0038] Among them, X f is the Hankel matrix of future state variables; U p and U f are the Hankel matrices of past and future control variables; Y p and Y f are the Hankel matrices of past and future output variables respectively; p represents the past time domain; f represents the future time domain; n x Represents the dimension of the state variable x, n u represents the total dimension of the control variables u and health parameters h, n y Indicates the dimension of the output variable y;

[0039] S2. Construct an optimal prediction output estimator of the following form:

[0040]

[0041] in, represents the future estimated output of the optimal predicted output estimator; the coefficient matrix and By solving for the future estimated output and the future actual output Y f The least squares problem of : get; is the Frobenius 2 norm;

[0042] S3, according to the optimal prediction output estimator, the prediction output equation for the next f steps is obtained:

[0043]

[0044] in, m is the control time domain;

[0045] Then, by solving the following estimated output values ​​for the next k+1 and k+2 steps: The actual output value y k+1 ,y k+2 The simultaneous equations of the state matrix A, input matrix B, output matrix C, and transfer matrix D of the aircraft engine state space model are obtained:

[0046]

[0047] Preferably, the least squares problem is solved using the following method:

[0048] Step 1: Convert the solution to the least squares problem into matrix Y f The row space of the matrix [W p U f ] T The orthogonal projection is:

[0049]

[0050] Where, “ / ” indicates orthographic projection;

[0051] Step 2: Use QR decomposition to decompose the following combined Hankel matrix:

[0052]

[0053] in, is an upper triangular matrix, is an orthogonal matrix;

[0054] Step 3: Get the coefficient matrix L by the following formula: w and L u Values:

[0055]

[0056] The superscript represents the generalized inverse.

[0057] Further preferably, the matrix R is updated in real time using the input-output data updated in real time by the auxiliary airborne model, and the coefficient matrix L is directly calculated according to the formula in step 3 using the updated matrix R w and L u The value of .

[0058] The digital twin-based aero-engine performance parameter perception method uses a main onboard model to perceive the aero-engine performance parameters; at the same time, according to the deviation between the main onboard model and the actual engine measurement, a linear Kalman filter is used to estimate the degradation of the aero-engine health parameters, and the main onboard model is updated using the estimated degradation of the health parameters; the aero-engine state space model matrix used by the linear Kalman filter is obtained using the following method:

[0059] Step 1: During the first flight, the state space model matrix of the aircraft engine is obtained by using the method as described in any one of claims 1 to 3 and provided to a linear Kalman filter; at the same time, the input parameters of each operating point of the aircraft engine are clustered, and a sample database is constructed using the obtained cluster centers of each class and the corresponding aircraft engine state space model matrix as samples;

[0060] Step 2: After the first flight, define the envelope area with frequent flights during the entire life cycle of the engine as the high flight area, and the envelope area with fewer flights as the low flight area. The samples in the current sample database are divided into high flight areas and low flight areas. The input parameters of the high flight area samples in the sample database are used as training inputs, and the corresponding aircraft engine state space model matrix is ​​used as the expected output to train the deep neural network and obtain a deep neural network model for the high flight area.

[0061] Step 3: When flying online again, determine whether the minimum distance between the input parameters of the current working point of the aircraft engine and the input parameters of the samples in each high flight area in the current sample database exceeds a preset threshold value. If so, use the method described in any one of claims 1 to 3 to obtain the aircraft engine state space model matrix and provide it to the linear Kalman filter; otherwise, input the input parameters of the current working point of the aircraft engine into the deep neural network model of the high flight area, and provide the aircraft engine state space model matrix output by the deep neural network model to the linear Kalman filter; at the same time, cluster the input parameters of the current working point of the aircraft engine, and use the obtained cluster center and its corresponding state space model matrix as new samples to update the sample database;

[0062] Step 4: After the flight, the input parameters of the high flight area samples in the updated sample database are used as training inputs, and the corresponding aircraft engine state space model matrix is ​​used as the expected output to retrain the deep neural network model offline to obtain an updated deep neural network model of the high flight area; Step 5: When flying again, go to step 3.

[0063] Preferably, the clustering is performed using a Gaussian mixture model clustering method.

[0064] Preferably, the distance between the input parameter of the current operating point of the aircraft engine and the input parameter of each high flight area sample in the current sample database is the Mahalanobis distance.

