Variable configuration aircraft anti-interference attitude control method based on aerodynamic parameter estimation

Abstract

The invention relates to an aerodynamic parameter estimation-based variable configuration aircraft anti-interference attitude control method, which comprises the following steps of: establishing a variable configuration aircraft attitude motion model containing unknown nonlinear dynamic, aerodynamic parameter uncertainty, deformation interference and external interference information, and converting the variable configuration aircraft attitude motion model into a second-order affine nonlinear form to obtain an attitude dynamics model containing total interference; performing aerodynamic parameter estimation on the attitude dynamics model based on a GRU neural network to obtain a variable configuration aircraft attitude dynamics equation; designing a disturbance observer based on a Koopman operator to estimate unknown nonlinear dynamics, external disturbance and deformation disturbance to obtain a disturbance estimation value; and based on the obtained aerodynamic parameter identification quantity and the disturbance estimation value, designing an attitude control algorithm based on improved finite time sliding mode control, and completing the design of the variable configuration aircraft attitude controller under the anti-interference condition. Rapid estimation and compensation of complex interference are realized, and autonomy, accuracy and anti-interference capability of a variable configuration control process of the aircraft are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of aircraft attitude control, and relates to an anti-interference attitude control method for variable configuration aircraft based on aerodynamic parameter estimation. Background Art

[0002] Variable configuration aircraft exhibit characteristics such as strong uncertainty, strong interference, strong coupling, strong nonlinearity, fast time-variation, and multiple constraints. Due to technical limitations, it is difficult to simulate the high Mach flight environment in ground wind tunnel experiments, resulting in large deviations in aerodynamic parameters. In addition, the variable configuration aircraft system exhibits strong coupling, strong nonlinearity, and multimodal characteristics, making it difficult to achieve precise control; internal and external disturbances during the deformation process are likely to cause serious uncertainties, leading to divergence of the closed-loop system. These characteristics greatly increase the difficulty of designing and analyzing the attitude control system of variable configuration aircraft, and traditional aircraft attitude control methods are difficult to directly apply. Therefore, it is important to design a high-precision attitude control method for variable configuration aircraft under the full flight profile in combination with existing technologies, which has broad application prospects.

[0003] Currently, research on attitude control methods for variable configuration aircraft under strong interference at home and abroad is mainly divided into two categories. One is to use common classical control theories such as sliding mode control and adaptive control to design high-performance robust controllers; the other is to combine optimal control and reinforcement learning methods to design intelligent controllers for online learning. However, these methods are difficult to solve the problem of precise attitude control of variable configuration aircraft based on aerodynamic parameter estimation under strong interference.

[0004] In summary, existing methods lack high-precision attitude control methods under multi-source interference of nonlinear dynamics, uncertain aerodynamic parameters, and external interference, and it is urgent to overcome the anti-interference attitude control technology of variable configuration aircraft in complex flight tasks. Summary of the Invention

[0005] The technical problem solved by the present invention is: aiming at the attitude control problem of variable configuration aircraft with unknown nonlinear dynamics, uncertain aerodynamic parameters, deformation interference, and external interference, an anti-interference attitude control method for variable configuration aircraft based on aerodynamic parameter estimation is proposed, which realizes the rapid estimation and compensation of complex interference, adopts a set of control strategies to track the attitude reference signal given by offline trajectory planning, and ensures certain dynamic performance of the controller, improving the autonomy, accuracy, and anti-interference ability of the aircraft variable configuration control process.

[0006] The solution of the present invention to solve the technical problem is: to propose an anti-interference attitude control method for variable configuration aircraft based on aerodynamic parameter estimation, including the following steps:

[0007] A variable-configuration aircraft attitude motion model containing unknown non-linear dynamic information, uncertain aerodynamic parameters, deformation disturbances, and external disturbance information is established and transformed into a second-order affine non-linear form. The uncertain aerodynamic parameters, external disturbances, deformation disturbances, and unknown non-linear dynamic information are called total disturbances, and an attitude dynamics model containing total disturbances is obtained.

[0008] Based on the GRU neural network, the aerodynamic parameters of the attitude dynamics model containing total disturbances are estimated to obtain the aerodynamic parameter identification quantity, reduce the influence of uncertain aerodynamic parameters in the model, and obtain a more accurate attitude dynamics equation of the variable-configuration aircraft.

[0009] Design an interference observer based on the Koopman operator to estimate the unknown non-linear dynamic information, external disturbances, and deformation disturbances contained in the attitude dynamics equation of the variable-configuration aircraft to obtain the disturbance estimation value.

[0010] Based on the obtained aerodynamic parameter identification quantity and disturbance estimation value, design an attitude control algorithm based on improved finite-time sliding mode control to complete the design of the attitude controller of the variable-configuration aircraft under anti-interference.

[0011] Furthermore, the variable-configuration aircraft attitude motion model is a six-degree-of-freedom model, where the six degrees of freedom are respectively composed of three degrees of freedom of the centroid translational motion and three degrees of freedom of the rotation around the centroid, and thus decomposed into a centroid motion model and an aircraft rotation motion model around the centroid.

[0012] Furthermore, the attitude dynamics model containing total disturbances is specifically:

[0013] (1) The centroid motion model represents the relevant state variables of the centroid motion in the form of a differential equation system:

[0014]

[0015] Among them, V is the aircraft speed, θ is the flight path inclination angle, and σ is the flight path deviation angle; are the change rates of V, θ, and σ respectively; α is the angle of attack, β is the sideslip angle, and γ v is the speed tilt angle, m is the aircraft mass, g is the acceleration due to gravity, and T is the engine thrust; D, L, and Z are the drag, lift, and lateral force respectively, and the expressions are as follows:

[0016]

[0017] Among them, ρ is the atmospheric density, S0 is the basic reference area of the aircraft without deformation, ξ is the configuration change amount of the aircraft, are the equivalent drag coefficient, lift coefficient, and lateral force coefficient respectively; Δ L , Δ D , Δ ZDenote the modeling errors of lift, drag, and side force;

[0018] (2) The rotational motion model about the center of mass is:

[0019]

[0020] where J x , J y , J z , J xy are the moments of inertia and products of inertia about the roll, yaw, and pitch axes respectively, and ω x , ω y , ω z are the angular velocities of the roll, yaw, and pitch attitudes respectively; ψ and γ are the pitch angle, yaw angle, and roll angle respectively; The aerodynamic moments are divided into the roll moment M x , the yaw moment M y , and the pitch moment M z , and their expressions are respectively:

[0021]

[0022] where b is the wingspan and l is the mean aerodynamic chord of the wing; are the equivalent roll, yaw, and pitch moment coefficients respectively; Denote the modeling errors of the roll moment, pitch moment, and yaw moment;

[0023] The aerodynamic force and aerodynamic moment coefficient models are as follows:

[0024]

[0025] In the formula, S(ξ) represents the actual reference area when there is deformation; C D (ξ), C L (ξ), C Z (ξ) are the drag coefficient, lift coefficient, and side force coefficient respectively, and m x (ξ), m y (ξ), m z (ξ) are the roll moment coefficient, yaw moment coefficient, and pitch moment coefficient respectively, and their calculation methods are as follows:

[0026]

[0027] where Ma is the Mach number; δ x , δ y , δ z are the roll, yaw, and pitch rudder deflection angles respectively; are the polynomial model coefficients;

[0028] (3) Model the attitude motion into the state - space form and organize to obtain:

[0029]

[0030] Wherein, Δd represents unknown internal and external disturbances; ω = [ω x , ω y , ω z T , u = [M cx , M cy , M cz T , M cx , M cy , M cz are the control force and torque of the roll, yaw, and pitch channels; the expressions of matrices R1, Φ, and J are respectively:

[0031]

[0032] (4) Let x = Ω and take the second - derivative. After linear transformation, it is transformed into the integral - cascade form as follows:

[0033]

[0034] Wherein, g(x, ξ) = R1J -1 , d(t) = R1Δd;

[0035] (5) Conduct attitude command conversion. The conversion method is:

[0036]

[0037] Wherein, M1[·], M2[·], M3[·] are elementary transformation matrices from the ground coordinate system to the body coordinate system. M1[γ] represents the elementary transformation matrix of rotating by γ angle around the x - axis. The subscript numbers 1, 2, and 3 respectively represent different coordinate axes;

[0038] Thus, the pitch - angle command the yaw - angle command ψ c and the roll - angle command γ c are calculated as:

[0039]

[0040] Wherein, B Gij represents the element in the i - th row and j - th column of the matrix ;

[0041] Expand the above formula and considering ψ c and γ​​c The possible quadrants result in the following relational expressions:

[0042]

[0043] When β c , γ Vc and σ are all small quantities, the above equation simplifies to:

[0044]

[0045] Furthermore, based on the GRU neural network, the aerodynamic parameter estimation is performed on the attitude dynamics model including the total interference, specifically:

[0046] The output of the GRU neural network is expressed as The estimated values of the corresponding aerodynamic parameters C D (ξ), C L (ξ), C Z (ξ), m x (ξ), m y (ξ), m z (ξ); the update gate z t and the reset gate r t are expressed as:

[0047] z t = σ(w z ·[h t-1 , x t + b z )

[0048] r t = σ(w r ·[h t-1 , x t + b r )

[0049] Where:

[0050]

[0051] In the above equation, x t , h t-1 are the input data of the input layer, x t is the sequence information composed of the three-axis attitude angles of the aircraft and the current dynamic pressure, and h t-1 is the output of the neuron at the previous moment; is the final output; σ represents the sigmoid function; w z , w r , represent the weight coefficient matrices; b z , b r , represents the offset; the calculation symbol · represents matrix multiplication; the calculation symbol ⊙ represents element-wise multiplication; represents the candidate hidden layer state, which is used to calculate the internal state of the neuron at the current moment;

[0052] The training method of the GRU neural network is based on the adaptive moment estimation algorithm. The CFD simulation results are used as the training labels of the aerodynamic force and aerodynamic moment prediction network model, and the mean square error is selected as the loss function. The expression is:

[0053]

[0054] In the formula, N is the amount of data for a single training; C i is the prediction value of the GRU neural network; is the aerodynamic parameter obtained by CFD simulation, which is used as the true value in model training; the weights in the network are continuously updated through training until the loss function or the number of training times reaches the preset value.

[0055] Furthermore, a disturbance observer based on the Koopman operator is designed, specifically:

[0056]

[0057] where ζ is an auxiliary variable, represents the disturbance prediction value, L represents the observer gain; Φ(d(t)) is the lifted space, is the real-valued observation function; K is the finite-dimensional approximation of the Koopman operator; t s is the step size;

[0058] Furthermore, the attitude controller of the variable configuration aircraft under anti-disturbance is designed as follows:

[0059]

[0060] where U is the control law, s represents the sliding surface, represents the disturbance prediction value; e = x c -x represents the error between the actual signal and the desired signal x c of, represents the first derivative of the error between the actual signal and the desired signal; represents the second derivative of the desired signal; k1, k2, k3 are control parameter matrices, which determine the convergence speed of the system when it is far from the equilibrium point; a1, a2 represent constants greater than zero.

[0061] The beneficial effects of the present invention compared with the prior art are:

[0062] A method for anti - interference attitude control of variable - configuration aircraft based on pneumatic parameter estimation is proposed in the present invention. Aiming at the shortcomings of the existing methods that in the case of unknown non - linear dynamics, uncertain pneumatic parameters, deformation interference and external interference, the attitude controller of variable - configuration aircraft lacks high - precision control ability, a relatively accurate attitude motion model of variable - configuration aircraft is obtained by predicting pneumatic parameters based on GRU neural network; a disturbance observer based on Koopman operator is designed to estimate the unknown internal and external disturbances of the system; aiming at the actuator fault situation, a control method based on improved finite - time sliding - mode algorithm is designed to complete the design of the attitude control system of variable - configuration aircraft under anti - interference, which has the characteristics of high precision and strong robustness, and is applicable to the design of the attitude control system for large - scale continuous variable - configuration aircraft in complex environments. Brief Description of the Drawings

[0063] Figure 1 It is a flowchart of the anti - interference attitude control method for variable - configuration aircraft based on pneumatic parameter estimation. Detailed Embodiments

[0064] The present invention will be further described below with reference to the drawings and embodiments.

