A Thrust Compensation Control Method for Aeroengine Performance Degradation

Abstract

The present invention belongs to the technical field of aeroengine control, and relates to a thrust compensation control method for aeroengine performance degradation, mainly including a robust adaptive Kalman filter and an error feedback controller. Among them, the robust adaptive Kalman filter is used for estimating the engine degradation factor, and can improve the estimation accuracy compared with the traditional Kalman filter; the multi-layer recursive error feedback controller is used for compensating and controlling the estimated degraded thrust, and can significantly eliminate the influence brought by errors and disturbances compared with the traditional controller. Compared with the traditional method, the present invention has the advantages of high estimation accuracy for the degradation factor and less influence of the controller by interference and errors.

Description

Technical Field

[0001] The present invention belongs to the technical field of aero-engine control, and relates to a thrust compensation control method for aero-engine performance degradation. Specifically, it is a method for estimating the unmeasurable performance parameters of an aero-engine with performance degradation and designing a thrust compensation controller. Background Art

[0002] An aero-engine is the power source for an aircraft's flight, and its structure is very precise and complex. Due to the influence of the harsh working environment of high temperature and high pressure and component wear during flight, the components of an aero-engine are prone to performance degradation, which in turn affects the thrust performance. Therefore, thrust compensation control for aero-engine performance degradation is very important.

[0003] In order to alleviate the adverse effects brought about by performance degradation, a large number of research works on performance degradation mitigation control have been carried out at home and abroad. To address the problem of thrust reduction caused by aero-engine performance degradation, methods such as increasing outer-loop control to adjust the speed control system command and then restoring the rated engine thrust have been adopted. Direct thrust control can also enable the engine to output the rated state thrust to the maximum extent when performance degradation occurs, but it requires large-scale improvement of the existing control method and is affected by the thrust estimation accuracy, and the technology maturity cannot meet the requirements. An effective way to solve this problem is to use the Kalman filter algorithm to estimate the measurable degradation factor of the aero-engine, and then compensate for the degraded thrust through an advanced controller according to the performance degradation situation. Existing literature shows that the classical Kalman filter algorithm has been widely used in the estimation of unmeasurable parameters of aero-engines. However, due to its nature as a model-based estimation method, factors such as model uncertainty, differences from the actual system, external disturbances, and abnormal fluctuations of estimated parameters will seriously affect its estimation performance. At the same time, most outer-loop control strategies have poor anti-interference and anti-noise performance and are greatly affected by the errors and uncertainties generated in the modeling process. Therefore, in view of the above problems, it is of great engineering significance to design a thrust compensation control method for aero-engine performance degradation. Summary of the Invention

[0004] Since the existing classical Kalman filter and H ∞The thrust compensation controller composed of a controller has problems such as poor estimation accuracy in its outer-loop control method, and there are also differences between the estimated performance degradation factor and the true degradation degree. Therefore, there is also an error between the output of the aero-engine on-board adaptive model and the reference command. To address the above problems, the present invention proposes a thrust compensation controller method for aero-engine performance degradation, aiming to estimate the degradation factor through a designed robust adaptive Kalman filter when components such as the low-pressure turbine and fan of the aero-engine experience performance degradation, and further design a thrust compensation controller to keep the output thrust of the degraded aero-engine the same as that in the non-degraded state.

[0005] To achieve the above object, the present invention first uses a state estimator to estimate the degradation factor. Based on the traditional Kalman filter algorithm, an improved robust adaptive Kalman filter algorithm is proposed to reduce the estimation error caused by the uncertainty of the process noise covariance Q and the observation noise covariance R, so as to improve the estimation accuracy of the degradation factor. Then, a multi-layer recursive error feedback controller is designed and selected according to the estimated degraded thrust for thrust compensation.

