Aircraft attack angle protection control method based on dynamic instruction correction

Abstract

The invention discloses an aircraft attack angle protection control method based on dynamic instruction correction, belongs to the field of flight control methods, and is used for solving the tracking control problem under the condition that the aircraft attack angle is limited. Firstly, based on an aircraft dynamics model, a backstepping method is utilized to design an adaptive controller to ensure stable tracking of a system, a virtual attack angle instruction part needs to be limited by adopting a saturation function, and a limited boundary value is provided by a subsequent protection system. And then, designing a prediction model representing correlation between the attack angle and the attack angle instruction and dynamics based on the designed closed-loop control system, reversely deducing a complex function relationship between the attack angle instruction and the system state and dynamics through the prediction model, and constructing an attack angle protection system. And finally, calculating an attack angle instruction boundary value through the prediction model and a system motion state trend under the current input, and correcting an original virtual attack angle instruction boundary so as to realize attack angle protection. According to the method, stable tracking of the system under attack angle constraint can be ensured, the residual attack angle in the envelope can be fully utilized, the maneuverability of the aircraft is improved, and the method has an engineering application reference value.

Description

Technical Field

[0001] The present invention relates to a state protection flight control method, in particular to an aircraft angle of attack protection control method based on dynamic instruction correction, and belongs to the field of flight control methods. Background Art

[0002] As one of the core key technologies of aircraft, the control system is directly related to flight performance. Angle of attack protection is a key factor in ensuring aircraft flight safety. Angle of attack protection is designed to prevent excessive angles of attack during flight and to provide optimized performance for maneuvers in extreme situations. As the angle of attack increases, the lift coefficient also increases until it reaches its maximum value, which is the stall angle of attack. Pilot error, environmental changes, or system failures may cause the aircraft's angle of attack to exceed its limit, threatening flight safety and, in severe cases, causing major accidents and aviation incidents. Therefore, angle of attack protection is necessary to keep the angle of attack within safe limits.

[0003] Currently, control methods that consider angle of attack constraints are mainly designed based on barrier Lyapunov function expansions. "Robust adaptive control of hypersonic flight vehicle with asymmetric AOA constraint" (YY Guo, B. Xu, WX Han, S. Li, YP Wang, Y. Zhang, Science China Information Sciences, 2020) proposes an angle of attack constraint control method based on barrier Lyapunov functions. This method implements angle of attack constraints by simultaneously limiting both system commands and tracking errors. The controller design must satisfy both initial condition constraints and feasibility condition constraints, resulting in the initial feasible solution set being only a subset of the original constraint solution set, which is highly conservative. Furthermore, when limiting system commands and tracking errors, the limit boundary values ​​typically require multiple manual adjustments to determine. Since the total boundary value composed of the command boundary and error boundary constitutes the state constraint boundary, the allocation of each boundary value also requires multiple manual adjustments to determine, making the design process extremely complex and tedious. In actual engineering, command restriction and control restriction are often used to achieve state protection, but these restrictions are usually hard restrictions, without considering the complex functional relationship between the limit boundary and the system dynamics and state. The fixed boundary value makes the state protection very conservative, which can easily lead to over-protection of the state, making it impossible to fully utilize the flight envelope within the limit range and losing the maneuverability of the aircraft. Therefore, it is urgent to carry out research on new constraint control methods to break through the conservatism of existing methods. Summary of the Invention

[0004] The present invention aims to provide an aircraft angle of attack protection control method based on dynamic command correction to address the angle of attack constraint problem during aircraft tracking control. To ensure aircraft safety, the flight angle of attack must be constrained within a safety boundary. To achieve angle of attack protection while fully utilizing the angle of attack within the envelope, a method for aircraft angle of attack protection control based on dynamic command correction is proposed. First, an adaptive controller is constructed based on a backstepping framework to stabilize the system. The virtual control variable angle of attack command in the controller design is constrained and limited using a saturation function, and the constraint boundary is obtained by a subsequently designed protection system. Then, the angle of attack protection system is established. A neural network is used to construct a complex functional relationship between the angle of attack command and the system dynamics and state. An angle of attack prediction model is designed, and the angle of attack command limit boundary is inferred by inverse analysis of the system state response. Finally, the angle of attack command boundary in the controller design is dynamically corrected based on the angle of attack command boundary value predicted by the protection system, thereby achieving angle of attack protection. This controller utilizes dynamic command correction to fully utilize the angle of attack within the envelope, improving aircraft maneuverability. Furthermore, the controller design does not require meeting initial value conditions or feasibility constraints, simplifying the design process.