[0065] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0066] The online identification method of the state space model matrix of an aircraft engine based on digital twin proposed in the present invention can perform online high-precision identification of the state space model matrix of an aircraft engine, solves the problem that it is difficult to obtain an accurate linear state space model within the full envelope range in the airborne adaptive model, and can realize online real-time high-precision perception of the performance parameters of the aircraft engine;

[0067] The present invention further combines the above-mentioned digital twin-based aero-engine state-space model matrix online identification method with a data-driven deep neural network. Through offline / online learning of the engine state-space model matrix, the real-time performance of the state-space model matrix calculation in the high flight envelope region of the engine is improved, and the waste of computing resources is avoided, thereby achieving a good coordination and unity of high precision and high real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a block diagram of the implementation principle of the method for sensing performance parameters of an aircraft engine according to the present invention;

[0069] Figure 2 Schematic diagram of the principle of high / low flight area division. DETAILED DESCRIPTION

[0070] In response to the technical problem of insufficient performance perception accuracy of existing airborne adaptive models based on linear Kalman filters, the solution of the present invention is to realize high-precision online identification of the aircraft engine state space model matrix through the limited input-output data set of the auxiliary airborne model based on digital twin technology, thereby realizing online and real-time high-precision perception of the performance parameters of the aircraft engine.

[0071] In order to facilitate public understanding, the technical solution of the present invention is described in detail below with reference to the accompanying drawings:

[0072] like Figure 1As shown, the present invention adds an auxiliary airborne model to the original airborne adaptive model architecture based on the linear Kalman filter. The auxiliary airborne model is the same as the main airborne model. It simulates the working state of the standard engine and can provide the output data of the standard engine in real time according to the same control input as the real engine. When the health parameter input excitation signal is given at the same time, the output data corresponding to the control input u and the health parameter excitation signal h as input can be obtained; then, based on the collected input of the auxiliary airborne model in the current period (control input u and health parameter excitation signal h) and the corresponding data set of the auxiliary airborne model output, an online finite data driven identification is designed. The model performs online identification on this finite input-output data set to obtain the engine state-space model matrix (state matrix A, input matrix B, output matrix C, transfer matrix D): first, based on the finite input-output data set, past and future row full-rank Hankel matrices are constructed, and an optimal prediction output estimator similar to the state-space model form is designed; then, the engine state-space model matrix is ​​solved according to the optimal prediction output estimator coefficient matrix; finally, the linear Kalman filter estimates the engine health parameters according to this state-space model matrix to update the main onboard model, thereby enabling the main onboard model to accurately estimate the true values ​​of the engine's performance parameters such as thrust and surge margin throughout its life cycle.

[0073] The basic implementation principle of the online limited data driven identification model proposed in this invention is as follows:

[0074] Using continuous excitation input signal u near a certain steady-state point of the engine k (k=1,2,…,N) acts on the system continuously, and obtains an input / output (I / O) data set D(U,Y ct ) Construct the past and future row full-rank Hankel matrices and obtain:

[0075]

[0076] Among them, X a,f is the Hankel matrix of future state variables; U p and U f are the Hankel matrices of past and future control variables; Y p and Y f are the Hankel matrices of past and future output variables respectively; p represents the past time domain; f represents the future time domain, generally p = f; x a is the augmented state vector of the engine high / low pressure shaft speed and all health parameters, n x Represents x a The dimension of u Indicates the dimension of the control variable u, n yRepresents the dimension of the output variable y (such as fan speed Nf and core engine speed Nc, duct inlet total pressure Pt17, high-pressure compressor inlet total temperature Tt25, total pressure Pt25 and outlet total temperature Tt30, static pressure Ps30, high-pressure turbine outlet total temperature Tt45, etc.).

[0077] The state space model expression of the engine steady-state point is as follows:

[0078]

[0079] Among them, A, B, C, and D matrices are state matrix A, input matrix B, output matrix C, and transfer matrix D respectively; k is the discrete time.

[0080] Therefore, according to the linear state space model, Y f With X f and U f The linear relationship is as follows:

[0081]

[0082] Among them, Φ x is the extended observable matrix; H u is a lower triangular Toeplitz matrix.

[0083] Therefore, according to the above formula, Y f The expression form of is to design the future optimal prediction output estimator based on the past and future Hankel matrices. The specific form is as follows:

[0084]

[0085] in, and The coefficient matrices of the online data-driven prediction estimator representing the past I / O data and future input data, respectively; W p is the combination of past input / output Hankel matrices, defined as

[0086]

[0087] Next, by constructing the estimated output and the future actual output Y f The least squares problem is used to solve L w and L u Values:

[0088]

[0089] in, is the Frobenius 2 norm.

