CN103235504B - Flight control method for large civil aircrafts on basis of direct adaptive control reconfiguration - Google Patents

Abstract

The invention discloses a flight control method for large civil aircrafts on the basis of direct adaptive control reconfiguration, and relates to the field of control technology for aircrafts. The flight control method has the advantages that a direct adaptive process and a model following process are combined with each other and are applied to B747-200 aircrafts, control reconfiguration flight simulation for the large civil aircrafts in an actuator failure mode is implemented, reference is provided for control reconfiguration for the large civil aircrafts in the actuator failure mode, and accordingly the flight control method has an important significance in improving the safety of the aircrafts, reducing the accident rate of the aircrafts and improving the survival ability of the aircrafts under failure conditions.

Description

Based on the Large Civil Aircraft flight control method of direct adaptive control reconstruct

Technical field

The present invention relates to a kind of method of flying vehicles control technical field, be specifically related to a kind of direct adaptive control restructing algorithm based on model following and the utilization in Large Civil Aircraft flight controls thereof.

Background technology

Reconfigurable Flight Control System (Reconfigurable Flight Control System, RFCS) controller architecture or computing controller gain can be redesigned when aircraft breaks down, utilize the control redundancy feature of aircraft, independently also ensure flight safety in time.

Classical control reconfiguration method can be divided into active reconstruct and passive type reconstruct.Passive type reconstruct is based on the method for designing of robust control, and the fault-tolerant ability by control law inherence allows that some kind and fault to a certain degree exist, but to the unknown, the fault that happens suddenly then lacks adaptive ability.Reconstruct different from controlled formula, active reconstruct, based on the redesign of controller, can process a lot of unknown failure.

Self-adapting reconstruction method belongs to active reconstruct.Compared with other control method, the sharpest edges of self-adapting reconstruction method are when in the face of a large amount of uncertain parameter, and still to ensure needed for aircraft asymptotic follows.Its design philosophy utilizes error to adjust controller parameter to realize adaptive control.Self-adapting reconstruction method is divided into again indirect self-adaptive method and direct adaptive method.Indirect self-adaptive method is in two steps: parameter identification and parameter calculate.Parameter identification needs the input stimulus that continues and level and smooth input-output data, and along with the increase of number of parameters, parameter identification becomes difficulty.For the ease of calculating, need to force restriction to the input-output data in past, and this may cause the deterioration of reconstruction property.Direct self-adaptation rule can avoid these problems, because it does not need the accurate information of identification system, but carries out direct estimation to the parameter of controller.

Through finding the retrieval of prior art, Zhang Weiguo etc., disclose a kind of flight control system actuator failure reconfiguration method for designing based on Robust tracking control in " the flight control system self-adapting reconstruction technical research based on Robust tracking control " (Xi'an, Shaanxi, Northwestern Polytechnical University Automation Institute 710072).The identification model that the method is parallel according to the randomness design series of fault, the corresponding a kind of fault of each model, corresponds to each identification model simultaneously and sets up a controller.But this technology is when processing unknown indeterminate fault, owing to not having corresponding Matching Model in model bank, cause occurring larger diagnosis deviation, thus control reconfiguration performance is under some influence.The present inventor, for the flight control system containing unknown indeterminate fault, follows and residual error logic analysis angle from nominal model, carries out adaptive failure detection and treatment.

Liu little Xiong etc. (control and decision-making " reconfigurable flight control system based on direct adaptive control is studied ", 2007) direct Adaptive Reconfigurable Control is proposed in, and adjust ginseng online by PI proportional plus integral control rule, this algorithm is when CONTROLLER DESIGN, have employed state and the input feedback of common Error Feedback and reference model, but this control mode is difficult to the stability ensureing controlled device, the present invention intends the feedback of status introducing controlled device in the controller, ensures the stability of controlled device.