[0005] The present invention solves the technical problem by adopting a technical solution: a method for controlling the angle of attack of an aircraft based on dynamic instruction correction, which is implemented by the following steps:

[0006] (a) Consider the following nonlinear system with external disturbances:

[0007]

[0008] in, x i ∈R represents the system state; u∈R represents the system input; y∈R represents the system output; Express about Smooth function of ; Denotes the control gain function. The disturbance d i (i=1,2,3) is assumed to be bounded, satisfying |d i |<d im For ease of analysis, it is assumed here that x2 is the system angle of attack.

[0009] (b) Design the controller and angle of attack protection system according to equation (1).

[0010] Step 1: Define tracking error as e1 = x1 - y d ,y d Indicates the expected output command. Define the virtual angle of attack command as x 2c , the corrected angle of attack command is x 2m , then the angle of attack correction error is c2 = x 2m -x2c , the virtual angle of attack command after passing through the command filter is x 2d , define the filtering error b2 = x 2d -x 2m To compensate for the limited influence of the instruction, the error z1 = e1-z is defined 10 , z 10 Is the compensation signal, and the design update law will be given in the subsequent design. Design virtual control quantity x 2c for

[0011]

[0012] Wherein, k1 and μ1 both represent positive constants; Indicates d 1m An estimate of ; the integral term is defined as ω0 represents any positive constant.

[0013] Design protection systems for Directive x 2c Limitations and corrections are made to ensure that the angle of attack is within the constraint range. The angle of attack prediction model based on neural network and linear dynamic compensator is constructed as follows:

[0014]

[0015] in, a c and b c Represents a positive constant; K f represents a positive constant; model error Using neural network approximation, that is Represents the ideal weight W of the neural network f The estimated value of θ f (x f )∈R N represents the neural network basis function, x f Represents the input vector of the neural network. The upper and lower boundary values ​​of the angle of attack command are reversed by the prediction model (3)

[0016]

[0017] in, and Represent the upper and lower boundary values ​​of the angle of attack prediction respectively.

[0018] The original angle of attack instruction (2) is corrected based on the angle of attack instruction limit boundary value (4), and the corrected virtual control instruction is

[0019]

[0020] in, x 2m This is the revised instruction.

[0021] Design the following instruction filter for the virtual instruction x 2m Filter

[0022]

[0023] Among them, a2 is a positive constant, x 2d is the filtered signal.

[0024] Design compensation signal z 10 Compensation command limited impact, compensation signal z 10 Designed for

[0025]

[0026] The design parameter update law is

[0027]

[0028] Among them, m1>0, Γ>0 and π1>0 represent positive constants to be designed.

[0029] Step 2: Define tracking error e2 = x2 - x 2d , define the error z2=e2-z 20 , z 20 Is the compensation signal of step 2, and the design update law will be given in the subsequent design. Design virtual control quantity x 3c for

[0030]

[0031] Among them, k2 and μ2 both represent positive constants to be designed; Indicates d 2m The error integral term is defined as

[0032] Design the following instruction filter for instruction x 3c Filter

[0033]

[0034] Among them, a3 represents a positive constant, x 3d is the filtered signal.

[0035] Design compensation signal z 20 for

[0036]

[0037] Where c3 = x 3d -x 3c , z 30is the compensation signal in step 3.

[0038] Design the adaptive update law as

[0039]

[0040] Wherein, m2 and π2 represent positive constants.

[0041] Step i: Define the tracking error e i =x i -x id , define the error z i =e i -z i0 , z i0 is the compensation signal of step i, and the update law is given in the subsequent design. Design the virtual control quantity x (i+1)c for

[0042]

[0043] Among them, k i and μ i All represent positive constants to be designed; Indicates d im The estimated value of ; the error integral term is defined as

[0044] Design the following instruction filter for instruction x (i+1)c Filter

[0045]

[0046] Among them, a i+1 represents a positive constant, x (i+1)d is the filtered signal.

[0047] Design compensation signal z i0 for

[0048]

[0049] Among them, c i+1 =x (i+1)d -x (i+1)c , z (i+1)0 is the compensation signal of the i+1th step.

[0050] Design the adaptive update law as

[0051]

[0052] Among them, m i and π i Represents a positive constant.

[0053] Step n: Define the tracking error en =x n -x nd , the actual control quantity u is designed to be

[0054]

[0055] Among them, k n and μ n All represent positive constants to be designed; Indicates d nm estimated value.

[0056] The design parameter update law is

[0057]

[0058] Among them, π n and m n Represents the positive constant to be designed.

[0059] (c) According to the virtual control quantity x obtained in (b) ic (i=2,…,n) and the actual control variable u, and return the obtained control variable to model (1) to achieve command tracking under angle of attack protection.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] (1) The present invention realizes angle of attack protection by means of dynamic instruction correction. The instruction boundary value is related to the system dynamics and changes with the change of state quantity, which can make fuller use of the remaining angle of attack within the envelope.

[0062] (2) The present invention does not impose strict restrictions on the initial value of the angle of attack and the initial value of the angle of attack command. Even if the initial angle of attack is outside the boundary value, the proposed control algorithm is still effective.

[0063] (3) The angle of attack protection control method designed in the present invention consists of two parts: a protection system and a flight control system. The protection system is an independent module outside the flight control system and interacts with it through angle of attack commands. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of an aircraft angle of attack protection control method based on dynamic instruction correction of the present invention. DETAILED DESCRIPTION

[0065] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0066] The flow chart of the aircraft angle of attack protection control method based on dynamic instruction correction of the present invention is as follows: Figure 1 As shown, the invention is used to achieve the problem of stable tracking control of the system under limited angle of attack, and its specific implementation steps are as follows:

[0067] (a) Consider the aircraft longitudinal dynamics model as follows:

[0068]

[0069]

[0070] where x = [V, h, γ, α, q] T Represents the aircraft state vector, where each element represents speed, altitude, track angle, angle of attack, and pitch velocity; m, I yy and g represent the mass of the aircraft, the moment of inertia about the y-axis, and the acceleration due to gravity, respectively; D, L, M yy and T represent drag, lift, pitching moment and engine thrust respectively, and their expressions are as follows

[0071]

[0072] Among them, ρ h represents the air density; S represents the wing area; represents the mean aerodynamic chord length; Indicates dynamic pressure; C L , C D and C M denote the total aerodynamic lift coefficient, drag coefficient, and pitching moment coefficient, respectively; represents the aerodynamic derivative; β i ,i=1,…,8 represents the thrust coefficient; z T Represents the thrust arm; u=[Φ,δ e ] T is the input variable, Φ represents the fuel equivalence ratio, δ e Indicates the elevator deflection angle.

[0073] In the controller design, only the attitude subsystem is designed. The speed and altitude are designed using PID, which will not be described here. Define x1 = γ, x2 = α, x3 = q. According to equations (21)-(23), the aircraft attitude subsystem can be converted into the following strict feedback form:

[0074]

[0075] in, y∈R represents output; u∈R represents input; represents an unknown nonlinear function; represents the control gain function, and The expression is:

[0076] (b) Design an angle of attack protection controller based on dynamic command optimization according to formula (24).

[0077] Step 1: Define tracking error as e1 = x1 - y d ,y d Indicates the expected output command. Define the virtual angle of attack command as x 2c , the corrected angle of attack command is x 2m , then the angle of attack correction error is c2 = x 2m -x 2c , the virtual angle of attack command after passing through the command filter is x 2d , define the filtering error b2 = x 2d -x 2m In order to compensate for the influence of instruction limitation on the system, the error z1 = e1-z 10 , z 10 Is the compensation signal, and the design update law will be given in the subsequent design. The virtual control quantity x of the design angle of attack 2c for

[0078]

[0079] Wherein, k1 and μ1 both represent positive constants; Indicates d 1m The estimated value of ; the error integral term is

[0080] The angle of attack prediction model in closed-loop state based on neural network and linear dynamic compensator is:

[0081]

[0082] in, a c and b c Represents a positive constant; K f represents a positive constant; model error Using neural network approximation, that is Represents the ideal weight W of the neural network f The estimated value of θ f (x f )∈R N represents the neural network basis function, x f Represents the input vector of the neural network. According to the prediction model (3), the upper limit value of the angle of attack instruction can be deduced and the lower boundary value for

[0083]

[0084] in, and They represent the upper and lower boundary values ​​of the angle of attack respectively.

[0085] Based on the angle of attack instruction boundary value (4), the original angle of attack virtual control instruction (25) is corrected, and the corrected virtual control instruction is

[0086]

[0087] in, x 2m This is the revised instruction.

[0088] Design compensation signal z 10 Compensation command limited impact, compensation signal z 10 Designed for

[0089]

[0090] Where c2 = x 2m -x 2c .

[0091] The design parameter update law is

[0092]

[0093] Among them, m1>0, π1>0, Γ>0 and Represents the positive constant to be designed.

[0094] Step 2: Define tracking error e2 = x2 - x 2d , the virtual control quantity x of the design angle of attack 3c for

[0095]

[0096] Wherein, k2 and μ2 are both positive constants; Indicates d 2m The estimated value of ; the error integral term is

[0097] Design an instruction filter for instruction x 3c Filter

[0098]

[0099] Wherein, a3>0 represents a positive constant to be designed.

[0100] Design compensation signal z 10 Compensation command limited impact, compensation signal z 10 Designed for

[0101]

[0102] Where c3 = x 3d -x 3c , z 30 Indicates the compensation signal.