[0006] The technical solution of the present invention:

[0007] A thrust compensation control method for aero-engine performance degradation, the specific steps are as follows:

[0008] Step 1: Establish a Kalman filter based on the aero-engine state space model to estimate the degradation factor:

[0009] Step 1.1: At a certain steady-state operating point, use the partial derivative method to obtain the state variable model of the aero-engine:

[0010]

[0011] Among them, A, B, C, and D represent the system matrices of the linear system, x, u, and y represent the state variable, input variable, and output variable respectively, represents the updated state of the state variable, where x = [N1 N2] T , u = [W f , A8, c elt T , y = [N1, N2, sP 31 , tT6] T , N1 is the low-pressure rotor speed, N2 is the high-pressure rotor speed, W f is the main fuel flow rate, A8 is the area of the tail pipe, c elt is the deviation of the low-pressure turbine efficiency, sP 31 is the total pressure at the compressor outlet, and tT6 is the total temperature at the low-pressure turbine outlet.

[0012] ​Step 1.2: Expand the aero-engine state variable model so that the degradation factor is an observable state variable:

[0013]

[0014] C ext = [C D(:,3)], D ext = [D(:,1:2)]

[0015] where A ext , B ext , C ext , D ext are the expanded system matrices. B(:,3) means taking the third column of matrix B, B(:,1:2) means taking the first two columns of matrix B, D(:,3) means taking the third column of matrix D, D(:,1:2) means taking the first two columns of matrix D, zeros(1,3) means a 1×3 zero matrix, and zeros(1,2) means a 1×2 zero matrix.

[0016] Step 1.3: Solve the Riccati equation to obtain the prediction error covariance P and the Kalman gain K, and form the traditional Kalman filter. The performance degradation estimation equation at a certain steady state point of the aero-engine can be expressed as follows:

[0017]

[0018]

[0019] where x ext is the expanded state variable, is the expanded input variable, K is the Kalman gain, y real is the true engine output value, y0 is the steady-state output corresponding to the steady state point where the state variable model is established. Since the output variable remains unchanged, y remains the same. Thus, the required degradation factor estimator is the component of the state variable x ext .

[0020] Step 2: Design an improved robust adaptive Kalman filter.

[0021] Step 2.1: Propose an adaptive factor Ω k , where k is the iteration number. Ω k is determined by the innovation sequence modulus ε k and the current covariance square root of the measurement noise :

[0022]

[0023] where the innovation sequence modulus Y kis the output of the current iterative system, C k is the current iterative system matrix is the predicted system state

[0024] Step 2.2: Solve the robust adaptive Kalman filter gain K k :

[0025]

[0026] where is the predicted error covariance, and R is the measurement noise covariance k Combined with the estimation equation of the adaptive factor:

[0027]

[0028] is the process noise covariance Q k Combined with the estimation equation of the adaptive factor:

[0029]

[0030] where the modulus of the residual sequence

[0031] Step 2.3: Design the structure of the robust adaptive Kalman filter. The performance degradation estimation equation at a certain steady state point of the aero-engine can be expressed as follows:

[0032]

[0033]

[0034] Step 3: Design a multi-layer recursive error feedback controller to track and eliminate the degraded thrust error

[0035] Step 3.1: First, establish the first-layer subsystem of the error feedback controller, that is, the inner loop control system of the real engine:

[0036]

[0037] where A z , B d , B z , C z are the system matrices of the linear subsystem, d is the interference signal, x z is the state variable of the subsystem, u z is the input variable of the subsystem, y z is the output variable of the subsystem. Let the input signal u z = W f , W f represents the fuel quantity, and let the output signal y z= Thrust, where Thrust represents the thrust force.

[0038] Define the tracking error:

[0039] e = r - y z

[0040] where r is the reference signal, i.e., the desired thrust.

[0041] Step 3.2: Design a recursive error compensation controller:

[0042]

[0043] where:

[0044] represents the input vector for the nth recursion, y n represents the output vector for the nth recursion, e n represents the error vector for the nth recursion, G = C(-A) -1 B z , β is the regularization parameter, and

[0045] The recursive error of the system is:

[0046] e n = e n-1 - y n

[0047] Since the output of each subsequent system attempts to track the error of the previous subsystem, e n is gradually converging. The outputs of each order system are respectively:

[0048]

[0049] Step 3.3: Add N subsystems to approximately cancel out the disturbances and errors existing in the original system, achieving the purpose of tracking thrust compensation.

[0050]

[0051] Step 4: Connect the engine model with the Kalman filter and the multi-layer recursive error feedback controller to form the outer loop control for thrust compensation of aero-engine performance degradation. Then, pass the estimated thrust through the multi-layer recursive error feedback controller designed in Step 3 to calculate the fuel flow rate that needs to be compensated, realizing closed-loop control. This enables the accurate adjustment of the fuel flow rate to maintain the output thrust the same as in the non-degraded state when the aero-engine degrades.

[0052] Advantages of the present invention: A thrust compensation control method for the performance degradation of an aeroengine proposed by the present invention mainly includes a robust adaptive Kalman filter and an error feedback controller. Specifically: The robust adaptive Kalman filter is used for estimating the engine degradation factor, and can improve the estimation accuracy compared with the traditional Kalman filter; The multi-layer recursive error feedback controller is used for compensating and controlling the estimated degraded thrust, and can significantly eliminate the influence of errors and disturbances compared with the traditional controller. Compared with the traditional method, the present invention has the advantages of high estimation accuracy for the degradation factor and less influence of the controller by disturbances and errors. Brief Description of the Drawings

[0053] Figure 1 is a structural diagram for compensating the degraded thrust based on the Kalman filter and the error feedback controller.

[0054] Figure 2 is a schematic structural diagram of the robust adaptive Kalman filter in the present invention.

[0055] Figure 3 is a schematic structural diagram of the degraded thrust error compensation controller in the present invention.

[0056] Figure 4 is the estimation result of the robust adaptive Kalman filter when the aeroengine degrades in the embodiment.

[0057] Figure 5 is the result of the thrust degradation situation of the aeroengine after degradation and the compensation of the multi-layer recursive error feedback controller in the embodiment. Detailed Embodiment

[0058] The present invention will be further described below in conjunction with the drawings and specific embodiments.

[0059] A thrust compensation control method for the performance degradation of an aeroengine according to the present invention specifically comprises the following steps:

[0060] Step 1: Establish a Kalman filter to estimate the degradation factor based on the state space model of the aeroengine:

[0061] Step 1.1: At a certain steady-state operating point, obtain the state variable model of the aeroengine by using the partial derivative method:

[0062]

[0063]

[0064] where x, u, and y respectively represent the state variable, input variable, and output variable, represents the updated state of the state variable, and x = [N1 N2] T , u = [W f,A8,c elt T ,y = [N1,N2,sP 31 ,tT6] T , where N1 is the low-pressure rotor speed, N2 is the high-pressure rotor speed, W f is the main fuel flow rate, A8 is the area of the tail nozzle, c elt is the deviation of the low-pressure turbine efficiency, sP 31 is the total pressure at the compressor outlet, and tT6 is the total temperature at the low-pressure turbine outlet.

[0065] Step 1.2: Expand the aero-engine state variable model so that the degradation factor is an observable state variable:

[0066]

[0067]

[0068] where A ext , B ext , C ext , D ext are the expanded system matrices. B(:,3) represents taking the third column of matrix B, B(:,1:2) represents taking the first two columns of matrix B, D(:,3) represents taking the third column of matrix D, D(:,1:2) represents taking the first two columns of matrix D, zeros(1,3) represents a 1×3 zero matrix, and zeros(1,2) represents a 1×2 zero matrix.

[0069] Step 1.3: Solve the Riccati equation to obtain the prediction error covariance P and the Kalman gain K, forming a traditional Kalman filter. The performance degradation estimation equation at a certain steady state point of the aero-engine can be expressed as follows:

[0070]

[0071]

[0072] where x ext is the expanded state variable, u~ is the expanded input variable, K is the Kalman gain, y real is the true engine output value, y0 is the steady-state output corresponding to the steady state point where the state variable model is established. Since the output variable does not change, y remains the same. Thus, the required degradation factor estimator is a component of the state variable x ext .

[0073] Step 2: Design an improved robust adaptive Kalman filter.

[0074] Step 2.1: Propose an adaptive factor Ω k ​, where k is the number of iterations. Ω is determined by the innovation sequence modulus ε k and the current covariance square root of the measurement noise as follows:

[0075]

[0076] where the innovation sequence modulus Y k is the output of the current iteration system, C k is the current iteration system matrix, and is the predicted system state.

[0077] Step 2.2: Solve the robust adaptive Kalman filter gain K k as follows:

[0078]

[0079] where, is the predicted error covariance, and the measurement noise covariance R k is combined with the estimation equation of the adaptive factor:

[0080]

[0081] The process noise covariance Q k is combined with the estimation equation of the adaptive factor:

[0082]

[0083] where the modulus of the residual sequence

[0084] The measurement noise covariance R is obtained through iterative update k as follows:

[0085]

[0086] The process noise covariance Q is obtained through iterative update k as follows:

[0087]

[0088] Step 2.3: Design the robust adaptive Kalman filter structure, as shown in Figure 2 . The performance degradation estimation equation at a certain steady state point of the aeroengine can be expressed as follows:

[0089]

[0090]

[0091] The estimation result of the degradation factor of the aeroengine is obtained asFigure 4 as shown

[0092] Step 3: Design a multi-layer recursive error feedback controller to track and eliminate the degraded thrust error.

[0093] Step 3.1: First, establish the first-layer subsystem of the error feedback controller, i.e., the real engine inner-loop control system:

[0094]

[0095] where A z , B d , B z , C z are the system matrices of the linear subsystem, d is the disturbance signal, x z is the state variable of the subsystem, u z is the input variable of the subsystem, and y z is the output variable of the subsystem. Let the input signal u z =W f , W f represents the fuel quantity, and let the output signal y z =Thrust, where Thrust represents the thrust.

[0096] Define the tracking error:

[0097] e = r - y z

[0098] where the reference signal, i.e., the desired thrust r = 11240.

[0099] Step 3.2: Design a recursive error compensation controller:

[0100]

[0101] where, represents the input vector of the nth recursion, y n represents the output vector of the nth recursion, e n represents the error vector of the nth recursion, G = C(-A) -1 B z = 0.4575, and β is the regularization parameter, taking β = 5.

[0102] Then

[0103] And the recursive error of the system is:

[0104] e n = e n-1 - y n

[0105] Since the output of each subsequent system attempts to track the error of the previous subsystem, e n gradually converges. The outputs of systems of each order are respectively:

[0106]

[0107] Step 3.3: Add N subsystems to approximately eliminate the interference and error existing in the original system, so as to achieve the purpose of tracking thrust compensation. The specific structure is as Figure 3 shown.

[0108]

[0109] Step 4: Connect the engine model with the Kalman filter and the multi-layer recursive error feedback controller to form the outer loop control for thrust compensation of aero-engine performance degradation, as Figure 1 shown. Calculate the fuel flow rate that needs to be compensated through the multi-layer recursive error feedback controller designed in the previous step for the estimated thrust, so as to achieve the effect of closed-loop control. When the aero-engine degrades, the fuel flow rate can be accurately adjusted to keep the output thrust the same as that in the non-degraded state, and the final effect is as Figure 5 shown.