Zhang Weiguo etc. are at " Fault Tolerance Flight Control System fault isolation and self-adapting reconstruction design " (Northcentral University's journal, 2007) a kind of Fault Tolerance Flight Control System algorithm for design based on multi-model self-adapting Quadratic optimal is disclosed in, this algorithm, by setting up multiple identification model and designing based on the model following controller retraining quadratic form, realizes the isolation of flight control system real time fail and reconstruct.But this technology adopts unified reference model to replace practical flight operating mode, when flying condition depart from reference model larger time reconfigurable control can be made to bring relatively large deviation; Secondly, during consideration multi-model switching stability, author has used desalination link to process, and the method can suppress high frequency noise, but the stability of system is also relevant with the stability margin of switched system, if switched system limit is unstable, even if there is desalination link to be also difficult to ensure to switch stablize.The present invention intends meeting Lyapunov stability angle from system and starts with, and carries out Stable Adaptive Control design to reconfigurable controller.

Summary of the invention

The present invention is directed to prior art above shortcomings, a kind of direct adaptive control reconstructing method controlled for Large Civil Aircraft flight is proposed, direct adaptive approach and model following methods combining are applied to B747-200 aircraft, realize the Large Civil Aircraft control reconfiguration flight simulation under actuator fault mode, for the control reconfiguration under Large Civil Aircraft actuator fault mode provides reference, significant to the viability improved in aircraft safety, reduction aircraft accident rate, raising airplane fault situation.

The present invention is achieved by the following technical solutions, the present invention includes following steps:

The first step, set up the kinetic model comprising controlled device linear model and aircraft reference model of Large Civil Aircraft, concrete steps comprise:

1.1) under the model of flight of setting, force analysis is carried out to aircraft, that set up the movement of aircraft barycenter with kinetics equation that is rotation around center of mass, use microvariations principle that nonlinear kinetics equation is done linearization process, again system of linear equations is configured to the form of matrix, just obtains the controlled device linear model under state space;

1.2) on the basis of plant model, use the aircraft reference model under the method design state space such as Method of Pole Placement, LQR (Linear-quadratic Regulator linearquadratic regulator) method, this aircraft reference model is stable closed-loop system, have desirable dynamic perfromance and follow target as controlled device.

Described plant model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_p=A_p{x}_p+B_p{u}_p\\ y_p=C_p{x}_p\end{array}\right.,$ Wherein: A _p∈ R ^{n × n}, B _p∈ R ^{n × m}, C _p∈ R ^{p × n}, ABC is normal parametric array, x _p∈ R ⁿfor system state, vector x _pin the longitudinal direction of aircraft controls, generally comprise forward speed, longitudinal velocity, pitch rate and the angle of pitch, in horizontal side direction controls, generally comprise yaw angle, roll angle speed, yawrate and roll angle, y _p∈ R ^pfor system exports, for control vector, in the longitudinal direction of aircraft controls, generally comprise the elevating rudder degree of bias and the engine oil-valve operation bar degree of bias, in horizontal side direction controls, generally comprise amount of rudder and the aileron degree of bias.

Described aircraft reference model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_m=A_m{x}_m+B_m{u}_m\\ y_m=C_m{x}_m\end{array}\right.,$ This model is stable closed-loop system, and has desirable dynamic perfromance, wherein: A _m∈ R ^{n × n}normal parameter stability matrix, B _m∈ R ^{n × l}, C _m∈ R ^{p × n}be all normal parametric array, for system state, y _m∈ R ^pfor system exports, for bounded reference input.

Second step, structure controlled device controller, failed controller and adaptive controller, wherein: the input of controlled device controller comprises: model following direct adaptive control input (hereinafter referred to as " adaptive control input ") contacted by matrix of coefficients and fail-safe control input; Failed controller is by the impact of fault with in the present control item of the form body of control inputs, and according to the difference of actuator failure condition, value when being presented as control inputs is also different; Adaptive controller comprises the control item of handling failure input, and comprises the item such as controlled device feedback of status, the feedforward of reference model state according to the thought of model following.