[0103] The design parameter update law is

[0104]

[0105] Wherein, m2 and π2 represent the positive constants to be designed.

[0106] Step 3: Define tracking error e3 = x3 - x 3d , z3=e3-z 30 , the actual control quantity u is designed to be

[0107]

[0108] Among them, k3 and μ3 both represent positive constants; Indicates d 3m The estimated value of ; the error integral term is

[0109] The design parameter update law is

[0110]

[0111] Here, m3>0 and π3>0 represent positive constants to be designed.

[0112] (c) According to the virtual control input x obtained in (b) 2c , x 3c and the actual control input u, and return the control input to the model (19)-(23) to achieve system instruction tracking control.

[0113] Parts of the present invention that are not described in detail belong to common knowledge among those skilled in the art.

Claims

1. A method for controlling an aircraft angle of attack protection based on dynamic correction of instructions is implemented by the following steps: (a) Consider the following nonlinear system with external disturbances: in, x i ∈R represents the system state; u∈R represents the system input; y∈R represents the system output; Express about Smooth function of ; Denotes the control gain function. The disturbance d i (i=1,2,3) is assumed to be bounded, satisfying |d i |<d im For ease of analysis, it is assumed here that x2 is the system angle of attack. (b) According to equation (1), design the controller and angle of attack protection system. Step 1: Define tracking error as e1 = x1 - y d ,y d Indicates the expected output command. Define the virtual angle of attack command as x 2c , the corrected angle of attack command is x 2m , then the angle of attack correction error is c2 = x 2m -x 2c , the virtual angle of attack command after passing through the command filter is x 2d , define the filtering error b2 = x 2d -x 2m To compensate for the limited influence of the instruction, the error z1 = e1-z is defined 10 , z 10 Is the compensation signal, and the design update law will be given in the subsequent design. Design virtual control quantity x 2c for Wherein, k1 and μ1 both represent positive constants; Indicates d 1m An estimate of ; the integral term is defined as ω0 represents any positive constant. Design protection systems for Directive x 2c Limitations and corrections are made to ensure that the angle of attack is within the constraint range. The angle of attack prediction model based on neural network and linear dynamic compensator is constructed as follows: in, a c and b c Represents a positive constant; K f represents a positive constant; model error Using neural network approximation, that is Represents the ideal weight W of the neural network f The estimated value of θ f (x f )∈R N represents the neural network basis function, x f Represents the input vector of the neural network. The upper and lower boundary values ​​of the angle of attack command are reversed by the prediction model (3) in, and Respectively represent the upper and lower boundary values ​​of the angle of attack prediction. The original angle of attack instruction (2) is corrected based on the angle of attack instruction limit boundary value (4), and the corrected virtual control instruction is in, x 2m This is the revised instruction. Design the following instruction filter for the virtual instruction x 2m Filter Among them, a2 is a positive constant, x 2d is the filtered signal. Design compensation signal z 10 Compensation command limited impact, compensation signal z 10 Designed for The design parameter update law is Among them, m1>0, Γ>0 and π1>0 represent positive constants to be designed. Step 2: Define tracking error e2 = x2 - x 2d , define the error z2=e2-z 20 , z 20 Is the compensation signal of step 2, and the design update law will be given in the subsequent design. Design virtual control quantity x 3c for Among them, k2 and μ2 both represent positive constants to be designed; Indicates d 2m The error integral term is defined as Design the following instruction filter for instruction x 3c Filter Among them, a3 represents a positive constant, x 3d is the filtered signal. Design compensation signal z 20 for Where c3 = x 3d -x 3c , z 30 is the compensation signal in step 3. Design the adaptive update law as Where m2 and π2 represent positive constants. Step i: Define the tracking error e i =x i -x id , define the error z i =e i -z i0 , z i0 is the compensation signal of step i, and the update law is given in the subsequent design. Design the virtual control quantity x (i+1)c for Among them, k i and μ i All represent positive constants to be designed; Indicates d im The estimated value of ; the error integral term is defined as Design the following instruction filter for instruction x (i+1)c Filter Among them, a i+1 represents a positive constant, x (i+1)d is the filtered signal. Design compensation signal z i0 for Among them, c i+1 =x (i+1)d -x (i+1)c , z (i+1)0 is the compensation signal of the i+1th step. Design the adaptive update law as Among them, m i and π i Represents a positive constant. Step n: Define the tracking error e n =x n -x nd , the actual control quantity u is designed to be Among them, k n and μ n All represent positive constants to be designed; Indicates d nm estimated value. The design parameter update law is Among them, π n and m n Represents the positive constant to be designed. (c) (c) According to the virtual control quantity x obtained in (b) ic (i=2,…,n) and the actual control variable u, and return the obtained control variable to model (1) to achieve command tracking under angle of attack protection.

Images