[0065] As Figure 1 shown, a method for anti - interference attitude control of variable - configuration aircraft based on pneumatic parameter estimation proposed in the present invention includes the following steps:

[0066] Step 1: Establish an attitude motion model of variable - configuration aircraft containing unknown non - linear dynamic information, uncertain pneumatic parameters, deformation interference and external interference information, and transform it into a second - order affine non - linear form. The uncertain pneumatic parameters, external interference, deformation interference and unknown non - linear dynamic information are called total disturbances, and an attitude dynamics model containing total disturbances is obtained.

[0067] Step 2: Based on GRU neural network, estimate the pneumatic parameters of the attitude dynamics model containing total disturbances to obtain pneumatic parameter identification quantities, reduce the influence of uncertain pneumatic parameters in the model, and obtain a relatively accurate attitude dynamics equation of variable - configuration aircraft.

[0068] Step 3: Design a disturbance observer based on Koopman operator to estimate the unknown non - linear dynamic information, external interference and deformation interference contained in the attitude dynamics equation of variable - configuration aircraft, and obtain the disturbance estimation value.

[0069] Step 4: Based on the pneumatic parameter identification quantities and disturbance estimation values obtained in Step 2 and Step 3, design an attitude control algorithm based on improved finite - time sliding - mode control to complete the design of the attitude controller of variable - configuration aircraft under anti - interference.

[0070] Embodiment 1

[0071] Step 1: Establish a variable-configuration aircraft attitude motion model containing unknown non-linear dynamic information, uncertain aerodynamic parameters, deformation disturbances, and external disturbance information, and transform it into a second-order affine non-linear form. The uncertain aerodynamic parameters, external disturbances, deformation disturbances, and unknown non-linear dynamic information are called the total disturbances, and an attitude dynamics model containing the total disturbances is obtained.

[0072] The variable-configuration aircraft attitude motion model is a six-degree-of-freedom model, where the six degrees of freedom are composed of three degrees of freedom of the center-of-mass translational motion and three degrees of freedom of the rotation about the center of mass. Therefore, it can be decomposed into a center-of-mass motion model and an aircraft rotation model about the center of mass.

[0073] (1) Center-of-mass motion model. Represent the relevant state variables of the center-of-mass motion in the form of a differential equation system:

[0074]

[0075] where V is the aircraft speed, with an initial value of 2720 m / s; θ is the flight-path angle, with an initial value of -0.035 rad; σ is the track deviation angle, with an initial value of 0 rad; are the change rates of V, θ, and σ respectively; α is the angle of attack, β is the sideslip angle, and γ v is the velocity inclination angle, with initial values of -0.02 rad, 0 rad, and 0 rad respectively; m is the aircraft mass, with an initial value of 960 kg; g is the acceleration due to gravity, with a value of 9.8 m / s 2 ; T is the engine thrust; D, L, and Z are the drag, lift, and side force respectively, and their expressions are as follows:

[0076]

[0077] where ρ is the atmospheric density, with a value of 1.225 kg / m3; S0 is the basic reference area of the aircraft without deformation, and its value is 0.2 m 2 ; are the equivalent drag coefficient, lift coefficient, and side force coefficient respectively, ξ is the aircraft deformation; Δ L , Δ D , Δ Z represent the modeling errors of the lift, drag, and side force.

[0078] (2) The model of the aircraft rotation about the center of mass is:

[0079]

[0080] In the formula, J x , J y , J z , J xy are the roll, yaw, pitch axis moments of inertia and products of inertia respectively, ωx , ω y , ω z are the roll, yaw, and pitch attitude angular velocities, respectively. ψ and γ are the pitch angle, yaw angle, and roll angle, respectively. The aerodynamic moments are divided into the roll moment M x , the yaw moment M y , and the pitch moment M z , and their expressions are as follows:

[0081]

[0082] where b is the wingspan and l is the mean aerodynamic chord of the wing. are the equivalent roll, yaw, and pitch moment coefficients, respectively. represents the modeling errors of the roll moment, pitch moment, and yaw moment.

[0083] The aerodynamic force and aerodynamic moment coefficient models are as follows:

[0084]

[0085] In the formula, S(ξ) represents the actual reference area when there is deformation; C L (ξ), C D (ξ), C Z (ξ) are the lift coefficient, drag coefficient, and side force coefficient, respectively; m x (ξ), m y (ξ), m z (ξ) are the roll moment coefficient, yaw moment coefficient, and pitch moment coefficient, respectively, and their calculation methods are as follows:

[0086]

[0087] where Ma is the Mach number, and its initial value is 8; δ x , δ y , δ z are the roll, yaw, and pitch rudder deflections, i.e., the control variables. are the polynomial model coefficients.

[0088] (3) Convert the above attitude motion model into the state - space form. The aerodynamic parameter uncertainties, external disturbances, and unknown nonlinear dynamic information are called the total disturbance, and we get:

[0089]

[0090] In the formula, ω = [ω x , ω y , ω z T , u = [M cx ​,M cy ,M cz T ,M cx ,M cy ,M cz are the control torques for the roll, yaw, and pitch channels respectively; Δd represents unknown internal and external disturbances; the expressions of matrices R1, Φ, and J are respectively

[0091]

[0092] (4) Take the second derivative of the state variables in the above model, and after linear transformation, it is converted into an integral cascade form as follows:

[0093]

[0094] where g(x, ξ) = R1J -1 , and d(t) = R1Δd.

[0095] (5) Perform attitude command conversion; generally, the guidance commands during attitude control are usually the angle of attack α, the sideslip angle β, and the bank angle γ V , since the control variables used in the above high-order attitude model of the aircraft are the pitch angle the yaw angle ψ, and the roll angle γ, so the guidance commands [α c β c γ Vc T need to be converted into The conversion method is:

[0096]

[0097] In the formula, M1[·], M2[·], M3[·] are the elementary transformation matrices from the ground coordinate system to the body coordinate system. M1[γ] represents the elementary transformation matrix of rotating by γ angle around the x-axis, where the subscript numbers 1, 2, and 3 represent different coordinate axes respectively. The meanings of M2[ψ] and are the same by analogy.

[0098] Thus, the pitch angle command the yaw angle command ψ c and the roll angle command γ c can be calculated as follows:

[0099]

[0100] In the formula, B Gij represents the element in the i-th row and j-th column of the matrix .

[0101] ​​Expand the above formula and considering ψ c and γ c the possible quadrants they may be in, the following relational expressions are obtained

[0102]

[0103] When β c , γ Vc and σ are all small quantities, the above formula can be simplified to:

[0104]

[0105] Step 2: Based on the GRU neural network, perform aerodynamic parameter estimation on the attitude dynamics model containing total interference to obtain the aerodynamic parameter identification quantity, reduce the influence of aerodynamic parameter uncertainty in the model, and obtain a more accurate attitude dynamics equation of the variable configuration aircraft.

[0106] Collect sufficient flight state and attitude data in the standard flight trajectory of the variable configuration of the aircraft, organize the data, and generate an input database for each moment.

[0107] During training, the input layer inputs data x t , h t-1 , x t is the sequence information composed of the three-axis attitude angles of the aircraft and the current flight dynamic pressure, and h t-1 is the output of the neuron at the previous moment. The output of the GRU neural network is expressed as That is, the estimated value of the aerodynamic parameter. The expression of the aerodynamic parameter estimation method based on the GRU neural network is:

[0108] z t =σ(w z ·[h t-1 , x t +b z )

[0109] r t =σ(w r ·[h t-1 , x t +b r )

[0110]

[0111] In the above formula, z t , r t are the update gate and the reset gate respectively; σ represents the sigmoid function; w z , w r , represent the weight coefficient matrices; b z , br , represents the offset; the calculation symbol · represents matrix multiplication; the calculation symbol represents element-wise multiplication; represents the candidate hidden layer state and is used to calculate the internal state of the neuron at the current moment.

[0112] Considering the real-time prediction requirements, after testing, the established prediction model adopts a two-layer GRU structure, with the first layer containing 96 nodes and the second layer containing 64 nodes.

[0113] The GRU network is trained using the Adaptive Moment Estimation (Adam) algorithm. The learning rate is set to 0.001 at the beginning of training to accelerate the convergence speed of the deep network in the initial stage of training; during the training process, the learning rate gradually decreases to 1×10 -5 to obtain a higher prediction accuracy.

[0114] The CFD simulation results are used as the training labels for the aerodynamic force and aerodynamic moment coefficient prediction network model, and the mean squared error is selected as the loss function, and its expression is:

[0115]

[0116] In the formula, N is the amount of data for a single training; C i is the predicted value of the GRU neural network; is the aerodynamic coefficient obtained from the CFD simulation and is used as the true value in the model training. The weights in the network are continuously updated through training until the loss function or the number of training times reaches the preset value.

[0117] Step 3: Design an interference observer based on the Koopman operator to estimate the unknown nonlinear dynamic information, external interference, and deformation interference contained in the attitude dynamics equation of the variable configuration aircraft, and obtain the disturbance estimation value.

[0118] Assume that the nonlinear model of the unknown interference d(t) is:

[0119]

[0120] In the formula, represents the known and unknown external interference vectors.

[0121] For the real-valued observation function there is:

[0122]

[0123] In the formula, t0 is the initial time; is the function composition operator; is the Koopman operator.

[0124] Assume that the lifted space is:

[0125]

[0126] Calculate the finite-dimensional projection on the finite-dimensional subspace of all observable values using the Extended Dynamic Mode Decomposition (EDMD) algorithm, and determine the finite-dimensional approximation of the Koopman operator through linear regression applied to the observed data, and obtain the finite-dimensional approximation K of:

[0127]

[0128] where M is the total number of sampling points.

[0129] The solution of the above equation can be expressed as:

[0130]

[0131] Then, the linearized equation of the nonlinear interference model in the lifted space is:

[0132]

[0133] where r1 is the residual term; t s is the step size;

[0134] Design the observer model of the unknown interference d(t) as follows:

[0135]

[0136] Then, the final model of the interference observer based on the Koopman operator is as follows:

[0137]

[0138] where ζ is the auxiliary variable, represents the interference prediction value, L represents the observer gain, L = -3I 3×3 ; Φ is the lifting function, set as Φ(d(t)) = [d1, d2, d3] T .

[0139] Step 4: Based on the aerodynamic parameter identification quantity and the disturbance estimation value obtained in Step 2 and Step 3, design an attitude control algorithm based on improved finite-time sliding mode control to complete the design of the attitude controller of the variable configuration aircraft under anti-interference.

[0140] The design of the attitude controller of the variable configuration aircraft based on improved finite-time sliding mode control is as follows:

[0141]

[0142] Among them, s represents the sliding mode surface, and U is the control law, represents the predicted value of the disturbance; e = x c -x represents the error between the actual signal and the desired signal x c of, represents the first-order derivative of the error between the actual signal and the desired signal; represents the second-order derivative of the desired signal; k1, k2, k3 are control parameter matrices, which determine the convergence speed when the system is far from the equilibrium point. k1 = diag(0.7, 0.6, 0.5), k2 = diag(0.7, 0.6, 0.5), k3 = diag(0.6, 0.5, 0.35); a1, a2 represent constants greater than zero, a1 = diag(1.5, 1.5, 1.5), a2 = diag(1.2, 1.2, 1.2).

[0143] Using the method of the present invention for full-profile control, a set of attitude control strategies can be used during the large-scale cross-domain variable configuration flight of a variable configuration aircraft, and good anti-disturbance performance and dynamic performance of the controller can be maintained, without steady-state error. At the same time, compared with a controller without disturbance estimation and compensation, it can withstand greater aerodynamic parameter uncertainties and deformation disturbance amounts, and the disturbance estimation error is stable within 2 s, meeting the requirements of high precision and strong robustness.

[0144] The present application provides a computer-readable storage medium, and the computer-readable storage medium stores computer instructions. When the computer instructions run on a computer, the computer is made to execute Figure 1 the method described above.

[0145] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories and optical memories, etc.) containing computer-usable program codes.

[0146] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be realized by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for realizing in the process Figure 1one or more processes and / or blocks Figure 1 means for the functions specified in one or more blocks.

[0147] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device, and the instruction device implements the process Figure 1 one or more processes and / or blocks Figure 1 the functions specified in one or more blocks.

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Therefore, the instructions executed on the computer or other programmable device provide for implementing the process Figure 1 one or more processes and / or blocks Figure 1 the steps of the functions specified in one or more blocks.

[0149] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to include these changes and variations.

[0150] The content not described in detail in the specification of the present invention belongs to the well-known technology of those skilled in the art.

Claims

1. A disturbance rejection attitude control method for variable configuration aircraft based on pneumatic parameter estimation, characterized in that It includes the following steps: Establish a variable-configuration aircraft attitude motion model containing unknown non-linear dynamic information, uncertain aerodynamic parameters, deformation interference, and external interference information, and transform it into a second-order affine non-linear form. The uncertain aerodynamic parameters, external interference, deformation interference, and unknown non-linear dynamic information are called the total interference, and an attitude dynamics model containing the total interference is obtained; Based on the GRU neural network, estimate the aerodynamic parameters of the attitude dynamics model containing the total interference to obtain the aerodynamic parameter identification quantity, reduce the influence of uncertain aerodynamic parameters in the model, and obtain a more accurate variable-configuration aircraft attitude dynamics equation; Design an interference observer based on the Koopman operator to estimate the unknown non-linear dynamic information, external interference, and deformation interference contained in the variable-configuration aircraft attitude dynamics equation to obtain the disturbance estimation value; Based on the obtained aerodynamic parameter identification quantity and disturbance estimation value, design an attitude control algorithm based on improved finite-time sliding mode control to complete the design of the variable-configuration aircraft attitude controller under anti-interference.

2. The anti-interference attitude control method for a variable configuration aircraft based on pneumatic parameter estimation according to claim 1, characterized in that The variable-configuration aircraft attitude motion model is a six-degree-of-freedom model, where the six degrees of freedom are respectively composed of three degrees of freedom of the center-of-mass translational motion and three degrees of freedom of the rotation around the center of mass, and thus decomposed into a center-of-mass motion model and an aircraft rotation motion model around the center of mass.

3. The anti-interference attitude control method for a variable configuration aircraft based on pneumatic parameter estimation according to claim 2, characterized in that, The attitude dynamics model containing the total interference is specifically: (1) Center-of-mass motion model, expressing the relevant state variables of the center-of-mass motion in the form of a differential equation system: where V is the aircraft speed, θ is the flight path inclination angle, and σ is the flight path deviation angle; are the rates of change of V, θ, and σ respectively; α is the angle of attack, β is the sideslip angle, and γ v is the speed inclination angle, m is the mass of the aircraft, g is the acceleration due to gravity, and T is the engine thrust; D, L, and Z are the drag, lift, and side force respectively, and the expressions are as follows: where ρ is the atmospheric density, S0 is the basic reference area of the aircraft without deformation, and ξ is the configuration change amount of the aircraft. are the equivalent drag coefficient, lift coefficient, and side force coefficient respectively; Δ L , Δ D , Δ Z represent the modeling errors of lift, drag, and side force. (2) The rotation motion model around the center of mass is: Among them, J x , J y , J z , J xy are the roll, yaw, pitch axis moments of inertia and products of inertia respectively, and ω x , ω y , ω z are the roll, yaw, pitch attitude angular velocities respectively; ψ, γ are the pitch angle, yaw angle and roll angle respectively; The aerodynamic moments are divided into the roll moment M x , the yaw moment M y , the pitch moment M z , and the expressions are respectively: where \(b\) is the wingspan and \(l\) is the mean aerodynamic chord of the wing; are the equivalent roll, yaw, and pitch moment coefficients, respectively; represent the modeling errors of the roll moment, pitch moment, and yaw moment; The aerodynamic force and aerodynamic moment coefficient models are as follows: Where, S(ξ) represents the actual reference area during deformation; C D (ξ), C L (ξ), C Z (ξ) are the drag coefficient, lift coefficient, and side force coefficient respectively, m x (ξ), m y (ξ), m z (ξ) are the rolling moment coefficient, yaw moment coefficient, and pitch moment coefficient respectively, and the calculation methods are as follows: where Ma is the Mach number; δ x , δ y , δ z are the roll, yaw, and pitch rudder deflection angles, respectively; is the polynomial model coefficient; (3) Convert the attitude motion model into a state space form and organize to obtain: where Δd represents unknown internal and external disturbances; ω = [ω x , ω y , ω z T , u = [M cx , M cy , M cz T , M cx , M cy , M cz are the control force torques of the roll, yaw, and pitch channels; the expressions of matrices R1, Φ, and J are as follows:​​ (4) Let x = Ω and take the second derivative. After linear transformation, it is converted into an integral cascade type as follows: Among them, g(x, ξ) = R1J -1 , d(t) = R1Δd; (5) Perform attitude command conversion, and the conversion method is: In the formula, M1[·], M2[·], M3[·] are elementary transformation matrices from the ground coordinate system to the body coordinate system, and M1[γ] represents the elementary transformation matrix for rotating γ angles around the x-axis. The subscript numbers 1, 2, and 3 respectively represent different coordinate axes; Thus, the pitch angle command is obtained. The yaw angle command ψ c and the roll angle command γ c are calculated as follows: where B Gij represents the element in the i-th row and j-th column of the matrix . Expanding the above equation and considering ψ c and γ c may be in the quadrants, the following relational expressions are obtained: When β c , γ Vc and σ are all small quantities, the above equation simplifies to:

4. The anti-interference attitude control method for a variable configuration aircraft based on pneumatic parameter estimation according to claim 3, characterized in that Based on the GRU neural network, estimate the aerodynamic parameters of the attitude dynamics model containing the total interference, specifically: The output of the GRU neural network is expressed as corresponding to the aerodynamic parameter C D (ξ), C L (ξ), C Z (ξ), m x (ξ), m y (ξ), m z (ξ), the estimated value; the update gate z of the network t and the reset gate r t The expressions are as follows: z t = σ(w z · [h t-1 , x t + b z ) r t = σ(w r · [h t-1 , x t + b r ) Among them: In the above formula, x t , h t-1 are the input data of the input layer, where x t is the sequence information composed of the three-axis attitude angle of the aircraft and the current dynamic pressure of flight, and h t-1 is the output of the neuron at the previous moment; is the final output; σ represents the sigmoid function; w z , w r , represent the weight coefficient matrices; b z , b r , represent the bias quantities; the calculation symbol · represents matrix multiplication; the calculation symbol ⊙ represents element-wise multiplication; represents the candidate hidden layer state, which is used to calculate the internal state of the neuron at the current moment; The training method of the GRU neural network is based on the adaptive moment estimation algorithm. Use the CFD simulation results as the training labels of the aerodynamic force and aerodynamic moment prediction network model, and select the mean square error as the loss function. The expression is: where N is the amount of data for a single training; C i is the predicted value of the GRU neural network; is the aerodynamic parameter obtained by CFD simulation and used as the true value in model training; the weights in the network are continuously updated through training until the loss function or the number of training times reaches the preset value.

5. The anti-interference attitude control method for a variable configuration aircraft based on pneumatic parameter estimation according to claim 4, characterized in that Design an interference observer based on the Koopman operator, specifically: where ζ is an auxiliary variable, represents the interference prediction value, L represents the observer gain; Φ(d(t)) is the lifted space, is a real-valued observation function; K is a finite-dimensional approximation of the Koopman operator; t s is the step size; 6. The anti-interference attitude control method for a variable configuration aircraft based on pneumatic parameter estimation according to claim 5, characterized in that, The design of the variable-configuration aircraft attitude controller under anti-interference is as follows: Among them, U is the control law, s represents the sliding mode surface, represents the interference prediction value; e = x c -x represents the error between the actual signal and the desired signal x c ; represents the first derivative of the error between the actual signal and the desired signal; represents the second derivative of the desired signal; k1, k2, k3 are control parameter matrices, which determine the convergence rate when the system is far from the equilibrium point; a1, a2 represent constants greater than zero.

7. A computer-readable storage medium storing a computer program, characterized in that, When the described computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the described processor executes the described computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images