Iterative satellite propulsion system fault diagnosis method based on Luenberger observer

Abstract

The invention discloses an iterative satellite propulsion system fault diagnosis method based on a Luenberger observer, and the method comprises the steps: building a service satellite quality-fault-control input coupling model based on a service satellite relative pose equation and a service satellite propulsion system fault model, and then building a satellite state observation equation; based on the above, the Kalman filter based on the fault model is designed. The filter carries out noise reduction processing on relative pose data observed by a sensor on a service satellite to obtain a filtering result. And obtaining a fusion result from the filtering result through a fusion algorithm. And the fault diagnosis observer obtains a residual error based on the residual error, and the residual error is input into the fault estimator to obtain a fault estimator. After the fault propeller is judged, the fault estimator of the fault propeller is used for correcting the quality and rotational inertia of the service satellite, and the next round of diagnosis is turned to. According to the method, additional sensors are not needed; meanwhile, the method has the characteristics of low operand and high interpretability; and the method has robustness of causing changes to the work of the propeller.

Description

Technical Field

[0001] The present invention belongs to the field of advanced satellite platforms, and in particular relates to an iterative satellite propulsion system fault diagnosis method based on a Lumberg observer. Background Art

[0002] On-orbit service satellites use their onboard propulsion systems to achieve relative orbital maneuvers and attitude control, thereby completing complex on-orbit missions. When a spacecraft is in a complex space environment, subject to adverse factors such as high temperature and vacuum, a failure in the propulsion system can lead to mission losses or even failure.

[0003] The design of propulsion system diagnostic tasks requires comprehensive consideration of fault diagnosis principles, the type of data used, the acquisition method, and the sensor types and characteristics. Current methods for diagnosing propulsion systems on in-orbit servicing satellites face challenges such as limited accuracy of onboard equipment, limited available data types, high latency associated with offline diagnostics, and the inability to correct satellite thrust in real time.

[0004] In their paper "State Memory-Based Autonomous Fault Diagnosis Method for Spacecraft," Jin Yang et al. improved diagnostic efficiency by introducing a state memory mechanism, avoiding the drawback of traditional methods that ignore prior diagnostic results. However, this method is computationally intensive and difficult to apply on satellite platforms. In their paper "Application of Kernel Principal Component Analysis in Autonomous Fault Diagnosis of Spacecraft Flywheels," Nie Xiaohui et al. used Kernel Principal Component Analysis to process satellite quaternions and their derivatives, designing a diagnostic method suitable for a variety of fault scenarios, but not for those involving mass changes. Summary of the Invention

[0005] The present invention proposes an iterative satellite propulsion system fault diagnosis method based on the Lumberg observer, which solves the problems of existing satellite fault diagnosis methods being sensitive to satellite state changes and susceptibility to noise interference, and achieves the purpose of completing propulsion system diagnosis by only using the relative position of the service satellite and the target satellite when considering the mass moment of inertia and changes caused by propellant consumption.

[0006] The technical solution to implement the present invention is: an iterative satellite propulsion system fault diagnosis method based on a Lumberg observer, comprising the following steps:

[0007] Step 1: Based on the CW equation, construct the relative position equation of the service satellite and the target satellite; based on the working principle and structure of the satellite propulsion system, establish a propulsion system fault model and proceed to step 2.

[0008] Step 2: Based on the relative posture equation and the propulsion system fault model, a service satellite quality-fault-control input coupling model considering mass consumption is established. After linearization, the service satellite state observation equation is established. Based on the coupling model, a Kalman filter based on the fault model is designed, and then the process goes to step 3.

[0009] Step 3: Use the Kalman filter based on the fault model to perform noise reduction on the relative posture data observed by the sensors on the service satellite to obtain the filtering result, and then go to step 4.

[0010] Step 4: Process the filtering results to obtain the fusion results, and then go to step 5.

[0011] Step 5: Based on the service star mass and moment of inertia, design a fault diagnosis observer and a fault estimator. Input the fusion result into the fault diagnosis observer to generate a residual. Input the residual into the fault estimator to obtain the fault estimate, and then go to step 6.

[0012] Step 6: Use the threshold discrimination method to screen the faulty thrusters. If the residual meets the discrimination conditions, go to step 7; otherwise, complete the diagnosis.

[0013] Step 7: Use the fault estimation quantity to estimate the mass and moment of inertia of the service star, implement the mass and moment of inertia correction of the service star, and return to step 2.

[0014] Compared with the prior art, the present invention has the following significant advantages:

[0015] (1) The present invention is designed based on modern control theory and has higher interpretability than deep learning methods. The convergence and error term size of the present invention can be derived and proved by mathematical methods. Therefore, the obtained results are highly credible and more conducive to control engineering practices such as fault-tolerant control.

[0016] (2) The method of the present invention is easy to implement and requires little computation. It is suitable for satellite platforms with weak computing power and has high real-time performance. It can be applied as a real-time onboard diagnostic method without the need for ground station processing and other steps, thus avoiding adverse consequences such as greater losses to satellite missions caused by fault diagnosis delays.

[0017] (3) The present invention uses the fault detection results to update the calculation of the satellite mass and moment of inertia, and performs a new round of fault diagnosis and fault estimation based on the updated results. During the iterative process, the problem of changes in mass and moment of inertia caused by the propellant consumption of the propulsion system is gradually reduced, thereby ensuring the diagnostic convergence under the condition of considering the changes in mass and moment of inertia. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The figure is a flow chart of an iterative satellite propulsion system fault diagnosis method based on a Lumberg observer according to the present invention.

[0019] Figure 2 This is a schematic diagram of the on-orbit service scenario of the present invention.

[0020] Figure 3 This is a simulation result diagram of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The technical solutions between the various embodiments of the present invention can be combined with each other, but they must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0023] The following will further introduce the specific implementation methods, as well as the technical difficulties and inventive points of this invention in combination with this design example.

[0024] Combine Figure 1 , an iterative satellite propulsion system fault diagnosis method based on Lumberg observer, comprising the following steps:

[0025] In step 1, Figure 2 In the on-orbit service scenario shown, the relative posture equation between the service satellite and the target satellite is constructed according to the CW equation, as follows:

[0026]

[0027] in:

[0028]

[0029] Where: ρ(t) is the position of the service satellite relative to the target satellite, ρ(t) = r c (t)-r t (t), r c (t) is the position vector of the service satellite in the inertial system, r t (t) is the position vector of the target star in the inertial system, is the first-order derivative of ρ(t), A1 and A2 are coefficient matrices, ω is the angular velocity of the service satellite in the inertial system; u F (t) is the acceleration generated by the service satellite, u F (t)=[fx (t),f y (t),f z (t)] T , f x (t), f y (t), f z (t) are the accelerations along the coordinate axes in the service star system; t represents time, and T represents transposition.

[0030] Based on the working principle and structure of the satellite propulsion system, a propulsion system fault model is established, as follows:

[0031]

[0032] Where: F r is the actual control quantity generated by the thruster in the propulsion system; e(t) is the actual efficiency coefficient; F expect is the expected control quantity in each control cycle of the service star under fault-free conditions, and const is a constant. That is, when the thruster has an efficiency degradation fault that causes the valve to fail to reach the maximum opening stroke, the actual thrust generated is determined by the actual efficiency coefficient e(t); when the thruster has a stuck fault that causes the valve to be unable to close, the actual control quantity generated is a constant value.

[0033] Step 2: Based on the relative posture equation and the propulsion system fault model, a service satellite quality-fault-control input coupling model considering mass consumption is established. After linearization, the service satellite state observation equation is established. Based on the coupling model, a Kalman filter based on the fault model is designed, as follows:

[0034] According to the relative posture equation and propulsion system fault model, a service satellite quality-fault-control input coupling model considering mass consumption is established:

[0035] x(t+1)=Fx(t)+BE(t)u(t)+Γ x (t)w(t),

[0036] Among them: x(t+1) is the system state at time t+1, x(t) is the system state at time t, F is the state transfer matrix, B is the control matrix, E(t) is the mass change effect matrix, is the angular velocity of the service star relative to the target star in its own system; Γ x (t) is the process noise influence matrix; w(t) is the process noise; u(t) is the control instruction, u(t)=[u1(t),u2(t),…,u i (t),…,u n (t)] T ;u i(t) is the control instruction of the i-th thruster, and its physical meaning is the opening time of the jet valve of the i-th thruster, i = 1, 2, …, n, where n is the total number of thrusters.

[0037] The mass change effect matrix E(t) is as follows:

[0038]

[0039] Among them, the mass change effect coefficient of the i-th thruster is The effect coefficient of the moment of inertia change of the i-th thruster e i (t) is the actual efficiency coefficient of the i-th propeller, To estimate the quality of the service star; is the estimated moment of inertia of the service satellite; k is the mass of propellant ejected per unit time; f is the propellant ejection speed.

