An adaptive fault-tolerant control method for actuator failures of a quadrotor aircraft

Abstract

The present invention discloses an adaptive fault-tolerant control method for actuator failures of a quadrotor aircraft. In the present invention, by defining intermediate control input variables, unknown actuator parameters are extracted from the original control input to form an unknown input gain matrix. The unknown input gain matrix is ​​decomposed using a gain matrix decomposition method to obtain linear parameterized uncertainty terms containing unknown system body parameters and actuator parameters. A filter-based immersion and invariant adaptive method is used to design an adaptive fault-tolerant compensation controller for the uncertain system physical parameters and actuator partial failure coefficients. This method ensures the robustness of the body under parameter perturbations and actuator partial failures, while also improving the transient response characteristics of the system, thereby enhancing the flight safety of the UAV.

Description

Technical Field

[0001] The present invention relates to an actuator fault control method for an aircraft in the field of aircraft control technology design, and in particular to an actuator fault adaptive fault-tolerant control method for a quadrotor aircraft. Background Art

[0002] Due to their low development and maintenance costs, high hovering accuracy, and strong maneuverability, quadrotors have garnered widespread attention and application in military and civilian fields, including 3D mapping, military surveillance, emergency rescue, traffic monitoring, and agricultural surveys. However, due to negative factors such as measurement errors and device losses, the physical parameters of quadrotors are likely to be uncertain, posing a challenge to the design of their controllers. In particular, when quadrotors perform missions in complex and harsh environments, their actuators are likely to be disturbed by external forces, causing damage to components such as propellers and motors, seriously affecting their flight quality. Therefore, to further improve the flight quality of quadrotors and expand their potential applications, it is of great significance to design a high-performance fault-tolerant control algorithm for their actuators.

[0003] To address the problem of partial actuator failure in quadrotor aircraft, researchers have applied a range of control algorithms to the system, including gain-scheduled PID control, sliding mode control, backstepping, adaptive control, and observer-based control methods. Among them, Lyapunov-based adaptive control techniques have been widely used to design nonlinear adaptive fault estimation laws to compensate for unknown actuator failure coefficients. This fault-tolerant control method is characterized by the absence of fault diagnosis and isolation. Upon the occurrence of a fault, the adaptive estimation law responds immediately to compensate for the uncertainty. Although Lyapunov adaptive control can achieve asymptotic tracking of the system to achieve good steady-state performance under unknown system parameters, its transient performance is difficult to guarantee. When a sudden actuator failure occurs in a quadrotor system, the relevant physical parameters undergo instantaneous changes. Consequently, the dynamic response of the system when recovering to steady-state using the Lyapunov adaptive method is likely to be suboptimal. Therefore, based on an analysis of existing results, it is crucial to design an adaptive fault-tolerant controller for quadrotors with uncertain actuator parameters that ensures good transient response. Summary of the Invention

[0004] In order to solve the problems in the background technology, the present invention provides an adaptive fault-tolerant control method for actuator faults of a quadrotor aircraft to eliminate the impact of uncertain and faulty actuators on the quadrotor aircraft, thereby achieving stable control of the quadrotor drone.

[0005] The technical solution adopted in the present invention is as follows:

[0006] Step 1: Establish a quadrotor height and attitude model with uncertain actuator parameters or unknown actuator failure factors;

[0007] Step 2: Extract the unknown actuator parameters from the original control input of the actuator and use them as the unknown input gain matrix. The original control input after the unknown actuator parameters are extracted is used as the control input to be designed.

[0008] Step 3: Determine the initial tracking error variable based on the measured displacement, linear velocity, angle, and angular velocity data of the quadrotor;

[0009] Step 4: Based on the quadrotor altitude and attitude model, the initial tracking error variables, and the control input to be designed, the unknown input gain matrix is ​​decomposed using the gain matrix decomposition method to obtain linear parameterized uncertainties. The linear parameterized uncertainties consist of the initial regression vector and the unknown parameters.

[0010] Step 5: Perform low-pass filtering on the initial tracking error variable and the initial regression vector to obtain filtered tracking error variables and filtered regression vectors respectively;

[0011] Step 6: Construct a filter-based immersion and invariant adaptive controller based on the initial tracking error variable and the filtered tracking error variable, the initial regression vector, and the filtered regression vector to achieve adaptive control of the quadrotor aircraft.

[0012] In step 1, the formulas for the height and attitude model of a quadrotor with uncertain actuator parameters or unknown actuator failure factors are as follows:

[0013]

[0014]

[0015] in, Represents the first flight parameter x of the rotorcraft i1 The differential of x 11 =φ、x 21 =θ, x 31 =ψ and x 41 = z, φ, θ, ψ and z are roll angle, pitch angle, yaw angle and height respectively, x i2 represents the second flight parameter of the rotorcraft, wherein and and are the roll angle, pitch angle, yaw angle and linear velocity in the height direction, i=1,...,4 represents the roll, pitch, yaw and height subsystems respectively, f iThe function and g of the corresponding subsystem including gyroscopic effect and gravity effect are represented by i represents the control gain function, and the formula is as follows:

[0016]

[0017] Among them, J x , J y , J z , J r are the first to third moments of inertia and propeller moment of inertia, Ω R is the rotor speed difference, m is the machine mass, g is the acceleration of gravity, τ1, τ2, τ3 are the first to third control torques.

[0018] In step 2, the original control input U, the unknown input gain matrix G, and the control input v to be designed satisfy the following relationship:

[0019] U=Gv

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028] Among them, θ 11 ,θ 22 ,θ 33 ,θ 44 is the control gain of each subsystem including unknown actuator parameters of roll, pitch, yaw and altitude, θ 12 ,θ 13 ,θ 14 is the unknown actuator parameter in the rolling rotor system affected by the pitch, yaw, and altitude subsystems, θ 21 ,θ 23 ,θ 24 is the unknown actuator parameter in the pitch subsystem affected by the roll, yaw, and altitude subsystems, θ 31 ,θ 32 ,θ 34is the unknown actuator parameter in the yaw subsystem affected by the roll, pitch, and altitude subsystems, θ 41 ,θ 42 ,θ 43 is the unknown actuator parameter in the altitude subsystem affected by the roll, pitch, and yaw subsystems, to Represents the element containing unknown actuator parameters; i, j = 1, ..., 4, T represents transpose, c ti and c di are the lift and anti-torque coefficients, K mi is the voltage-speed gain coefficient, i=1,...,4 represent the roll, pitch, yaw and altitude subsystems respectively.

[0029] In step 3, the initial tracking error variable is the first tracking error variable z i1 and the second tracking error variable z i2 Composition, i=1,…,4, the formula is as follows:

[0030] z i1 =x i1 -x id

[0031] z i2 =x i2 -α i

[0032]

[0033] Among them, x i1 Represents the first flight parameter of the rotorcraft, x i2 Represents the second flight parameter of the rotorcraft, x id is the preset reference signal, α i represents the virtual control signal, k i1 represents the first design parameter, Indicates the differential of the preset reference signal.

[0034] The step 4 is specifically as follows:

[0035] First, the unknown input gain matrix G is decomposed into G = SDQ by applying the gain matrix decomposition method, where S is a positive definite symmetric matrix, D is a diagonal matrix to be designed, satisfying D = D1D2, D1 is an arbitrary positive definite diagonal matrix, and D2 is a matrix containing the symbols of the principal minors of each order in the unknown input gain matrix G, satisfying D2 = diag{sgn(Δ1),sgn(Δ2 / Δ1),…,sgn(Δ q / Δ q-1 )}, diag{·} represents element diagonalization, sgn(·) represents the sign function, Δ1,…,Δ qis the non-zero ordered principal minor of matrix G, Q is an upper triangular matrix;

[0036] Solve the following formula based on the quadrotor height and attitude model, the initial tracking error variables, the control input to be designed, the positive symmetric matrix S, the diagonal matrix D, and the upper triangular matrix Q to obtain the initial regression vector μ and the unknown parameters Θ:

[0037]

[0038]

[0039]

[0040] z1=[z 11 ,z 21 ,z 31 ,z 41 ] T

[0041] z2=[z 12 ,z 22 ,z 32 ,z 42 ] T

[0042]