Described adaptive control is input as u _p, wherein:

σ＝diag{σ ₁，…，σ _m}

σ is matrix of coefficients,

U is adaptive controller,

for failed controller, the controlled quentity controlled variable value of corresponding common actuator fault is: $u_{p_j}\left(t\right)={\stackrel{-}{u}}_{p_j},\∀t\≥t_j,j\∈\{1,\·\·\·,m\}$

wherein: t _jrepresent the moment that a jth actuator breaks down, x _mmatrix of coefficients K ₀constant gain amplifier battle array, when other gain battle array all time become, constant gain amplifier battle array K ₀be conducive to increasing system robustness; V item, in order to offset fault effects, is the core of u.Adaptive control gain battle array C ₀, G ₀and the calculating of adaptive control vector v belongs to the category of direct adaptive approach.By online updating C ₀, G ₀, v parameter, make u have adaptive ability to fault.

When actuator breaks down, matrix of coefficients by corresponding for shielding adaptive control input item, enable corresponding fail-safe control input item, other of simultaneously adaptive control input still works; When under non-failure conditions, the impact, only the adaptive control input that mask fail-safe control input are worked by matrix of coefficients.

Described fail-safe control input, for replacing existing fault model, directly acts on flight control system the impact of fault with the form of control inputs.

Described matrix of coefficients, when actuator fault, to the corresponding entry of actuator becoming 1 from 0, thus is converted into corresponding control inputs item the fault of actuator.

In reality, because not needing failure inspect and indentifition, so it is all unknown that matrix of coefficients in controlled device control inputs and fail-safe control input, adaptive controller according to the gain matrix of the output error on-line tuning adaptive controller of controlled device and reference model and gain vector, can realize controlled device following reference model.

3rd step, be embodied in 2 adaptive gain battle array (C due to the online adaptive faculty of adaptive controller to fault ₀, G ₀) and 1 adaptive gain vector (v) in, this adaptive gain battle array and adaptive gain vector are according to the output error on-line tuning size of controlled device and reference model; The turnover rate of adaptive gain battle array and gain vector is obtained according to Lyapunov stability theory, utilize turnover rate, adaptive gain battle array and gain vector then can realize online updating according to the output error of controlled device and reference model, thus realize the online adaptive to actuator fault.

Described output error is e, e=y _p-y _m=C _px _p-C _mx _m.

Described adaptive gain battle array and the turnover rate of gain vector value is $\left\{\begin{array}{c}{\stackrel{\·}{G}}_0=-{\mathrm{\γ}}_1{B}_m^T{C}_m^T{\mathrm{Pex}}_p^T\\ {\stackrel{\·}{C}}_0=-{\mathrm{\γ}}_2{C}_0{B}_m^T{C}_m^T{\mathrm{Peu}}^T{C}_0\\ \stackrel{\·}{v}=-{\mathrm{\γ}}_3{B}_m^T{C}_m^T\mathrm{Pe}\end{array}\right.,$ Wherein: γ ₁, γ ₂, γ ₃be respectively self-adaptation constant gain, P is symmetric positive definite matrix, and its determining step is as follows:

1) use POLE PLACEMENT USING thought, select for a post meaning stable matrix A _e, the same A of dimension _m, wherein: A _emeet error dynamics

$\stackrel{\·}{e}=A_{e}e+C_m{B}_m\mathrm{\ΔG}{x}_p+C_m{B}_m\mathrm{\Δv}+C_m{B}_m\mathrm{\Δ\ψu};$

2) meaning symmetric positive definite matrix Q is selected for a post

3) P is made to meet: $A_e^T+{\mathrm{PA}}_e=-Q.$

To achieve these goals, the innovative point of technical solution of the present invention is: 1) fault effects is put on controller with the form of control inputs, directly effectively reflects airplane fault to the impact controlled; Fail-safe control input consider control surface stuck, lost efficacy, the floating several typical fault pattern of pine; 2) when CONTROLLER DESIGN, the present invention selects the state with reference to model but not the part exported as controller, avoids the loss of learning being multiplied by observing matrix and may causing; 3) utilize the output error of system instead of state error to do solving of adaptive updates rate, more meet engineering reality.

Accompanying drawing explanation

Fig. 1 is schematic diagram of the present invention.

Fig. 2 is model following direct adaptive controller structural drawing.

