An Integrated Active-Passive Control Method for Hypersonic Vehicles with Rigid-Elastic Coupling

Abstract

This invention belongs to the field of hypersonic vehicle control technology, specifically relating to an integrated active-passive control method for hypersonic vehicles with rigid-elastic coupling. The purpose of this invention is to achieve stable tracking control of an elastic hypersonic vehicle. The method includes constructing a longitudinal dynamic model of the elastic hypersonic vehicle; adaptively identifying the elastic vibration frequencies using a cascaded adaptive filter; designing an elastic mode filtering estimation method to provide high-precision and low-cost elastic mode state variables for subsequent active feedback controller design; then performing rigid-elastic coupling model decomposition; using active disturbance rejection control (ADRC) to ensure control performance for the rigid body subsystem; and introducing an RBF neural network for the elastic subsystem, employing sliding mode control to actively suppress elastic modes and achieve stable tracking of reference commands. This method is an integrated active-passive control approach for hypersonic vehicles with rigid-elastic coupling and has broad application prospects.

Description

Technical Field

[0001] This invention belongs to the field of hypersonic vehicle control technology, and relates to an integrated active and passive control method for hypersonic vehicles with rigid-elastic coupling. Background Technology

[0002] Hypersonic vehicles often employ lightweight materials, slender fuselage designs, and statically unstable configurations, resulting in lower natural frequencies of the fuselage structure and increased coupling between rigid body motion and structural vibration. Sensor measurements contain elastic vibration information; directly using this signal in control system design can negatively impact closed-loop system performance. Furthermore, high-bandwidth attitude controllers cause the control system's operating bandwidth to approach the natural frequencies of the vehicle's lower-order elastic modes, making it easy for control forces to excite free motion in these elastic modes. Existing control methods aim to ensure the stability of the closed-loop system in the presence of elastic vibration. Therefore, actively suppressing elastic vibration is crucial for ensuring flight safety and stable control.

[0003] The patent "Anti-interference Control Method and System for Hypersonic Vehicles" (patent, National University of Defense Technology of the Chinese People's Liberation Army, CN202311492465.9, 20250103) proposes an anti-interference control method and system for hypersonic vehicles. It constructs a control-oriented attitude system model for the hypersonic vehicle, designs a dynamic sliding surface based on the terminal function and adaptive law, and designs a fixed-time fuzzy interference observer to compensate for interference in the flight control input. This method is used for attitude control of hypersonic vehicles and exhibits fast tracking speed, high tracking accuracy, and strong robustness. However, this patent does not consider the influence of structural elasticity on control, and the significant aerodynamic servo-elasticity problem remains unresolved.

[0004] The patent "An Active Disturbance Rejection Control Method for Hypersonic Vehicles" (Patent, Lanzhou University of Technology, CN202411921922.6, 20250404) proposes an active disturbance rejection control method for hypersonic vehicles. It establishes a new linear and nonlinear switching function and designs an extended state observer for the velocity and altitude subsystems based on it. This method can effectively suppress the influence of uncertainties on the control system and enhance the robustness of the control system. However, this method treats elastic disturbances as interferences for control and does not actively suppress elastic vibrations.

[0005] The patent "Aerodynamic Servo-Elastic Active Control System and Method for Air-breathing Hypersonic Vehicles" (Patent, Civil Aviation University of China, CN202411624444.2, 20250225) proposes an aerodynamic servo-elastic active control system and method for air-breathing hypersonic vehicles. It uses an elastic frequency estimator to obtain the elastic frequency estimate of the controlled object, and combines a state disturbance joint estimator and an attitude controller to perform stable control of the hypersonic vehicle. However, this method uses the elastic frequency identification result for compensation control and does not combine rigid body passive control and elastic active suppression control, so the control effect is generally poor.

[0006] The paper "Robust Adaptive Control of Elastic Hypersonic Vehicles Combining Active and Passive Control [J]" (Journal of Astronautics, 2024, 45(7): 1052-1064) proposes a robust adaptive control method combining active and passive control to address the significant aerodynamic servo-elasticity problem caused by the lightweight materials and slender body configuration of hypersonic vehicles. This method can effectively ensure the elastic suppression and rigid body control performance of hypersonic vehicles under the influence of aerodynamic servo-elasticity. However, this method lacks corresponding measurement and estimation methods for elastic mode estimation, which will lead to an increase in the active feedback control error and affect the overall control effect.

[0007] For the control of elastic hypersonic vehicles, existing methods primarily focus on ensuring the stability of the closed-loop system in the presence of elastic vibrations, rather than actively suppressing them. Without active vibration control, disturbances or self-excitation can lead to increased elastic vibration amplitude and excessively long vibration decay times, resulting in decreased performance of onboard equipment, reduced control accuracy, and even system instability. With technological advancements, artificial intelligence plays a crucial role in improving system performance. However, current technologies fail to utilize AI to optimize control strategies, particularly under the complex flight conditions of hypersonic vehicles, lacking real-time learning mechanisms. Summary of the Invention

[0008] To address the aforementioned problems, this invention provides an integrated active and passive control method for hypersonic vehicles with rigid-elastic coupling.

[0009] The technical solution of the present invention is as follows:

[0010] An integrated active-passive control method for rigid-elastic coupled hypersonic vehicles includes: constructing a longitudinal dynamic model of the elastic hypersonic vehicle, online estimation of elastic vibration modes, decomposition of the rigid-elastic coupling model, design of an attitude active disturbance rejection controller for the elastic hypersonic vehicle, and combined active-passive vibration suppression control for rigid-elastic coupled hypersonic vehicles. Details are as follows:

[0011] Step 1: Construct a longitudinal dynamic model of an elastic hypersonic vehicle

[0012] Among all elastic modes, the first-order elastic mode has the greatest influence. As the order increases, its influence gradually decreases. The influence of the modes within the third order can almost represent the influence of all elastic modes. