An adaptive fault-tolerant control method for vertical take-off and landing aircraft

Abstract

The present invention provides an adaptive fault-tolerant control method for a vertical take-off and landing aircraft to compensate for actuator failures and model uncertainties. The control method is divided into two independent control modules: a high-level adaptive sliding mode control module and a low-level control allocation module. The low-level control allocation module is used to allocate virtual control signals generated by the high-level control module among redundant available actuators. The high-level control module is composed of an adaptive sliding mode controller, which is used to maintain the overall tracking performance of the system under failure and uncertainty conditions. In the case of actuator failures and model uncertainties, an adaptive scheme will be triggered to generate more virtual control signals. Using a comprehensive adaptive scheme, control parameters can be adaptively changed to compensate for virtual control errors. The method is effective and superior in both single actuator failure and concurrent actuator failure conditions.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicle control methods, and in particular relates to an adaptive fault-tolerant control method for a vertical take-off and landing aircraft. Background Art

[0002] In recent years, with the development of drone technology, more and more drones have been developed and used in various practical applications, such as payload transportation, aerial surveillance, and border monitoring. For many industrial applications, vertical take-off and landing is a basic requirement for drones. In addition, in order to complete the assigned tasks more efficiently, drones need to fly for a long time. In this context, different types of hybrid vertical take-off and landing drones have been developed, which combine the advantages of rotary-wing and fixed-wing drones to achieve a wider flight envelope.

[0003] As mission requirements become more complex, higher requirements are placed on the high reliability and high precision control of vertical take-off and landing UAV systems. The mission environment of UAVs is complex and changeable. During the autonomous flight of UAVs, there will always be some unpredictable and unavoidable situations, especially the situation where the actuator loses its effectiveness. This requires the emergence of active fault-tolerant control systems. People have conducted extensive research on the safe operation of vertical take-off and landing UAVs, but the research on adaptive fault-tolerant control of hybrid vertical take-off and landing UAVs with servo failures and model uncertainties in fixed-wing mode is far from enough.

[0004] Based on this, an adaptive fault-tolerant control method for vertical take-off and landing aircraft is proposed. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an adaptive fault-tolerant control method for a vertical take-off and landing aircraft in view of the deficiencies of the above-mentioned prior art, so as to simultaneously compensate for the adverse effects caused by actuator failures and model uncertainties without requiring any prior knowledge of failures and uncertainties. The proposed adaptive fault-tolerant control scheme can update control parameters in real time to compensate for virtual control errors, and redistribute control signals to available redundant actuators, thereby theoretically ensuring the stability of the closed-loop system to solve the problems raised in the above-mentioned background technology.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: an adaptive fault-tolerant control method for a vertical take-off and landing aircraft, comprising the following steps:

[0007] S1. Establish the dynamic model of vertical take-off and landing hybrid UAV:

[0008]

[0009] The origin of the fuselage fixed coordinate system coincides with the center of gravity of the hybrid UAV;

[0010] Where φ, θ, ψ are the roll, pitch, and yaw angles of the drone;

[0011] I xx ,I yy ,I zz is the moment of inertia of the drone;

[0012] I xz is the product of inertia;

[0013] M xb ,M yb ,M zb They are respectively applied to drone x b ,y b and z b The resultant moment on the shaft;

[0014] S2. Design an adaptive sliding mode control strategy that adapts to actuator faults and model uncertainty, and construct an integral sliding mode controller and control allocation scheme to ensure the tracking performance of the system under fault-free conditions;

[0015] Consider a full-chain nonlinear affine system with model uncertainty and actuator fault disturbances:

[0016]

[0017] in is the state vector;

[0018] is the control effectiveness matrix;

[0019] is the control input vector;

[0020] is a diagonal matrix;

[0021] represents an unknown perturbation whose bound is ||d(t)||≤D;

[0022] Vector F(x 1 (t),x 2 (t))∈R p is a nonlinear function that includes model uncertainty;

[0023] D f (t) = diag([d f1 (t),d f2 (t),...,d fm (t)]) is a diagonal matrix representing the effectiveness level of actuator control, where d fj (t)(j=1,2,...,m) satisfies 0≤d fj (t)≤1 if dfj (t) = 1, the jth actuator works normally, otherwise, the jth actuator will have a certain degree of failure;

[0024] S3. Based on the fact that the hybrid canard rotary wing (CRW) UAV under study is an overdriven system, when an actuator fails, it is necessary to obtain fault information and redistribute the control signal to the available actuators. Considering the application of the control scheme in the actual system, when an actuator fails, the redistribution of the control signal needs to be triggered immediately. In this case, an adaptive control strategy that does not require any fault information is proposed to adapt to the actuator failure of the overdriven system.

[0025] For the system considered above, the quadratic optimization algorithm is used to calculate the actual control input u so that the virtual control signal generated by the control allocation module can meet the requirements of the advanced sliding mode control to generate the virtual control signal;

[0026] The quadratic programming method based on minimizing the control input is described as:

[0027] J = arg minu T Qu

[0028] stv i =C ui u

[0029] The displayed solution is as follows:

[0030]

[0031] in is a symmetric positive definite weight matrix, corresponding to the residual control efficiency of the actuator.

[0032] Furthermore, in S1, due to I xz The absolute value is much smaller than I xx ,I yy ,I zz In order to facilitate the controller design, the absolute value of I xz The relevant components are considered as disturbances;

[0033] In practice, it is impossible to accurately obtain the torque acting on the drone, and the formula for the synthetic torque is as follows:

[0034]

[0035] Among them, M xknow ,M yknow and M zplane represents the known part of the resultant moment;

[0036] M xc ,M ycand M zc It represents the moment produced by the deflection of the control surface;

[0037] M xu ,M yu and M zu It is expressed as the unknown uncertain part of the resultant torque;

[0038] The control-oriented model looks like this:

[0039]

[0040] where δ ac is the aileron deflection of the canard control surface;

[0041] δ at is the aileron deflection of the horizontal tail control surface;

[0042] δ ec is the elevator deflection of the canard control surface;

[0043] δ et is the elevator deflection of the horizontal tail control surface;

[0044] δ e is the rudder deflection of the vertical tail control surface;

[0045] is a coefficient related to the torque generated.

[0046] Furthermore, in S2, in order to facilitate the design of the controller, the system state is defined as:

[0047]

[0048] The nonlinear affine system in S2 can be written as:

[0049]

[0050] Where i = 1, 2, 3 represents each subsystem;

[0051] x 2i-1 ∈[φ,θ,ψ] T ,

[0052] Will and Expressed as the desired trajectory, the tracking error is then defined as and In this sense, the integral sliding surface of the system is defined as:

[0053]

[0054] in is a nonlinear combination of tracking errors,

[0055] g i Including the integral term, t 0 is the initial moment, a i1 ,a i2 and c i is the design parameter;

[0056] In this case, the integral sliding surface can be expressed as:

[0057]

[0058] By making The continuous control part is obtained, and the disturbance d is not considered at this time i ;

[0059] At this time, the continuous control part can be designed as:

[0060]

[0061] Then, to compensate for the disturbance d i , the discontinuous control part is designed and integrated to ensure the ideal sliding motion, as shown in the following formula:

[0062] ζ i2 =-a i3 sign(s i )

[0063] In the formula, a i3 is a positive gain, which makes the sliding surface attractive to the sliding variable;

[0064] A further improvement of the smooth discontinuity is to add a thin boundary layer near the defined sliding surface. The boundary layer can be expressed as:

[0065]

[0066] Using the defined boundary layer, a saturation function is defined to replace ζ i2 =-a i3 sign(s i ), as shown in the following formula:

[0067]

[0068] The continuous control part and the discontinuous control part can be combined, and the developed control law is expressed as:

[0069]

[0070] Furthermore, in S3, when the actuator fails, the diagonal matrix Q will no longer be the unit matrix. Since the actual fault information cannot be obtained, if the matrix Q is still selected as the unit matrix to calculate the actual control input u, there will be a virtual control error. Under this condition, it can be obtained:

[0071] ζ i =C ui Eu-C ui (ED f )

[0072] Where E is the identity matrix;

[0073] Order ie =-C ui (ED f )u, the following formula

[0074]

[0075] can be rewritten as:

[0076]

[0077] where ζ id =C ui Eu represents the virtual control signal expected by the high-level controller.