[0043] f=[f1,f2,f3,f4] T ,

[0044]

[0045] v=[v1,v2,v3,v4] T

[0046] k1=diag{k 11 ,k 21 ,k 31 ,k 41}

[0047] k2=diag{k 12 ,k 22 ,k 32 ,k 42}

[0048] in, represents the differential of the first tracking error vector z1, represents the differential of the first tracking error vector z2, k1 represents the first design parameter diagonal matrix, k2 represents the second design parameter diagonal matrix, represents the control gain matrix, f represents the vector including gyroscopic effect and gravity effect, represents the virtual control signal differential vector of the corresponding subsystem, Represents the virtual control signal α of the corresponding subsystem i The differential of z i1 represents the first tracking error variable of the corresponding subsystem, z i2 represents the second tracking error variable of the corresponding subsystem, T represents the transpose, and f i represents the function of the corresponding subsystem including gyroscopic effect and gravity effect, g i represents the control gain function of the corresponding subsystem, v i represents the control input to be designed for the corresponding subsystem, v represents the control input vector to be designed, k i2 is the second design parameter of the corresponding subsystem, i=1,…,4.

[0049] In step 6, the formula of the filter-based immersion and invariant adaptive controller is as follows:

[0050]

[0051]

[0052]

[0053] in, is the integral part of the adaptive estimation of the unknown parameter Θ, is the integral part of the adaptive estimation of the unknown parameter Θ The differential of , β is the proportional part of the adaptive estimation of the unknown parameter Θ, is the differential of the proportional part β of the adaptive estimate of the unknown parameter Θ, is the regression vector μ after filtering f The differential of is the estimated value of Θ, γ is the positive adaptive gain, z 1,f represents the first tracking error variable after filtering, z 2,f represents the second tracking error variable after filtering, v represents the control input vector to be designed, and D is the diagonal matrix to be designed.

[0054] The beneficial effects of the present invention are:

[0055] This paper designs a unified adaptive control method for quadrotor aircraft with uncertain or faulty actuators. This method not only compensates for the steady-state errors caused by slight actuator uncertainty in quadrotor systems, but also effectively and quickly compensates for sudden partial efficiency losses in actuator failures. Furthermore, the designed filtering-based immersion and invariant adaptive controller achieves superior transient response results compared to traditional adaptive methods. By introducing filtering, the method avoids the difficulty in calculating partial differential equations in standard immersion and invariant adaptive methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a control block diagram of the present invention.

[0057] Figure 2 Schematic diagram of the quadrotor aircraft model used in the present invention.

[0058] Figure 3 It is a simulation diagram of the height tracking curve in an embodiment of the present invention.

[0059] Figure 4 4 is a simulation diagram of a roll angle tracking curve in an embodiment of the present invention.

[0060] Figure 5 It is a simulation diagram of the pitch angle tracking curve in an embodiment of the present invention.

[0061] Figure 6 It is a simulation diagram of the yaw angle tracking curve in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The following is a further description with reference to the accompanying drawings and embodiments.

[0063] A flowchart of a method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to an embodiment of the present invention is shown in FIG. Figure 1 As shown, it is used to solve the problem of actuator uncertainty or efficiency loss failure in the model. The schematic diagram of the quadrotor model is shown in Figure 2 As shown, the method includes the following steps:

[0064] Step 1: Establish a quadrotor height and attitude model based on the quadrotor with uncertain actuator parameters or unknown actuator failure factors;

[0065] In step 1, the equations for the quadrotor altitude and attitude models with uncertain actuator parameters or unknown actuator failure factors are as follows:

[0066]

[0067]

[0068] in, represents the first flight parameter x of the rotorcraft i1 The differential of x 11 =φ、x 21 =θ、x 31 =ψ and x 41 = z, φ, θ, ψ and z are roll angle, pitch angle, yaw angle and height respectively, x i2 represents the second flight parameter of the rotorcraft, wherein and and are the roll angle, pitch angle, yaw angle and linear velocity in the height direction, i=1,...,4 represents the roll, pitch, yaw and height subsystems respectively, f i The function and g of the corresponding subsystem including gyroscopic effect and gravity effect are represented by i represents the control gain function, and the formula is as follows:

[0069]

[0070]

[0071] Among them, J x , J y , J z , J r are the first to third moments of inertia and propeller moment of inertia, Ω R is the rotor speed difference, m is the body mass, g is the acceleration of gravity, τ1, τ2, τ3 are the first to third control torques, and the original control input U i It is given by:

[0072]

[0073]

[0074]

[0075]

[0076] Among them, c ti and c di are the lift and anti-torque coefficients of the corresponding subsystem, l is the quadrotor arm length, ω i is the rotor speed of the corresponding subsystem, i=1,...,4 represents the roll, pitch, yaw and altitude subsystems respectively. The relationship between voltage input and speed is:

[0077] ω i =K mi u i ,i=1,…,4

[0078] Among them, K mi is the voltage-speed gain coefficient.

[0079] In this embodiment, the system parameters are selected as m=1kg, J x =J y =0.05kg·m 2 , J z =0.1kg·m 2 , J r =0.0001kg·m 2 , g=9.81m / s 2 , c t =5×10 -4 N·s 2 / rad 2 , c d =1×10 -5 Nm·s 2 / rad 2 , l=0.25m, K m1 =100,K m2 =97,K m3 =102,K m4 =95.

[0080] Step 2: Extract unknown actuator parameters from the original control input of the actuator and use them as the unknown input gain matrix. The original control input after extracting the unknown actuator parameters is used as the control input to be designed. In the specific implementation, the unknown actuator parameters include the propeller lift coefficient and the voltage-speed gain coefficient.

[0081] In step 2, the original control input U, the unknown input gain matrix G, and the control input v to be designed satisfy the following relationship:

[0082] U=Gv

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] v=[v1 v2 v3 v4] T

[0092]

[0093]

[0094]

[0095]

[0096] Among them, θ 11 ,θ 22 ,θ 33 ,θ 44 is the control gain of each subsystem including unknown actuator parameters of roll, pitch, yaw and altitude, θ 12 ,θ 13 ,θ 14 is the unknown actuator parameter in the rolling rotor system affected by the pitch, yaw, and altitude subsystems, θ 21 ,θ 23 ,θ 24 is the unknown actuator parameter in the pitch subsystem affected by the roll, yaw, and altitude subsystems, θ 31 ,θ 32 ,θ 34 is the unknown actuator parameter in the yaw subsystem affected by the roll, pitch, and altitude subsystems, θ 41 ,θ 42 ,θ 43 is the unknown actuator parameter in the altitude subsystem affected by the roll, pitch, and yaw subsystems, to Represents the element containing unknown actuator parameters; i, j = 1, ..., 4, T represents transpose, c ti and c di are the lift and anti-torque coefficients, K mi is the voltage-speed gain coefficient, i=1,...,4 represents the roll, pitch, yaw and altitude subsystems respectively. i represents the control input to be designed for the corresponding subsystem, u i Represents the voltage input signal of the corresponding subsystem.

[0097] Step 3: Determine the initial tracking error variable z of the corresponding subsystem based on the measured displacement, linear velocity, angle, and angular velocity data of the quadrotor aircraft i1 ,z i2 , where z i1 ,z i2 ,i=1,…,4 represent the roll, pitch, yaw and altitude subsystems respectively;

[0098] In step 3, the displacement and linear velocity of the quadrotor are measured by GPS and optical flow sensor, and the angle and angular velocity data are measured by gyroscope, accelerometer and magnetometer. The initial tracking error variable of the corresponding subsystem is the first tracking error variable z i1 and the second tracking error variable z i2 Composition, i=1,…,4, the formula is as follows:

[0099] z i1 =x i1 -x id

[0100] z i2 =x i2 -α i

[0101]

[0102] Among them, x i1 represents the first flight parameter of the corresponding subsystem of the rotorcraft, x i2 represents the second flight parameter of the corresponding subsystem of the rotorcraft, x id is the preset reference signal of the corresponding subsystem, α i represents the virtual control signal of the corresponding subsystem, k i1 represents the first design parameter of the corresponding subsystem, Indicates the differential of the preset reference signal of the corresponding subsystem.