Fig. 3 is the derivation theory diagram of the turnover rate of adaptive gain battle array and gain vector.

Fig. 4 be yaw rudder stuck after, yaw angle response curve schematic diagram;

Fig. 5 be yaw rudder stuck after, roll angle rate response curve synoptic diagram;

Fig. 6 be yaw rudder stuck after, yawrate response curve schematic diagram;

Fig. 7 be yaw rudder stuck after, the response curve schematic diagram of roll angle;

In figure: dotted line (annotation is 1) represents the response of reference model; The response of controlled device after the direct adaptive approach reconstruct of solid line (annotation is 2) expression model following of the present invention; Dotted line (annotation is 3) represents with the response of controlled device under not having the model following method of adaptive ability to control.

Fig. 8 be aileron stuck after, yaw angle response curve schematic diagram.

Fig. 9 be aileron stuck after, roll angle rate response curve synoptic diagram.

Figure 10 be aileron stuck after, yawrate response curve schematic diagram.

Figure 11 be aileron stuck after, the response curve schematic diagram of roll angle.

Embodiment

Elaborate to embodiments of the invention below, the present embodiment is implemented under premised on technical solution of the present invention, give detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Embodiment 1

Embodiment comprises the following steps:

The first step, set up the kinetic model comprising controlled device linear model and aircraft reference model of Large Civil Aircraft, concrete steps comprise:

1.1) under the model of flight of setting, force analysis is carried out to aircraft, that set up the movement of aircraft barycenter with kinetics equation that is rotation around center of mass, use microvariations principle that nonlinear kinetics equation is done linearization process, again system of linear equations is configured to the form of matrix, just obtains the controlled device linear model under state space;

1.2) on the basis of plant model, use Method of Pole Placement, LQR (Linear-quadratic Regulator, linearquadratic regulator) aircraft reference model under the method design state space such as method, this aircraft reference model is stable closed-loop system, have desirable dynamic perfromance and follow target as controlled device.

Described plant model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_p=A_p{x}_p+B_p{u}_p\\ y_p=C_p{x}_p\end{array}\right.,$ Wherein: A _p∈ R ^{n × n}, B _p∈ R ^{n × m}, C _p∈ R ^{p × n}a _p, B _p, C _prepresent controlled aircraft PARAMETERS IN THE LINEAR MODEL respectively, and be normal parametric array, x _p∈ R ⁿfor n ties up system state, vector x _pin the longitudinal direction of aircraft controls, generally comprise forward speed, longitudinal velocity, pitch rate and the angle of pitch, in horizontal side direction controls, generally comprise yaw angle, roll angle speed, yawrate and roll angle, y _p∈ R ^pexport for p maintains system, for m ties up control vector, in the longitudinal direction of aircraft controls, generally comprise the elevating rudder degree of bias and the engine oil-valve operation bar degree of bias, in horizontal side direction controls, generally comprise amount of rudder and the aileron degree of bias.

Described aircraft reference model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_m=A_m{x}_m+B_m{u}_m\\ y_m=C_m{x}_m\end{array}\right.,$ This model is stable closed-loop system, and has desirable dynamic perfromance, wherein: A _m∈ R _{n × n}normal parameter stability matrix, B _m∈ R ^{n × l}for n is capable, l arranges normal parametric array, C _m∈ R ^{p × n}for p is capable, n arranges normal parametric array, for m ties up system state, y _m∈ R ^pexport for p maintains system, for l ties up bounded reference input.

Second step, structure controlled device controller, failed controller and adaptive controller, wherein: the input of controlled device controller comprises: model following direct adaptive control input (hereinafter referred to as " adaptive control input ") contacted by matrix of coefficients and fail-safe control input; Failed controller is by the impact of fault with in the present control item of the form body of control inputs, and according to the difference of actuator failure condition, value when being presented as control inputs is also different; Adaptive controller comprises the control item of handling failure input, and comprises the item such as controlled device feedback of status, the feedforward of reference model state according to the thought of model following.

Described adaptive control is input as u _p, wherein

σ is matrix of coefficients,

σ＝diag{σ ₁，…，σ _m}

U is adaptive controller,

for failed controller, the controlled quentity controlled variable value of corresponding common actuator fault is: $u_{p_j}\left(t\right)={\stackrel{-}{u}}_{p_j},\∀t\≥t_j,j\∈\{1,\·\·\·,m\}$

wherein: t _jrepresent the moment that a jth actuator breaks down, x _mmatrix of coefficients K ₀constant gain amplifier battle array, when other gain battle array all time become, constant gain amplifier battle array K ₀be conducive to increasing system robustness; V item, in order to offset fault effects, is the core of u.Adaptive control gain battle array C ₀, G ₀and the calculating of adaptive control vector v belongs to the category of direct adaptive approach.By online updating C ₀, G ₀, v parameter, make u have adaptive ability to fault.

Fig. 2 is the structural drawing (consistent with formula (a)) of adaptive controller u.C in dotted line frame ₀, G ₀, v is the target that adaptive algorithm will obtain, and is gating matrix and the control vector of online updating, it upgrades according to being adaptive gain battle array turnover rate in solid wire frame

3rd step, solve adaptive gain battle array C in formula (a) ₀, G ₀with the more new law of gain vector v

(I) output error e, e=y is defined _p-y _m=C _px _p-C _mx _m;

(II) self-adaptation constant gain γ is selected ₁, γ ₂, γ ₃;

(III) symmetric positive definite matrix P is determined:

1. use POLE PLACEMENT USING thought, select for a post meaning stable matrix A _e, the same A of dimension _m, wherein, A _emeet error dynamics

$\stackrel{\·}{e}=A_{e}e+C_m{B}_m\mathrm{\ΔG}{x}_p+C_m{B}_m\mathrm{\Δv}+C_m{B}_m\mathrm{\Δ\ψu}.---\left(b\right)$

Wherein $\left\{\begin{array}{c}\mathrm{\ΔG}=G_0-G_0^{*}\\ \mathrm{\Δ\ψ}=C_0^{*-1}-C_0^{-1}\\ \mathrm{\Δv}=v-v^{*}\end{array}\right.$ For systematic error matrix and error vector.

2. meaning symmetric positive definite matrix Q is selected for a post;

3. P is made to meet following equation $A_e^{T}P+{\mathrm{PA}}_e=-Q.$

(IV) the more new law of adaptive gain battle array and gain vector value is

$\left\{\begin{array}{c}{\stackrel{\·}{G}}_0=-{\mathrm{\γ}}_1{B}_m^T{C}_m^T{\mathrm{Pex}}_p^T\\ {\stackrel{\·}{C}}_0=-{\mathrm{\γ}}_2{C}_0{B}_m^T{C}_m^T{\mathrm{Peu}}^T{C}_0\\ \stackrel{\·}{v}=-{\mathrm{\γ}}_3{B}_m^T{C}_m^T\mathrm{Pe}\end{array}\right.,$ C the resolution principle of () formula (b) is shown in block diagram 3.

4th step, the above-mentioned equation (a) of simultaneous, (c) etc., independent variable comprises controlled device state x _p, reference model state x _m, the adaptive gain battle array C in adaptive controller ₀, G ₀with adaptive gain vector v, each independent variable of line solver.

Illustrate 1: structure Lyapunov function

V=e ^tpe+tr [Δ G ^tΔ G/ γ ₁]+tr [Δ ψ ^tΔ ψ/γ ₂]+Δ v ^tΔ v/ γ ₃≤ 0, it is asymptotically stable for can obtaining error system (b).

Illustrate 2: in adaptive gain battle array and each parameter needed for gain vector renewal process, the A of positive definite matrix P by setting in advance _ecalculate with Q, u _m, x _mbe all known reference model parameter, only have x _pneed online measuring.Adaptive law solve the parameter not needing system, be the very large advantage of direct adaptive method.