[0090] According to matrix theory, the solution to the above equation can be understood as the matrix Y f The row space of the matrix [W p U f ] T The orthogonal projection is:

[0091]

[0092] Here, “ / ” indicates orthographic projection.

[0093] The following combined Hankel matrix is ​​decomposed using the QR decomposition method

[0094]

[0095] in, is an upper triangular matrix, is an orthogonal matrix.

[0096] So, L w and L u The values ​​are:

[0097]

[0098] The superscript represents the generalized inverse.

[0099] The prediction output equation for the next f steps is:

[0100]

[0101] in, m is the control time domain.

[0102] Therefore, the estimated output values ​​for the next k+1 and k+2 steps are The actual output value y k+1 ,y k+2 By combining the equations of , we can get:

[0103]

[0104] According to this equation, the C matrix can be solved: Cx k+1 =L u (1:n y ,:)w p , and then, solve this equation according to the solved C matrix to obtain the A matrix: CAx k+1 =L u (n y +1:2n y ,:)w p ; D = L u (1:n y ,1:nu ), according to the solved C matrix, solve this equation to obtain the B matrix: CB = L u (n y +1:2n y ,1:2n u ).

[0105] The linear Kalman filter is designed using the obtained A, B, C, and D matrices. m ) and the airborne model estimated parameters (y e ) is used to estimate the engine health parameters, thereby enabling the main onboard model to adapt to the entire life cycle of the engine and obtain the performance parameter values ​​such as thrust and surge margin required for engine control.

[0106] In order to adapt to the changes in the nonlinear working state of the engine and avoid the calculation of high-dimensional matrices at every moment, the present invention further uses the newly obtained input-output data at each moment to update the R matrix, and then uses the updated matrix R to directly calculate the coefficient matrix L w and L u The value of .

[0107] Generally speaking, within the flight envelope of an engine throughout its life cycle, some areas fly frequently, while others fly less frequently. In particular, in areas with frequent flights, the engine state space model matrix will be repeatedly calculated based on the online finite data driven identification model during each flight. In order to avoid the waste of computing resources caused by repeated calculation of the state space model matrix at the same operating point during each flight, the present invention further combines the above-mentioned online identification method of the state space model matrix of an aviation engine based on digital twins with a data-driven deep neural network, and through offline / online learning of the engine state space model matrix, the real-time performance of the calculation of the state space model matrix in the high flight envelope area of ​​the engine is improved to avoid waste of computing resources.

[0108] like Figure 1 As shown, in the process of sensing the performance parameters of the aircraft engine, the present invention uses the main onboard model to sense the performance parameters of the aircraft engine; at the same time, according to the deviation between the main onboard model and the actual engine measurement, the linear Kalman filter is used to estimate the degradation amount of the health parameters of the aircraft engine, and the estimated degradation amount of the health parameters is used to update the main onboard model; the aircraft engine state space model matrix used by the linear Kalman filter is obtained using the following method:

[0109] Step 1: During the first flight, an online finite data driven identification model is used to obtain an aircraft engine state space model matrix and provide it to a linear Kalman filter; at the same time, the input parameters of each operating point of the aircraft engine are clustered, and a sample database is constructed using the obtained class centers of each class and the corresponding aircraft engine state space model matrix as samples;

[0110] Step 2: After the first flight, define the envelope area with frequent flights during the engine's life cycle as the high flight area, and the envelope area with fewer flights as the low flight area. The samples in the current sample database are divided into high flight areas and low flight areas (the division principle can be found in Figure 2 ), and use the input parameters of the high flight area samples in the sample database as training inputs, and use the corresponding aircraft engine state space model matrix as the expected output to train the deep neural network and obtain a deep neural network model for the high flight area;

[0111] Step 3: When flying online again, determine whether the minimum distance between the input parameters of the current working point of the aircraft engine and the input parameters of the samples in each high flight area in the current sample database exceeds a preset threshold value. If so, use the online finite data driven identification model to obtain the aircraft engine state space model matrix and provide it to the linear Kalman filter; otherwise, input the input parameters of the current working point of the aircraft engine into the deep neural network model of the high flight area, and provide the aircraft engine state space model matrix output by the deep neural network model to the linear Kalman filter; at the same time, cluster the input parameters of the current working point of the aircraft engine, and use the obtained cluster center and its corresponding state space model matrix as new samples to update the sample database;

[0112] Determine whether the minimum distance between the input parameters of the current working point of the aircraft engine and the input parameters of each high-flight area sample in the current sample database exceeds a preset threshold value, which is essentially to determine whether the deep neural network model of the current high-flight area is suitable for the current working point; the distance can be measured by Euclidean distance, Mahalanobis distance, Manhattan distance and other measurement methods, and the Mahalanobis distance is used in this embodiment; the data can be clustered by various existing clustering methods, and the Gaussian mixture model clustering method is used in this embodiment;