[0103] In this embodiment, the desired attitude angle and height are step signals, specifically x 1d =x 2d =x 3d = 0.1 and x 4d =1. In this embodiment, the first design parameter is selected as k i1 =2.

[0104] Step 4: According to the quadrotor height and attitude model, the initial tracking error variable z of the corresponding subsystem i1 ,z i2 , i=1,…,4 and the control input to be designed, the unknown input gain matrix is ​​decomposed by the gain matrix decomposition method to obtain the linear parameterized uncertainty term; the linear parameterized uncertainty term consists of the initial regression vector and the unknown parameters;

[0105] Step 4 is as follows:

[0106] First, the unknown input gain matrix G is decomposed into G = SDQ by applying the gain matrix decomposition method, where S is a positive definite symmetric matrix, D is a diagonal matrix to be designed, satisfying D = D1D2, D1 is an arbitrary positive definite diagonal matrix, and D2 is a matrix containing the symbols of the principal minors of each order in the unknown input gain matrix G, satisfying D2 = diag{sgn(Δ1),sgn(Δ2 / Δ1),…,sgn(Δ q / Δ q-1 )}, diag{·} represents element diagonalization, sgn(·) represents the sign function, Δ1,…,Δ q is the non-zero ordered principal minor of matrix G, Q is an upper triangular matrix;

[0107] According to the quadrotor height and attitude model, the initial tracking error variable z i1 ,z i2 , i=1,…,4 and the control input to be designed as well as the positive definite symmetric matrix S, the diagonal matrix D and the upper triangular matrix Q to solve the following formula to obtain the initial regression vector μ and the unknown parameters Θ, where μ is known and Θ is unknown:

[0108]

[0109]

[0110]

[0111] z1=[z 11 ,z 21 ,z 31 ,z 41 ] T

[0112] z2=[z 12 ,z 22 ,z 32 ,z 42 ] T

[0113]

[0114] f=[f1,f2,f3,f4] T ,

[0115]

[0116] v=[v1,v2,v3,v4] T

[0117] k1=diag{k 11 ,k 21 ,k 31 ,k 41}

[0118] k2=diag{k 12 ,k 22 ,k 32 ,k 42}

[0119] in, represents the differential of the first tracking error vector z1, represents the differential of the first tracking error vector z2, k1 represents the first design parameter diagonal matrix, k2 represents the second design parameter diagonal matrix, represents the control gain matrix, f represents the vector including gyroscopic effect and gravity effect, represents the virtual control signal differential vector of the corresponding subsystem, Represents the virtual control signal α of the corresponding subsystem i The differential of k i2 is the second design parameter of the corresponding subsystem, i=1,…,4. In this embodiment, the second design parameter k i2 =10;

[0120] Step 5: Initial tracking error variable z for the corresponding subsystem i1 ,z i2 , i=1,…,4 and the initial regression vector μ, the initial regression vector μ is a known function value, and low-pass filtering is performed respectively to obtain the filtered tracking error variables z of the corresponding subsystems. i1,f ,z i2,f , i=1,…,4 and the filtered regression vector μ f ;

[0121] In step 5, use a low-pass filter For variable z i1 ,z i2 ,1,…,4 and μ for filtering, the formula is as follows:

[0122]

[0123] Step 6: According to the initial tracking error variable z of the corresponding subsystem i1 ,z i2 , i=1,…,4 and the filtered tracking error variable z i1,f ,z i2,f ,i=1,…,4, initial regression vector μ, filtered regression vector μ f A filtering-based immersed and invariant adaptive controller is constructed, and the filtering-based immersed and invariant adaptive method is used to design an adaptive controller for a quadrotor aircraft to offset the influence of unknown parameterization uncertainty on the system and realize adaptive control of the quadrotor aircraft.