[0040] The service satellite state observation equation is established based on the linearization of the coupling model:

[0041] y(t+1)=Ηx(t+1)+Γ y v(t+1),

[0042] Where: y(t+1) is the observation information; Η is the observation matrix; Γ y is the observation noise influence matrix; v(t+1) is the observation noise.

[0043] The Kalman filter based on the fault model is designed based on the coupling model, and its structure is as follows:

[0044]

[0045]

[0046] in: represents the predicted value of the service star status at time t+1, represents the predicted state quantity of the service star at time t, It represents the predicted value of the service star status at time t; is the prediction covariance matrix at time t+1; P(t+1) is the covariance matrix at time t+1, P(t) is the covariance matrix at time t; Q(t) is the process noise covariance matrix; K(t+1) is the Kalman filter gain; R(t) is the observation noise covariance matrix; ε is the intermediate function, I is the unit matrix, and x(t+1) is the filtering result.

[0047] The observation noise w(t-1) and system noise v(t) are both column vectors, and they satisfy:

[0048]

[0049] w(j1) and v(j1) represent the observation noise and system noise at time j1, respectively; w(j2) and v(j2) represent the observation noise and system noise at time j2, respectively; δ j1j2 represents the Kroneck function, and v(j)~N(0,Σ v ), w(j)~N(0,Σ w ),Σ v is the covariance matrix of the system noise v(t), Σ w is the covariance matrix of the observation noise w(t). In this step, the fault estimator is introduced into the Kalman filter, allowing the Kalman filter to dynamically adapt to changes in the service star state and change the state prediction steps. This avoids problems such as decreased fault estimation accuracy or even divergence caused by increased state prediction errors after a fault occurs.

[0050] Step 3: Use the Kalman filter based on the fault model to perform noise reduction on the relative posture data observed by the sensors on the service satellite to obtain a filtering result.

[0051] Step 4: Process the filtering results to obtain the fusion results, as follows:

[0052] y best (t+1)=W1x1(t+1)+W2x2(t+1)+…W p x p (t+1);

[0053] Where: y best (t+1) is the fusion result; x l (t+1) is the optimal pose observation data of filter l, l=1,2,…,p, p is the total number of filters, the number of filters is equal to the number of sensors, that is, each sensor runs a filter independently, x l (t+1) is x(t+1) output in step 2.2; W m is the fusion weight coefficient matrix of the mth filter, m=1,2,…,p, and satisfies: [W1,W2,…,W l ,…,W p ] T =Σ1 -1 e(e T Σ1 -1 e) -1 ; Auxiliary matrix Σ1=diag(P1,P2,…,P m ,…,P p ), P m Represents the posterior covariance matrix of the filter of the mth sensor, the auxiliary matrix e=[I1,I2,…,I m ,…,Ip ] T , I m is the unit matrix. In this step, by fusing multiple sensors, we can make full use of the equipment on the service satellite to obtain the most reliable observation data, i.e., the fusion result.

[0054] In step 5, design the fault diagnosis observer and fault estimator as follows:

[0055] The fault diagnosis observers for stuck faults or efficiency reduction faults are as follows:

[0056] Stuck type:

[0057] r i (t) = Mz i (t)+Hy best (t),

[0058] Efficiency reduction category:

[0059] r i (t) = Mz i (t)+Hy best (t),

[0060] Where: r i (t) is the residual error of the i-th thruster; y best (t) is the fusion result obtained in step 4, B is the control matrix, E(t-1) is the mass change effect matrix, and H is the observation matrix; i (t) is the state vector of the Lumberg observer of the i-th thruster, which should be an observable that can reflect the working condition of the i-th thruster; if the i-th thruster completely and only affects ρ(t) when it is working, then z(t) takes ρ(t) instead of θ(t); u(t-1) represents the control variable of the i-th thruster; represents the estimated control quantity of the i-th thruster, which is estimated in step 5.2; Indicates that the i e The efficiency of the propellers decreased due to failure, Pick The value of is the estimated value of the actual efficiency coefficient, and for i=1,2,…,p and i≠i e There is e i (t) = 1, and calculate the mass change effect matrix based on this, and the auxiliary matrix F is to be determined z , G, L, M, C must meet the following conditions:

[0061] F z G+LC-GF=0,

[0062] MG+HC=0,

[0063] Re(λ(F z ))<0,

[0064] MG(BE) i ≠0,i=1,2,…,n,

[0065] Among them, Re() means taking the real part of a variable, and (BE) means taking the eigenvalue of the matrix; i Represents the i-th column of the matrix BE; λ() represents taking the eigenvalue of the variable.