Implementation result is assessed: cruise under mode at height 40000 feet, speed 0.8Ma, ask for controlled device B747-200 Aircraft Lateral lateral dynamic model.Wherein, state vector x _pfor yaw angle, roll angle speed, yawrate and roll angle, control vector u _pfor amount of rudder and the aileron degree of bias, C _pfor quadravalence unit matrix,

$A_p=\left[\begin{array}{cccc}-0.0558& 0.0802& -0.9968& 0.0415\\ -3.0500& -0.4650& 0.3880& 0\\ 0.5980& -0.0316& -0.1150& 0\\ 0& 1.0000& 0.0804& 0\end{array}\right],$ $B_p=\left[\begin{array}{cc}0& 0.0073\\ 0.1430& 0.1530\\ 0.0077& -0.4750\\ 0& 0\end{array}\right];$

In reference model, x _m, u _mthe same controlled device of parameter, C _mfor quadravalence unit matrix,

$A_m=\left[\begin{array}{cccc}-0.1303& 0.1033& -0.9362& 0.05787\\ -4.081& -0.3221& 1.401& 0.07537\\ 5.476& -1.552& -4.07& -1.08\\ 0& 1& 0.0804& 0\end{array}\right],$ $B_m=\left[\begin{array}{cc}0& 0.0073\\ 0.143& 0.153\\ 0.0077& -0.475\\ 0& 0\end{array}\right].$

The state of initialization controlled device and reference model is x _m(0)=x _p(0)=[0,0,0,0] ^t.In simulation result, dotted line (being labeled as 1) represents the response of reference model, the response of controlled device after the reconstruct of solid line (being labeled as 2) expression model following of the present invention direct adaptive approach, dotted line (being labeled as 3) represents the response by the controlled device under not having the model following method of adaptive ability to control.

Setting direction rudder is stuck on+10 ° of positions the 2nd second time, and Fig. 4-Fig. 7 is the response curve of yaw angle, roll angle speed, yawrate and roll angle respectively.Arrange aileron stuck on+15 ° of positions the 2nd second time, Fig. 8-Figure 11 is the response curve of yaw angle, roll angle speed, yawrate and roll angle respectively.

From Fig. 4-Fig. 7, the yaw rudder fault of controlled device occurs in the adjustment process of dynamic response, before system enters stable state, just arrange fault, is intended to the re-configurability verifying context of methods.Yaw rudder stuck on+10 ° of positions after, all there is oscillation on large scale in each state of object under not having the model following method of adaptive ability to control, departs from the response of reference model far away; Again following reference model can be completed in 3 seconds by each state of the object after context of methods reconstruct, in Fig. 4, the situation of following of yaw angle is very little by fault effects, all the time good following state is in, in Fig. 7 there is of short duration crest in the response of yawrate after fault occurs, and is because rudder kick directly affects the change in course.

From Fig. 8-Figure 11, the aileron time of failure of controlled device is also selected in adjustment process, the equal oscillation on large scale of each state of the object under not having the model following method of adaptive ability to control, and firmly followed by reference model by each state of the object after context of methods reconstruct, in Fig. 8 and Figure 10, the response curve of yaw angle and yawrate almost overlaps with reference model.

Claims (7)

1., based on a Large Civil Aircraft flight control method for direct adaptive control reconstruct, it is characterized in that, comprise the following steps:

The first step, set up the kinetic model comprising controlled device linear model and aircraft reference model of Large Civil Aircraft, concrete steps comprise:

1.1) under the model of flight of setting, force analysis is carried out to aircraft, that set up the movement of aircraft barycenter with kinetics equation that is rotation around center of mass, use microvariations principle that nonlinear kinetics equation is done linearization process, again system of linear equations is configured to the form of matrix, just obtains the controlled device linear model under state space;

1.2) on the basis of controlled device linear model, use Method of Pole Placement, aircraft reference model under linearquadratic regulator method design point space, this aircraft reference model is stable closed-loop system, have desirable dynamic perfromance and follow target as controlled device;