[0124] In step 6, the formula of the filter-based immersed and invariant adaptive controller is as follows:

[0125]

[0126]

[0127]

[0128] in, is the integral part of the adaptive estimation of the unknown parameter Θ, is the integral part of the adaptive estimation of the unknown parameter Θ The differential of , β is the proportional part of the adaptive estimation of the unknown parameter Θ, is the differential of the proportional part β of the adaptive estimate of the unknown parameter Θ, is the regression vector μ after filtering f The differential of is the estimated value of Θ, γ is the positive adaptive gain, z 1,f represents the first tracking error variable after filtering, z 2,f represents the second tracking error variable after filtering.

[0129] In this embodiment, the design parameter value γ is selected as γ=1.

[0130] Defining parameter estimation error Taking the derivative of it, we can get

[0131]

[0132] To verify the stability of the closed-loop system formed by the quadcopter and the controller, the following Lyapunov function is designed.

[0133]

[0134] Wherein, V represents the Lyapunov function of the entire closed-loop system, and ε>0 is a positive constant.

[0135] Its derivative is

[0136]

[0137] in and λ gS are k2 and The minimum characteristic root of as well as Then the filtered tracking error z i1,f ,z i2,f It will eventually converge to zero, so the tracking error z i1 ,z i2will also converge to zero.

[0138] In order to further verify the significant substantive features of the present invention, a simulation experiment was carried out on the present invention.

[0139] like Figures 3 to 6 The tracking curves for altitude, roll angle, pitch angle, and yaw angle are shown. The simulation results show that even with an actuator failure occurring at the sixth second, the proposed design can still accurately track the desired signal, and its transient response is superior to traditional Lyapunov-based adaptive and PID control methods.

Claims

1. A method for adaptive fault-tolerant control of actuator faults of a quadrotor aircraft, characterized in that: The following steps are involved: Step 1: Establish a quadrotor height and attitude model with uncertain actuator parameters or unknown actuator failure factors; Step 2: Extract the unknown actuator parameters from the original control input of the actuator and use them as the unknown input gain matrix. The original control input after the unknown actuator parameters are extracted is used as the control input to be designed. Step 3: Determine the initial tracking error variable based on the measured displacement, linear velocity, angle, and angular velocity data of the quadrotor; Step 4: Based on the quadrotor altitude and attitude model, the initial tracking error variables, and the control input to be designed, the unknown input gain matrix is ​​decomposed using the gain matrix decomposition method to obtain linear parameterized uncertainties. The linear parameterized uncertainties consist of the initial regression vector and the unknown parameters. Step 5: Perform low-pass filtering on the initial tracking error variable and the initial regression vector to obtain filtered tracking error variables and filtered regression vectors respectively; Step 6: Construct a filter-based immersion and invariant adaptive controller based on the initial tracking error variable and the filtered tracking error variable, the initial regression vector, and the filtered regression vector to achieve adaptive control of the quadrotor aircraft.

2. The method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to claim 1, characterized in that: In step 1, the formulas for the height and attitude model of a quadrotor with uncertain actuator parameters or unknown actuator failure factors are as follows: in, Represents the first flight parameter x of the rotorcraft i1 The differential of x 11 =φ、x 21 =θ, x 31 =ψ and x 41 = z, φ, θ, ψ and z are roll angle, pitch angle, yaw angle and height respectively, x i2 represents the second flight parameter of the rotorcraft, wherein and and are the roll angle, pitch angle, yaw angle and linear velocity in the height direction, i=1,...,4 represents the roll, pitch, yaw and height subsystems respectively, f i The function and g of the corresponding subsystem including gyroscopic effect and gravity effect are represented by i represents the control gain function, and the formula is as follows: f4=-g, Among them, J x , J y , J z , J r are the first to third moments of inertia and propeller moment of inertia, Ω R is the rotor speed difference, m is the machine mass, g is the acceleration of gravity, τ1, τ2, τ3 are the first to third control torques.

3. The method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to claim 1, characterized in that: In step 2, the original control input U, the unknown input gain matrix G, and the control input v to be designed satisfy the following relationship: U=Gv in, are the control gains of the roll, pitch, yaw, and altitude subsystems containing unknown actuator parameters, is the unknown actuator parameter in the rolling rotor system affected by the pitch, yaw, and altitude subsystems, is the unknown actuator parameter in the pitch subsystem affected by the roll, yaw, and altitude subsystems, is the unknown actuator parameter in the yaw subsystem affected by the roll, pitch, and altitude subsystems, is the unknown actuator parameter in the altitude subsystem affected by the roll, pitch, and yaw subsystems, to Represents the element containing unknown actuator parameters; i, j = 1,,4, T represents transpose, c ti and c di are the lift and anti-torque coefficients, K mi is the voltage-speed gain coefficient, i=1,...,4 represent the roll, pitch, yaw and altitude subsystems respectively.

4. The method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to claim 1, characterized in that: In step 3, the initial tracking error variable is the first tracking error variable z i1 and the second tracking error variable z i2 Composition, i=1,…,4, the formula is as follows: z i1 =x i1 -x id z i2 =x i2 -α i Among them, x i1 Represents the first flight parameter of the rotorcraft, x i2 Represents the second flight parameter of the rotorcraft, x id is the preset reference signal, α i represents the virtual control signal, k i1 represents the first design parameter, Indicates the differential of the preset reference signal.

5. The method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to claim 1, characterized in that: The step 4 is specifically as follows: First, the unknown input gain matrix G is decomposed into G = SDQ by applying the gain matrix decomposition method, where S is a positive definite symmetric matrix, D is a diagonal matrix to be designed, satisfying D = D1D2, D1 is an arbitrary positive definite diagonal matrix, and D2 is a matrix containing the symbols of the principal minors of each order in the unknown input gain matrix G, satisfying D2 = diag{sgn(Δ1),sgn(Δ2 / Δ1),…,sgn(Δ q / Δ q-1 )}, diag{·} represents element diagonalization, sgn(·) represents the sign function, Δ1,…,Δ q is the non-zero ordered principal minor of matrix G, Q is an upper triangular matrix; According to the quadrotor height and attitude model, the initial tracking error variables and the control input to be designed, as well as the positive definite symmetric matrix S, the diagonal matrix D and the upper triangular matrix Q, the initial regression vector μ and the unknown parameters Θ are obtained: z1=[z 11 ,With 21 ,With 31 ,With 41 ] T z2=[z 12 ,With 22 ,With 32 ,With 42 ] T <h2 style=";text-align:left;direction:ltr">f = [f1,f2,f3,f4]<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> , v=[v1,v2,v3,v4] T k1=diag{k 11 ,k 21 ,k 31 ,k 41 } k2=diag{k 12 ,k 22 ,k 32 ,k 42 } in, represents the differential of the first tracking error vector z1, represents the differential of the first tracking error vector z2, k1 represents the first design parameter diagonal matrix, k2 represents the second design parameter diagonal matrix, represents the control gain matrix, f represents the vector including gyroscopic effect and gravity effect, represents the virtual control signal differential vector of the corresponding subsystem, Represents the virtual control signal α of the corresponding subsystem i The differential of z i1 represents the first tracking error variable of the corresponding subsystem, z i2 represents the second tracking error variable of the corresponding subsystem, T represents the transpose, and f i represents the function of the corresponding subsystem including gyroscopic effect and gravity effect, g i represents the control gain function of the corresponding subsystem, v i represents the control input to be designed for the corresponding subsystem, v represents the control input vector to be designed, k i2 is the second design parameter of the corresponding subsystem, i=1,…,4.

6. The method for adaptive fault-tolerant control of actuator failures of a quadrotor aircraft according to claim 1, characterized in that: In step 6, the formula of the filter-based immersion and invariant adaptive controller is as follows: in, is the integral part of the adaptive estimation of the unknown parameter Θ, is the integral part of the adaptive estimation of the unknown parameter Θ The differential of , β is the proportional part of the adaptive estimation of the unknown parameter Θ, is the differential of the proportional part β of the adaptive estimate of the unknown parameter Θ, is the regression vector μ after filtering f The differential of is the estimated value of Θ, γ is the positive adaptive gain, z 1,f represents the first tracking error variable after filtering, z 2,f represents the second tracking error variable after filtering, v represents the control input vector to be designed, and D is the diagonal matrix to be designed.

Images