[0113] like Figure 1 As shown, if the judgment is no, switches 1 and 2 are both turned to state 1, and the environment and control input u such as altitude (H), Mach number (Ma), ambient temperature (T0) / pressure (P0) are input into the deep neural network model, and the state space model matrix is ​​extracted to the linear Kalman filter estimator;

[0114] If the judgment is yes, then switches 1 and 2 are both turned to state 2, and the state space model matrix is ​​solved by the online finite data driven identification model;

[0115] At the same time, Gaussian mixture model clustering is performed on the environment and control inputs such as altitude (H), Mach number (Ma), ambient temperature (T0) / pressure (P0), and the obtained class center and its corresponding state space model matrix are used as new samples to update the sample database to obtain an updated sample database; Step 4: After the flight, the input parameters of the high flight area samples in the updated sample database are used as training inputs, and the corresponding aircraft engine state space model matrix is ​​used as the expected output, and the deep neural network model is retrained offline to obtain an updated deep neural network model of the high flight area;

[0116] In the offline state, switch 3 is turned to the offline state, and the data samples in the updated sample database are divided into high / low flight areas, which can be divided manually or by software; then the deep neural network model is retrained offline with the high flight area samples, so as to obtain an updated deep neural network model of the high flight area;

[0117] Step 5. When flying again, go to step 3.

[0118] Through the above method, the sample database is continuously updated during the cyclical flight of the engine throughout its life cycle. Finally, the high / low flight area division tends to be stable, and the deep neural network model of the high flight area also tends to be stable, which can realize the rapid extraction of the state space model matrix of the high flight area system.

Claims

1. An online identification method for the state space model matrix of an aero-engine based on digital twins, characterized in that: The control input u and the health parameter excitation signal h are input online and in real time into an auxiliary airborne model that is identical to the main airborne model to obtain the corresponding output data of the auxiliary airborne model; and based on the input-output data of the auxiliary airborne model, the aircraft engine state space model matrix is ​​identified online by the following method: S1. Construct past and future row full-rank Hankel matrices based on the finite input-output data set of the auxiliary airborne model in the current finite time: Among them, X f is the Hankel matrix of future state variables; U p and U f are the Hankel matrices of past and future control variables; Y p and Y f are the Hankel matrices of past and future output variables respectively; p represents the past time domain; f represents the future time domain; n x Represents the dimension of the state variable x, n u represents the total dimension of the control variables u and health parameters h, n y Indicates the dimension of the output variable y; S2. Construct an optimal prediction output estimator of the following form: in, represents the future estimated output of the optimal predicted output estimator; the coefficient matrix and By solving for the future estimated output and the future actual output Y f The least squares problem of : get; is the Frobenius 2 norm; S3, according to the optimal prediction output estimator, the prediction output equation for the next f steps is obtained: in, m is the control time domain; Then, by solving the following estimated output values ​​for the next k+1 and k+2 steps: The actual output value y k+1 ,y k+2 The simultaneous equations of the state matrix A, input matrix B, output matrix C, and transfer matrix D of the aircraft engine state space model are obtained:

2. The method for online identification of the state space model matrix of an aircraft engine based on digital twins according to claim 1, characterized in that: The least squares problem is solved using the following method: Step 1: Convert the solution to the least squares problem into matrix Y f The row space of the matrix [W p U f ] T The orthogonal projection is: Where " / " indicates orthographic projection; Step 2: Use QR decomposition to decompose the following combined Hankel matrix: in, is an upper triangular matrix, is an orthogonal matrix; Step 3: Get the coefficient matrix L by the following formula: w and L u Values: The superscript represents the generalized inverse.

3. The method for online identification of the state space model matrix of an aircraft engine based on digital twins as claimed in claim 2, characterized in that: The matrix R is updated in real time using the input-output data updated in real time by the auxiliary airborne model, and the coefficient matrix L is directly calculated according to the formula in step 3 using the updated matrix R w and L u The value of .