[0078] It can be seen that in order to maintain the performance of the closed-loop system, the parameter h needs to be adaptively adjusted. i Eliminate the error of the virtual control signalζ ie ;

[0079] set up Then the following formula

[0080] J = arg minu T Qu

[0081] stv i =C ui u

[0082] Refactored to:

[0083]

[0084] Using estimates taking into account model uncertainty and actuator failures and To derive the corresponding high-level control law;

[0085] In this case, the formula

[0086]

[0087] Refactored to:

[0088]

[0089]

[0090] Then, by expressing and ζ i =C ui Eu-C ui (ED f )u can be rearranged as:

[0091]

[0092] in

[0093] Furthermore, in order to compensate for actuator failures and model uncertainties, the estimated parameters need to be updated online. and Use an online adaptive scheme to estimate uncertain parameters:

[0094]

[0095]

[0096] In the formula, Δs i =s i -Φ i sat(s i ) represents the distance between the sliding variable and the defined boundary layer. The adaptive scheme is triggered only when the sliding variable is outside the defined boundary layer, that is, when the system tracking performance is not ideal. By defining the variable, the overestimation of the uncertain parameters can be avoided, which helps to alleviate the system control chattering.

[0097] Compared with the prior art, the present invention has the following advantages:

[0098] 1. The control method in the present invention is divided into two independent control modules, a high-level adaptive sliding mode control module and a low-level control allocation module. The low-level control allocation module is used to allocate the virtual control signal generated by the high-level control module among the redundant available actuators. The high-level control module is composed of an adaptive sliding mode controller, which is used to maintain the overall tracking performance of the system under fault and uncertainty conditions. In the case of actuator failure and model uncertainty, an adaptive scheme will be triggered to generate more virtual control signals. Using a comprehensive adaptive scheme, the control parameters can be adaptively changed to compensate for the virtual control error. This method is effective and superior in both single actuator failure and concurrent actuator failure situations, and can simultaneously compensate for the adverse effects caused by actuator failure and model uncertainty without any prior knowledge of failure and uncertainty.

[0099] 2. Compared with the existing fault-tolerant control method based on sliding mode, the control method in the present invention can avoid overestimation of control parameters, thereby helping to alleviate unexpected control chattering.

[0100] 3. The adaptive fault-tolerant control scheme proposed in the present invention can update the control parameters in real time to compensate for the virtual control error and redistribute the control signal to the available redundant actuators, thus theoretically ensuring the stability of the closed-loop system. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 This is a flow chart of the adaptive fault-tolerant control method proposed by the present invention.

[0102] Figure 2 This is a speed tracking performance diagram of a UAV when there is an elevator failure and model uncertainty in the experimental example of the present invention.

[0103] Figure 3 This is the input diagram of the canard control surface when there is an elevator fault and model uncertainty in the experimental example of the present invention.

[0104] Figure 4 This is the input diagram of the elevator control surface when there is an elevator fault and model uncertainty in the experimental example of the present invention. DETAILED DESCRIPTION

[0105] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0106] In order to verify the effectiveness of the proposed adaptive fault-tolerant control (AFTC) scheme: In the simulation experiment, the overdriven hybrid canard rotary wing (CRW) UAV under study was compared with the non-adaptive conventional sliding mode control (CSMC) in fixed-wing mode. The effectiveness of the proposed AFTC scheme was verified by considering the model uncertainty and actuator failure of the hybrid CRW UAV. Taking the longitudinal direction as an example, the UAV was controlled to track a set of pre-designed sinusoidal signals, and the proposed AFTC tracking of the speed and the deflection angle curves of the elevator and canard were observed after disturbances occurred at different times, and compared with CSMC to verify the effectiveness of this technology.

[0107] Examples, such as Figure 1As shown, the present invention provides a technical solution: an adaptive fault-tolerant control method for a vertical take-off and landing aircraft, characterized in that it includes the following steps:

[0108] S1. Establish the dynamic model of vertical take-off and landing hybrid UAV:

[0109] In order to model the CRW UAV, a fuselage fixed coordinate system is used, and the origin of the fuselage fixed coordinate system coincides with the center of gravity of the CRW UAV. The dynamic equation of the CRW UAV can be obtained by using the Newton-Euler formula:

[0110]

[0111] Where φ, θ, ψ are the roll, pitch, and yaw angles of the drone;

[0112] I xx ,I yy ,I zz is the moment of inertia of the drone;

[0113] I xz is the product of inertia;

[0114] M xb ,M yb ,M zb They are respectively applied to drone x b ,y b and z b The resultant moment on the shaft;

[0115] Because I xz The absolute value is much smaller than I xx ,I yy ,I zz In order to facilitate the controller design, the absolute value of I xz The relevant components are considered as disturbances;

[0116] In practice, it is impossible to accurately obtain the torque acting on the drone, and the formula for the synthetic torque is as follows:

[0117]

[0118] Among them, M xknow ,M yknow and M zplane represents the known part of the resultant moment;

[0119] M xc ,M yc and M zc It represents the moment produced by the deflection of the control surface;

[0120] M xu ,M yu and Mzu It is expressed as the unknown uncertain part of the resultant torque;

[0121] The control-oriented model looks like this:

[0122]

[0123] where δ ac is the aileron deflection of the canard control surface;

[0124] δ at is the aileron deflection of the horizontal tail control surface;

[0125] δ ec is the elevator deflection of the canard control surface;

[0126] δ et is the elevator deflection of the horizontal tail control surface;

[0127] δ e is the rudder deflection of the vertical tail control surface;

[0128] is a coefficient related to the torque generated.

[0129] S2. Design an adaptive sliding mode control strategy that adapts to actuator faults and model uncertainties, and construct an integral sliding mode controller and control allocation scheme to ensure the tracking performance of the system under fault-free conditions;

[0130] Consider a full-chain nonlinear affine system with model uncertainty and actuator fault disturbances:

[0131]

[0132] in is the state vector;

[0133] is the control effectiveness matrix;

[0134] is the control input vector;

[0135] is a diagonal matrix;

[0136] represents an unknown perturbation whose bound is ||d(t)||≤D;

[0137] Vector F(x 1 (t),x 2 (t))∈R p is a nonlinear function that includes model uncertainty;

[0138] D f(t) = diag([d f1 (t),d f2 (t),...,d fm (t)]) is a diagonal matrix representing the effectiveness level of actuator control, where d fj (t)(j=1,2,...,m) satisfies 0≤d fj (t)≤1 if d fj (t) = 1, the jth actuator works normally, otherwise, the jth actuator will have a certain degree of failure;

[0139] To facilitate the design of the controller, the system state is defined as:

[0140]

[0141] The nonlinear affine system in S2 can be written as:

[0142]

[0143] Where i = 1, 2, 3 represents each subsystem;

[0144] x 2i-1 ∈[φ,θ,ψ] T ,

[0145] The design of a sliding mode controller usually consists of two steps. The first is to construct a sliding surface on which the desired system performance can be maintained; then a suitable control law is selected to force the sliding variable to reach the designed sliding surface and keep the sliding motion near the designed sliding surface.

[0146] Will and Expressed as the desired trajectory, the tracking error is then defined as and In this sense, the integral sliding surface of the system is defined as:

[0147]

[0148] in is a nonlinear combination of tracking errors,

[0149] g i Including the integral term, t 0 is the initial moment, a i1 ,a i2 and c i is the design parameter;

[0150] In this case, the integral sliding surface can be expressed as:

[0151]

[0152] The next step is to design a suitable control law to keep the sliding variable near the defined sliding surface. The sliding mode control law usually consists of two control parts, namely the continuous control part and the discontinuous control part. The continuous control part is obtained, and the disturbance d is not considered at this time i ;

[0153] At this time, the continuous control part can be designed as:

[0154]

[0155] Then, to compensate for the disturbance d i , the discontinuous control part is designed and integrated to ensure the ideal sliding motion, as shown in the following formula:

[0156] ζ i2 =-a i3 sign(s i )

[0157] In the formula, a i3 is a positive gain, making the sliding surface attractive to the sliding variable.

[0158] However, in order to achieve anti-interference capability, a larger a i3 This will increase the discontinuity of the sliding mode control and may cause unexpected control chattering. Smoothing discontinuities is done by adding a thin boundary layer near the defined sliding surface. The boundary layer can be expressed as:

[0159]

[0160] Using the defined boundary layer, a saturation function is defined to replace ζ i2 =-a i3 sign(s i ), as shown in the following formula:

[0161]

[0162] The continuous control part and the discontinuous control part can be combined, and the developed control law is expressed as:

[0163]

[0164] S3. Based on the fact that the hybrid CRW UAV under study is an over-actuated system, when an actuator fails, it is necessary to obtain fault information and redistribute the control signal to the available actuators. Considering the application of the control scheme in the actual system, when an actuator fails, it is necessary to immediately trigger the redistribution of the control signal. In this case, an adaptive control strategy that does not require any fault information is proposed to adapt to the actuator failure of the over-actuated system.

[0165] For the system considered above, the quadratic optimization algorithm is used to calculate the actual control input u so that the virtual control signal generated by the control allocation module can meet the requirements of the advanced sliding mode control to generate the virtual control signal;

[0166] The quadratic programming method based on minimizing the control input is described as:

[0167] J = arg minu T Qu

[0168] stv i =C ui u

[0169] The displayed solution is as follows:

[0170]

[0171] Where Q = Q T =diag([q 1 ,q 2 ,...,q m ]) is a symmetric positive definite weight matrix, corresponding to the residual control efficiency of the actuator. A common method to achieve fault-tolerant control of overdriven systems is to change the weighting matrix Q, which requires obtaining fault information from the so-called fault detection and diagnosis module. However, in practical applications, there may be a large time delay in obtaining the required fault information. In addition, there may be fault estimation errors, which will also affect the corresponding control performance. Therefore, an adaptive control strategy that does not require any fault information is proposed to adapt to actuator faults in overdriven systems.

[0172] When the actuator fails, the diagonal matrix Q will no longer be the unit matrix. Since the actual fault information cannot be obtained, if the matrix Q is still selected as the unit matrix to calculate the actual control input u, there will be a virtual control error. Under this condition, it can be obtained:

[0173] ζ i =C ui Eu-C ui (ED f )

[0174] Where E is the identity matrix;

[0175] Orderie =-C ui (ED f )u, the dynamic equation of the system can be rewritten as:

[0176]

[0177] where ζ id =C ui Eu represents the virtual control signal expected by the high-level controller.

[0178] It can be seen that in order to maintain the performance of the closed-loop system, the parameter h needs to be adaptively adjusted. i Eliminate the error of the virtual control signalζ ie ;

[0179] set up Then the following formula

[0180] J = arg minu T Qu

[0181] stv i =C ui u

[0182] Refactored to:

[0183]

[0184] Using estimates taking into account model uncertainty and actuator failures and To derive the corresponding high-level control law;

[0185] In this case, the formula

[0186]

[0187] Refactored to:

[0188]

[0189]

[0190] Then, by expressing and ζ i =C ui Eu-C ui (ED f )u can be rearranged as:

[0191]

[0192] in

[0193] To compensate for actuator failures and model uncertainties, the estimated parameters need to be updated online. and Use an online adaptive scheme to estimate uncertain parameters:

[0194]

[0195]

[0196] In the formula, Δs i =s i -Φ i sat(s i ) represents the distance between the sliding variable and the defined boundary layer. The adaptive scheme is triggered only when the sliding variable is outside the defined boundary layer, that is, when the system tracking performance is not ideal. By defining the variable, the overestimation of the uncertain parameters can be avoided, which helps to alleviate the system control chattering.

[0197] After the above steps, an adaptive fault-tolerant control method for model uncertainty and actuator failure is designed, and this method is applied to the hybrid CRW UAV dynamics model mentioned in S1. In order to verify the performance of the proposed adaptive fault-tolerant control (AFTC) scheme, simulations are carried out under different uncertainty and fault scenarios based on the studied overdriven hybrid CRW UAV in fixed-wing flight mode. In the following simulation scenario, CSMC is demonstrated as a combination of a high-level controller and a low-level control distribution module. The UAV is controlled to track a set of pre-designed pitch commands, which are given as sinusoidal signals with an amplitude of 6° and a period of 10s.

[0198] In addition to the uncertainty of the resultant moment acting on the studied UAV, the uncertainty of the roll, pitch and yaw moments of inertia is set to 10% of their measured values. For the actuator failure mode, a 60% control effectiveness loss fault is injected into the elevator control surface at 15s, while a 70% control effectiveness loss fault is injected into the canard control surface.

[0199] When a 60% control failure occurs in the elevator and a 70% control failure occurs in the canard, the tracking performance of the UAV speed is as follows: Figure 2 shown.

[0200] When the actuator failure occurs at 15 seconds, the adaptive fault-tolerant control method proposed in this embodiment can still maintain the original tracking performance of the speed by adaptively changing the control parameters, however, the compared CSMC cannot maintain the original tracking performance.