[0066] The fault estimators for stuck fault and efficiency degradation fault are:

[0067] Stuck type:

[0068] Efficiency reduction category:

[0069] Where: is the estimated value of the stuck fault of the i-th thruster, which represents the actual thrust value estimated when the stuck fault occurs. is its first-order derivative; is the efficiency degradation failure estimator of the i-th thruster, which represents the estimated value of the ratio of the actual thrust to the expected thrust when the efficiency degradation failure occurs. for The first derivative of ; and Collectively referred to as fault estimator; i is the auxiliary vector of the i-th thruster, and when M columns are full rank, ζ i =M(M T M) -1 P p G(BE) i , otherwise we can get ζ by estimating i The value of η i is the undetermined auxiliary vector of the i-th thruster, satisfying M T η i =PG(BE) i ρ i is the adaptive rate coefficient of the i-th thruster, the auxiliary matrix P p is a symmetric positive definite matrix, which should satisfy: Π is an arbitrary positive definite matrix. In this step, a fault estimator is introduced to dynamically update the observer, and its convergence can be proved.

[0070] Step 6: The threshold discrimination method is as follows. The residual is further processed by the following formula to obtain the fault separation result:

[0071]

[0072] Where: S i (t) is the diagnostic index of the i-th thruster at time t, For the i e The diagnostic index of the thruster at time t is greater than the threshold; N s is the number of samples in the sliding window, S th is the threshold; the diagnostic parameter of the i-th thruster σ i When there is no fault i The standard deviation of (t).

[0073] Calculate the diagnostic index of each thruster separately and compare the diagnostic index of all thrusters in the manner shown above; if there is only one Satisfy the above formula, that is, judge the i-th e This step dynamically collects the residual variance using a sliding window method to prevent diagnostic errors caused by noise fluctuations in the service satellite sensor under certain circumstances.

[0074] Step 7: Use the fault estimation to estimate the mass and moment of inertia of the service star, and implement the mass and moment of inertia correction of the service star, as follows:

[0075] When an efficiency drop fault occurs:

[0076]

[0077] When a stuck fault occurs:

[0078]

[0079] in: is the estimated service quality at time t+1; is the moment of inertia of the service star at time t+1, m c (0) is the initial mass of the service star; t s is the time when the fault occurs, ||d||2 is the distance from the propellant tank to the satellite's center of mass, is the i-th e The control quantity of the faulty thruster, u i (t1) is the control quantity of the i-th thruster at time t1.

[0080] Example 1: When the efficiency drop fault occurs at t=100 and the actual efficiency coefficient is 0.5, the effect is as follows Figure 3As shown in the figure, at the beginning of the diagnosis, the modeling error is large due to the changes in mass and moment of inertia, and the error in the actual efficiency coefficient estimate is also large. Subsequently, as the iterations proceed, the modeling error gradually decreases, and the actual efficiency coefficient estimate approaches the true value.

Claims

1. An iterative satellite propulsion system fault diagnosis method based on Lumberg observer, characterized in that: The following steps are involved: Step 1: Based on the CW equation, construct the relative position equation of the service satellite and the target satellite; based on the working principle and structure of the satellite propulsion system, establish the propulsion system fault model and proceed to step 2; Step 2: Based on the relative posture equation and the propulsion system fault model, a service satellite quality-fault-control input coupling model considering mass consumption is established, and the service satellite state observation equation is established after linearization; Design a Kalman filter based on the fault model based on the coupling model and proceed to step 3; Step 3: Use the Kalman filter based on the fault model to perform noise reduction on the relative posture data observed by the sensor on the service satellite to obtain the filtering result, and then go to step 4; Step 4: Process the filtering results to obtain the fusion results, and then go to step 5; Step 5: Based on the mass and moment of inertia of the service satellite, design a fault diagnosis observer and a fault estimator. Input the fusion result into the fault diagnosis observer to generate a residual. Input the residual into the fault estimator to obtain the fault estimate, and then proceed to step 6. Step 6: Use the threshold discrimination method to screen the faulty thrusters. If the residual meets the discrimination condition, go to step 7; otherwise, complete the diagnosis. Step 7: Use the fault estimation quantity to estimate the mass and moment of inertia of the service star, implement the mass and moment of inertia correction of the service star, and return to step 2.

2. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 1 is characterized in that: In step 1, in the on-orbit service scenario shown in Figure 2, the relative pose equations of the service satellite and the target satellite are constructed according to the CW equation, as follows: in: Where: ρ(t) is the position of the service satellite relative to the target satellite, ρ(t) = r c (t)-r t (t), r c (t) is the position vector of the service satellite in the inertial system, r t (t) is the position vector of the target star in the inertial system, is the first-order derivative of ρ(t), A1 and A2 are coefficient matrices, ω is the angular velocity of the service satellite in the inertial system; u F (t) is the acceleration generated by the service satellite, u F (t)=[f x (t),f y (t),f z (t)] T , f x (t), f y (t), f z (t) are the accelerations along the coordinate axes in the service star system; t represents time, and T represents transposition.

3. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 2 is characterized in that: In step 1, a propulsion system fault model is established based on the working principle and structure of the satellite propulsion system, as follows: Where: F r is the actual control quantity generated by the thruster in the propulsion system; e(t) is the actual efficiency coefficient; F expect is the expected control quantity of the service star under no fault conditions, and const is a constant. That is, when the thruster has an efficiency degradation fault that causes the valve to fail to reach the maximum opening stroke, the actual thrust generated is determined by the actual efficiency coefficient e(t). When the thruster has a stuck fault that causes the valve to be unable to close, the actual control quantity generated is a constant value.

4. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 3 is characterized in that: In step 2, based on the relative posture equation and the propulsion system fault model, a service satellite quality-fault-control input coupling model considering mass consumption is established. After linearization, the service satellite state observation equation is established. Based on the coupling model, a Kalman filter based on the fault model is designed as follows: Step 2.1: Based on the relative posture equation and propulsion system fault model, establish a service satellite quality-fault-control input coupling model considering mass consumption: x(t+1)=Fx(t)+BE(t)u(t)+Γ x (t)w(t), Among them: x(t+1) is the system state at time t+1, x(t) is the system state at time t, F is the state transfer matrix, B is the control matrix, E(t) is the mass change effect matrix, is the angular velocity of the service star relative to the target star in its own system; Γ x (t) is the process noise influence matrix; w(t) is the process noise; u(t) is the control quantity, u(t)=[u1(t),u2(t),…,u i (t),…,u n (t)] T ;u i (t) is the control variable of the i-th thruster, i = 1, 2, ..., n, n is the total number of thrusters; Step 2.2: Establish the service satellite state observation equation based on the linearization of the coupling model: y(t+1)=Ηx(t+1)+Γ y v(t+1), Where: y(t+1) is the observation information; Η is the observation matrix; Γ y is the observation noise influence matrix; v(t+1) is the observation noise; Step 2.3: Design a Kalman filter based on the fault model based on the coupling model. Its structure is as follows: in: represents the predicted value of the service star status at time t+1, It represents the predicted value of the service star status at time t; is the prediction covariance matrix at time t+1; P(t+1) is the covariance matrix at time t+1, P(t) is the covariance matrix at time t; Q(t) is the process noise covariance matrix; K(t+1) is the Kalman filter gain; R(t) is the observation noise covariance matrix; ε is the intermediate function, I is the unit matrix, and x(t+1) is the filtering result.

5. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 4 is characterized in that: In step 2.1, the details are as follows: The mass change effect matrix E(t) is as follows: Among them, the mass change effect coefficient of the i-th thruster is The effect coefficient of the moment of inertia change of the i-th thruster e i (t) is the actual efficiency coefficient of the i-th propeller, To estimate the quality of the service star; is the estimated moment of inertia of the service satellite; k is the mass of propellant ejected per unit time; f is the propellant ejection velocity; The observation noise w(t-1) and system noise v(t) are both column vectors, and they satisfy: w(j1) and v(j1) represent the observation noise and system noise at time j1, respectively; w(j2) and v(j2) represent the observation noise and system noise at time j2, respectively; represents the Kroneck function, and v(j)~N(0,Σ v ), w(j)~N(0,Σ w ),Σ v is the covariance matrix of the system noise v(t), Σ w is the covariance matrix of the observation noise w(t).

6. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 5 is characterized in that: In step 4, the filtering results are processed to obtain the fusion results, as follows: y best (t+1)=W1x1(t+1)+W2x2(t+1)+…W p x p (t+1); Where: y best (t+1) is the fusion result; x l (t+1) is the optimal pose observation data of filter l, l=1,2,…,p, p is the total number of filters, the number of filters is equal to the number of sensors, that is, each sensor runs a filter independently, x l (t+1) is x(t+1) output in step 2.2; W m is the fusion weight coefficient matrix of the mth filter, m=1,2,…,p, and satisfies: [W1,W2,…,W l ,…,W p ] T =Σ1 -1 e(e T Σ1 -1 e) -1 ; Auxiliary matrix Σ1=diag(P1,P2,…,P m ,…,P p ), P m Represents the posterior covariance matrix of the filter of the mth sensor, the auxiliary matrix e=[I1,I2,…,I m ,…,I p ] T , I m is the identity matrix.

7. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 6, characterized in that: In step 5, the fault diagnosis observer and fault estimator are designed as follows: Step 5.1: Use the fault diagnosis observer for the stuck fault or efficiency drop fault as follows: Stuck type: r i (t)=Mz i (t)+Hy best (t), Efficiency reduction category: r i (t)=Mz i (t)+Hy best (t), Where: r i (t) is the residual error of the i-th thruster; y best (t) is the fusion result obtained in step 4, B is the control matrix, E(t-1) is the mass change effect matrix, and H is the observation matrix; i (t) is the state vector of the fault diagnosis observer of the i-th thruster, which should be an observable that can reflect the working condition of the i-th thruster; if the i-th thruster completely and only affects ρ(t) when it is working, then z(t) takes ρ(t) instead of θ(t); u(t-1) represents the control variable of the i-th thruster; represents the estimated value of the stuck fault of the i-th thruster, which is estimated in step 5.2; Indicates that the i e The efficiency of the propellers decreased due to failure, Pick The value of For the i e The actual efficiency coefficient of the propeller is For the i e The actual efficiency coefficient estimate of the propeller is as follows: e There is e i (t) = 1, and calculate the mass change effect matrix based on this, and the auxiliary matrix F is to be determined z , G, L, M, C must meet the following conditions: F z G+LC-GF=0, MG+HC=0, Re(λ(F z ))<0, MG(BE) i ≠0,i=1,2,…,n, Among them, Re() means taking the real part of a variable, and (BE) means taking the eigenvalue of the matrix; i represents the i-th column of the matrix BE; λ() represents taking the eigenvalue of the variable; Step 5.2: Design the Fault Estimator: The fault estimators for stuck fault and efficiency degradation fault are: Stuck type: Efficiency reduction category: Where: is the estimated value of the stuck fault of the i-th thruster, which represents the actual thrust value estimated when the stuck fault occurs. for The first derivative of ; is the estimated value of the actual efficiency coefficient of the i-th thruster, which represents the estimated value of the ratio of the actual thrust to the expected thrust when the efficiency degradation failure occurs, for The first derivative of ; and Collectively referred to as fault estimator; i is the auxiliary vector of the i-th thruster, and when M columns are full rank, ζ i =M(M T M) -1 P p G(BE) i , otherwise we can get ζ by estimating i The value of η i is the undetermined auxiliary vector of the i-th thruster, satisfying M T η i =PG(BE) i ρ i is the adaptive rate coefficient of the i-th thruster, the auxiliary matrix P p is a symmetric positive definite matrix, which should satisfy: Π is an arbitrary positive definite matrix.

8. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 7, characterized in that: In step 6, the threshold determination method is as follows Step 6: Further process the residual using the following formula to obtain the fault separation result: Where: S i (t) is the diagnostic index of the i-th thruster at time t, For the i e The diagnostic index of the thruster at time t is greater than the threshold; N s is the number of samples in the sliding window, S th is the threshold; the diagnostic parameter of the i-th thruster σ i When there is no fault i Standard deviation of (t); Calculate the diagnostic index of each thruster separately and compare the diagnostic index of all thrusters in the manner shown above; if there is only one Satisfy the above formula, that is, judge the i-th e A thruster malfunctioned.

9. The iterative satellite propulsion system fault diagnosis method based on the Lumberg observer according to claim 8, characterized in that: In step 7, the mass and moment of inertia of the service star are estimated using the fault estimation quantity, and the mass and moment of inertia of the service star are corrected as follows: When an efficiency drop fault occurs: When a stuck fault occurs: in: is the estimated service quality at time t+1; is the moment of inertia of the service star at time t+1, m c (0) is the initial mass of the service star; t s is the time when the failure occurs, ||d||2 is the distance from the propellant tank to the center of mass of the service satellite, For the i e The estimated failure rate of each thruster, is the i-th e The control quantity of the faulty thruster, u i (t1) is the control quantity of the i-th thruster at time t1.

Images