Claims

1. A thrust compensation control method for aircraft engine performance degradation, characterized in that The specific steps are as follows: Step 1: Establish a Kalman filter to estimate the degradation factor based on the state space model of the aero-engine: Step 1.1: At a certain steady-state operating point, obtain the state variable model of the aero-engine using the partial derivative method: Wherein, A, B, C, and D represent the system matrices of the linear system, and x, u, and y represent the state variable, input variable, and output variable, respectively. represents the updated state of the state variable, where x = [N1 N2] T , N1 is the low-pressure rotor speed, N2 is the high-pressure rotor speed, W f is the main fuel flow rate, A8 is the area of the tail nozzle, c elt is the deviation of the low-pressure turbine efficiency, sP 31 is the total pressure at the compressor outlet, and tT6 is the total temperature at the low-pressure turbine outlet; Step 1.2: Expand the state variable model of the aero-engine so that the degradation factor is used as an observable state variable: C ext = [CD(:,3)], D ext = [D(:,1:2)] Among them, A ext , B ext , C ext , D ext are the expanded system matrices. B(:,3) means taking the third column of matrix B, B(:,1:2) means taking the first two columns of matrix B, D(:,3) means taking the third column of matrix D, D(:,1:2) means taking the first two columns of matrix D, zeros(1,3) means a 1×3 zero matrix, and zeros(1,2) means a 1×2 zero matrix; Step 1.3: Solve the Riccati equation to obtain the prediction error covariance P and the Kalman gain K, constituting a traditional Kalman filter; the performance degradation estimation equation at a certain steady-state point of the aero-engine is expressed as follows: where x ext is the expanded state variable, is the expanded input variable, K is the Kalman gain, y real is the true engine output value, y0 is the steady-state output corresponding to the steady-state point where the state variable model is established. Since the output variable has not changed, y remains the same; thus, the required degradation factor estimator is a component of the state variable x ext ; Step 2: Design an improved robust adaptive Kalman filter; Step 2.1: Propose the adaptive factor Ω k , where k is the number of iterations; Ω k is determined by the innovation sequence modulus ε k and the current covariance square root of the measurement noise : Among them, the innovation sequence modulus Y k is the output of the current iteration system, and C k is the matrix of the current iteration system, is the predicted system state; Step 2.2: Solve the robust adaptive Kalman filter gain K k : wherein, is the prediction error covariance, and the measurement noise covariance R k Combined with the estimation equation of the adaptive factor: Process noise covariance Q k Estimation equation combined with the adaptive factor: Among them, the modulus of the residual sequence Step 2.3: Design the structure of the robust adaptive Kalman filter, and the performance degradation estimation equation at a certain steady-state point of the aero-engine is expressed as follows: Step 3: Design a multi-layer recursive error feedback controller to track and eliminate the degraded thrust error; Step 3.1: First, establish the first-layer subsystem of the error feedback controller, that is, the inner loop control system of the real engine: Among which A z , B d , B z , C z are the system matrices of the linear subsystems, d is the disturbance signal, x z is the state variable of the subsystem, u z is the input variable of the subsystem, y z is the output variable of the subsystem; Let the input signal u z =W f , W f represents the fuel quantity, let the output signal y z =Thrust, Thrust represents the thrust; Define the tracking error: e = r - y z where r is the reference signal, that is, the desired thrust; Step 3.2: Design a recursive error compensation controller: where: represents the input vector for the n - th recursion, y n represents the output vector for the n - th recursion, e n represents the error vector for the n - th recursion, G = C(-A) -1 B z , β is the regularization parameter, and And the recursive error of the system is: e n = e n-1 -y n Since the output of each subsequent system attempts to track the error of the previous subsystem, e n gradually converges; the outputs of each order system are respectively: y0 = r - e0 Step 3.3: Add N subsystems to approximately eliminate the interference and error existing in the original system, so as to achieve the purpose of tracking thrust compensation; Step 4: Connect the engine model with the Kalman filter and the multi-layer recursive error feedback controller to form the outer loop control for the thrust compensation of the aero-engine performance degradation. Then, calculate the fuel flow rate that needs to be compensated through the multi-layer recursive error feedback controller designed in Step 3 for the estimated thrust, and realize closed-loop control; when the aero-engine degrades, accurately adjust the fuel flow rate to keep the output thrust the same as the non-degraded state.

Images