[0201] When the actuator fails, the control inputs of the canard and elevator are as follows Figure 3 and Figure 4As shown in Figure 2, when a canard and elevator failure occurs, their control inputs will increase to eliminate the adverse effects of the failure. In addition, to compensate for the failure, the control inputs of the canard and elevator used by the proposed AFTC are significantly smaller than those of the compared CSMC, which indicates that the proposed control scheme does not use more control efficiency when compensating for the failure.

[0202] The present invention proposes an adaptive fault-tolerant control method for an overdriven hybrid UAV to compensate for actuator faults and model uncertainties without requiring fault and uncertainty information.

[0203] The control method is divided into two independent control modules: a high-level adaptive sliding mode control module and a low-level control allocation module.

[0204] The low-level control distribution module is used to distribute the virtual control signals generated by the high-level control module among the redundant available actuators. The high-level control module is composed of an adaptive sliding mode controller to maintain the overall tracking performance of the system under fault and uncertain conditions.

[0205] In the case of actuator faults and model uncertainty, the adaptive scheme will be triggered to generate more virtual control signals. Using the comprehensive adaptive scheme, the control parameters can be adaptively changed to compensate for the virtual control errors.

[0206] Comparative simulation results show that this method is effective and superior in both single actuator failure and concurrent actuator failure cases.

[0207] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0208] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive fault-tolerant control method for a vertical take-off and landing aircraft. Features: The following steps are involved: S1. First, establish the dynamic model of vertical take-off and landing hybrid rotary wing UAV; S2. Design an adaptive sliding mode control strategy that adapts to actuator failures and model uncertainty, construct an integral sliding mode controller and control allocation scheme to ensure the tracking performance of the system under fault-free conditions; Consider a full-chain nonlinear affine system with model uncertainty and actuator fault disturbances: where x(t) = [x 1 (t),x 2 (t)] T ∈ n is the state vector; C u ∈ p*m is the control effectiveness matrix; u(t)∈ m is the control input vector; H∈ p*p is a diagonal matrix; d(t)∈ p represents an unknown perturbation whose bound is ||d(t)||≤D; Vector F(x 1 (t),x 2 (t))∈R p is a nonlinear function that includes model uncertainty; D f (t) = diag([d f1 (t),d f2 (t),...,d fm (t)]) is a diagonal matrix representing the effectiveness level of actuator control, where d fj (t)(j=1,2,…,m) satisfies 0≤d fj (t)≤1 if d fj (t) = 1, the jth actuator works normally, otherwise, the jth actuator will have a certain degree of failure; S3. Combining the hybrid rotary wing UAV as an overdrive system, an adaptive control method that does not require any fault information is proposed to adapt to the actuator failure of the overdrive system. Specifically, when an actuator fails, the proposed adaptive fault-tolerant control scheme can update the control parameters in real time to compensate for the virtual control error and redistribute the control signal to the available redundant actuators. For the drive system, a quadratic optimization algorithm is used to calculate the actual control input u, so that the virtual control signal generated by the control allocation module can meet the requirements of the advanced sliding mode control to generate the virtual control signal; The quadratic programming method based on minimizing the control input is described as: Jarg minu T What s.t.v i =C ui u The displayed solution is as follows: Where Q = Q T =diag([q 1 ,q 2 ,...,q m ]) is a symmetric positive definite weight matrix, corresponding to the residual control efficiency of the actuator.

2. The adaptive fault-tolerant control method for a vertical take-off and landing aircraft according to claim 1, It is characterized in that In S1, the dynamic model of the vertical take-off and landing hybrid rotary wing UAV is shown in the following formula: Where φ, θ, ψ are the roll, pitch, and yaw angles of the drone; I xx ,I yy ,I zz is the moment of inertia of the drone; I xz is the product of inertia; M xb ,M yb ,M zb They are respectively applied to drone x b ,y b and z b The resultant moment on the shaft; Among them, I xz The absolute value is much smaller than I xx ,I yy ,I zz The absolute value of I xz The relevant components are regarded as disturbances, and the formula for the resultant torque is as follows: Among them, M xknow ,M yknow and M zplane represents the known part of the resultant torque; M xc ,M yc and M zc It represents the moment produced by the deflection of the control surface; M xu ,M yu and M zu It is expressed as the unknown uncertain part of the resultant torque; The control-oriented model is as follows: where δ ac is the aileron deflection of the canard control surface; δ at is the aileron deflection of the horizontal tail control surface; δ ec is the elevator deflection of the canard control surface; δ et is the elevator deflection of the horizontal tail control surface; δ e is the rudder deflection of the vertical tail control surface; is a coefficient related to the torque generated.

3. The adaptive fault-tolerant control method for a vertical take-off and landing aircraft according to claim 2, It is characterized in that In S2, the whole chain nonlinear affine system with actuator failure and model uncertainty is specifically expressed as follows: where x(t) = [x 1 (t),x 2 (t)] T ∈ n is the state vector; C u ∈ p*m is the control effectiveness matrix; u(t)∈ m is the control input vector; H∈ p*p is a diagonal matrix; d(t)∈ p represents an unknown perturbation whose bound is ||d(t)||≤D; Vector F(x 1 (t),x 2 (t))∈R p is a nonlinear function that includes model uncertainty; D f (t) = diag([d f1 (t),d f2 (t),...,d fm (t)]) is a diagonal matrix representing the effectiveness level of actuator control; where d fj (t)(j=1,2,…,m) satisfies 0≤d fj (t)≤1 if d fj (t) = 1, the jth actuator works normally, otherwise, the jth actuator fails; The system state is then defined as: The nonlinear affine system can be written as: Where i = 1, 2, 3 represents each subsystem; Will and Expressed as the desired trajectory, the tracking error is then defined as and Then the integral sliding surface of the system is defined as: in is a nonlinear combination of tracking errors, g i Including the integral term, t 0 is the initial moment, a i1 ,a i2 and c i is the design parameter; Then the integral sliding surface can be expressed as: By making Get the continuous control part without considering the disturbance d i ; The continuous control part is: To compensate for the disturbance d i , the discontinuous control part is integrated to ensure the ideal sliding motion, as shown in the following formula: g i2 =-a i3 sign(s) i ) Among them, a i3 is a positive gain, making the sliding surface attractive to the sliding variable.

4. The adaptive fault-tolerant control method for a vertical take-off and landing aircraft according to claim 3, It is characterized in that Add a boundary layer around the defined sliding surface. The boundary layer is shown in the following formula: Using the boundary layer, a saturation function is defined to replace ζ i2 =-a i3 sign(s i ), as shown in the following formula: The continuous control part and the discontinuous control part can be combined, and the developed control law is expressed as:

5. The adaptive fault-tolerant control method for a vertical take-off and landing aircraft according to claim 4, It is characterized in that The quadratic programming method to minimize the control input is described as: Jarg minu T What s.t.v i =C ui u The displayed solution is: Where Q = Q T =diag([q 1 ,q 2 ,...,q m ]) is a symmetric positive definite weight matrix, corresponding to the residual control efficiency of the actuator; In S3, when the actuator fails, the diagonal matrix Q will no longer be the unit matrix. Since the actual fault information cannot be obtained, if the matrix Q is still selected as the unit matrix to calculate the actual control input u, there will be a virtual control error. Under this condition, it can be obtained: ζ i =C ui I-C ui (ED f )u Where E is the identity matrix; Order ie =-C ui (ED f )u, the following formula It can be written as: where ζ id =C ui Eu represents the virtual control signal expected by the high-level controller; That is, in order to maintain the performance of the closed-loop system, the parameter h needs to be adaptively adjusted. i Eliminate the error of the virtual control signalζ ie ; set up Then the following formula Jarg minu T What s.t.v i =C ui u Can be refactored to: Incorporating model uncertainty and actuator failures, using estimates and To derive the advanced control law, we can use the formula Refactored to: Then, by expressing and ζ i =C ui Eu-C ui (ED f )u can be rearranged as: in 6. The adaptive fault-tolerant control method for a vertical take-off and landing aircraft according to claim 5, It is characterized in that To compensate for actuator failures and model uncertainties, the estimated parameters need to be updated online. and Use an online adaptive scheme to estimate uncertain parameters: In the formula, Δs i =s i -Φ i sat(s i ) represents the distance between the sliding variable and the boundary layer; When the sliding variable is outside the boundary layer, that is, the system tracking performance is not ideal, the adaptive scheme is triggered. By defining the variables, the overestimation of the uncertain parameters can be avoided, which is conducive to alleviating the system control chattering.

Images