4. An online identification device for the state space model matrix of an aircraft engine based on digital twins, characterized in that: include: The auxiliary onboard model is the same as the main onboard model and is used to obtain the corresponding output data according to the control input u and health parameter excitation signal h input online in real time; The online finite data driven identification model is used to perform online identification of the aircraft engine state space model matrix according to the input-output data of the auxiliary airborne model by the following method: S1. Construct past and future row full-rank Hankel matrices based on the finite input-output data set of the auxiliary airborne model in the current finite time: Among them, X f is the Hankel matrix of future state variables; U p and U f are the Hankel matrices of past and future control variables; Y p and Y f are the Hankel matrices of past and future output variables respectively; p represents the past time domain; f represents the future time domain; n x Represents the dimension of the state variable x, n u represents the total dimension of the control variables u and health parameters h, n y Indicates the dimension of the output variable y; S2. Construct an optimal prediction output estimator of the following form: in, represents the future estimated output of the optimal predicted output estimator; the coefficient matrix and By solving for the future estimated output and the future actual output Y f The least squares problem of : get; is the Frobenius 2 norm; S3, according to the optimal prediction output estimator, the prediction output equation for the next f steps is obtained: in, m is the control time domain; Then, by solving the following estimated output values ​​for the next k+1 and k+2 steps: The actual output value y k+1 ,y k+2 The simultaneous equations of the state matrix A, input matrix B, output matrix C, and transfer matrix D of the aircraft engine state space model are obtained:

5. The online identification device for the state space model matrix of an aircraft engine based on digital twin as claimed in claim 4, characterized in that: The least squares problem is solved using the following method: Step 1: Convert the solution to the least squares problem into matrix Y f The row space of the matrix [W p U f ] T The orthogonal projection is: Where " / " indicates orthographic projection; Step 2: Use QR decomposition to decompose the following combined Hankel matrix: in, is an upper triangular matrix, is an orthogonal matrix; Step 3: Get the coefficient matrix L by the following formula: w and L u Values: The superscript represents the generalized inverse.

6. The online identification device for the state space model matrix of an aircraft engine based on digital twin according to claim 5, characterized in that: The matrix R is updated in real time using the input-output data updated in real time by the auxiliary airborne model, and the coefficient matrix L is directly calculated according to the formula in step 3 using the updated matrix R w and L u The value of .

7. The digital twin-based aero-engine performance parameter perception method uses a main onboard model to perceive the aero-engine performance parameters; at the same time, according to the deviation between the main onboard model and the actual engine measurement, a linear Kalman filter is used to estimate the degradation of the aero-engine health parameters, and the main onboard model is updated using the estimated degradation of the health parameters; it is characterized in that: The aircraft engine state space model matrix used by the linear Kalman filter is obtained using the following method: Step 1: During the first flight, the state space model matrix of the aircraft engine is obtained by using the method as described in any one of claims 1 to 3 and provided to a linear Kalman filter; at the same time, the input parameters of each operating point of the aircraft engine are clustered, and a sample database is constructed using the obtained cluster centers of each class and the corresponding aircraft engine state space model matrix as samples; Step 2: After the first flight, define the envelope area with frequent flights during the entire life cycle of the engine as the high flight area, and the envelope area with fewer flights as the low flight area. The samples in the current sample database are divided into high flight areas and low flight areas. The input parameters of the high flight area samples in the sample database are used as training inputs, and the corresponding aircraft engine state space model matrix is ​​used as the expected output to train the deep neural network and obtain a deep neural network model for the high flight area. Step 3: When flying online again, determine whether the minimum distance between the input parameter of the current working point of the aircraft engine and the input parameter of each high flight area sample in the current sample database exceeds a preset threshold value. If so, use the method as described in any one of claims 1 to 3 to obtain the aircraft engine state space model matrix and provide it to the linear Kalman filter; Otherwise, inputting the input parameters of the current operating point of the aircraft engine into the deep neural network model of the high flight area, and providing the aircraft engine state space model matrix output by the deep neural network model to the linear Kalman filter; At the same time, the input parameters of the current working point of the aircraft engine are clustered, and the obtained cluster centers and their corresponding state space model matrices are used as new samples to update the sample database; Step 4: After the flight, the input parameters of the high flight area samples in the updated sample database are used as training inputs, and the corresponding aircraft engine state space model matrix is ​​used as the expected output, and the deep neural network model is retrained offline to obtain an updated deep neural network model of the high flight area; Step 5. When flying again, go to step 3.

8. The method for sensing performance parameters of an aircraft engine based on digital twins as claimed in claim 7, characterized in that: The clustering was performed using a Gaussian mixture model clustering method.

9. The method for sensing performance parameters of an aircraft engine based on digital twins as claimed in claim 7, characterized in that: The distance between the input parameter of the current working point of the aircraft engine and the input parameter of each high flight area sample in the current sample database is the Mahalanobis distance.

Images