Second step, structure controlled device controller, failed controller and adaptive controller, wherein: the input of controlled device controller comprises: the model following direct adaptive control input contacted by matrix of coefficients and fail-safe control input; Failed controller is by the impact of fault with in the present control item of the form body of control inputs, and according to the difference of actuator failure condition, value when being presented as control inputs is also different; Adaptive controller comprises the control item of handling failure input, and comprises controlled device feedback of status, the feedforward of reference model state according to the thought of model following;

3rd step, be embodied in 2 adaptive gain battle array (C due to the online adaptive faculty of adaptive controller to fault ₀, G ₀) and 1 adaptive gain vector (v) in, this adaptive gain battle array and adaptive gain vector are according to the output error on-line tuning size of controlled device and aircraft reference model; The turnover rate of adaptive gain battle array and gain vector is obtained according to Lyapunov stability theory, utilize turnover rate, adaptive gain battle array and gain vector then can realize online updating according to the output error of controlled device and aircraft reference model, thus realize the online adaptive to actuator fault.

2. method according to claim 1, is characterized in that, described controlled device linear model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_p=A_p{x}_p+B_p{u}_p\\ y_p=C_p{x}_p\end{array}\right.,$ Wherein: A _p∈ R ^{n × n}, B _p∈ R ^{n × m}, C _p∈ R ^{p × n}, A B C is normal parametric array, x _p∈ R ⁿfor system state, vector x _pin the longitudinal direction of aircraft controls, generally comprise forward speed, longitudinal velocity, pitch rate and the angle of pitch, in horizontal side direction controls, generally comprise yaw angle, roll angle speed, yawrate and roll angle, y _p∈ R ^pfor system exports, for control vector, in the longitudinal direction of aircraft controls, generally comprise the elevating rudder degree of bias and the engine oil-valve operation bar degree of bias, in horizontal side direction controls, generally comprise amount of rudder and the aileron degree of bias.

3. method according to claim 1, is characterized in that, described aircraft reference model is: $\left\{\begin{array}{c}{\stackrel{\·}{x}}_m=A_m{x}_m+B_m{u}_m\\ y_m=C_m{x}_m\end{array}\right.,$ This model is stable closed-loop system, and has desirable dynamic perfromance, wherein: A _m∈ R ^{n × n}normal parameter stability matrix, B _m∈ R ^{n × l}, C _m∈ R ^{p × n}be all normal parametric array, for system state, y _m∈ R ^pfor system exports, $u_m={[u_{p_1},\·\·\·,u_{p_l}]}^T\∈R^l$ For bounded reference input.

4. method according to claim 1, is characterized in that, described adaptive control is input as u _p, wherein:

σ＝diag{σ ₁,…,σ _m}

σ is matrix of coefficients,

U is adaptive controller,

for failed controller, the controlled quentity controlled variable value of corresponding common actuator fault is:

$u_{p_j}\left(t\right)={\stackrel{\‾}{u}}_{p_j},\∀t\≥t_j,j\∈\{1,\·\·\·,m\}$

wherein: t _jrepresent the moment that a jth actuator breaks down, x _mmatrix of coefficients K ₀constant gain amplifier battle array, when other gain battle array all time become, constant gain amplifier battle array K ₀be conducive to increasing system robustness; V item, in order to offset fault effects, is the core of u, adaptive control gain battle array C ₀, G ₀and the calculating of adaptive control vector v belongs to the category of direct adaptive approach, by online updating C ₀, G ₀, v parameter, make u have adaptive ability to fault;

When actuator breaks down, matrix of coefficients σ by corresponding for shielding adaptive control input item, enable corresponding fail-safe control input item, other of simultaneously adaptive control input still works; When under non-failure conditions, the impact, only the adaptive control input that mask fail-safe control input are worked by matrix of coefficients.

5. method according to claim 1, is characterized in that, described fail-safe control input, for replacing existing fault model, directly acts on flight control system the impact of fault with the form of control inputs.

6. method according to claim 1, is characterized in that, described matrix of coefficients, when actuator fault, to the corresponding entry of actuator becoming 1 from 0, thus is converted into corresponding control inputs item the fault of actuator.

7. method according to claim 1, is characterized in that, described output error is e, e=y _p-y _m=C _px _p-C _mx _m.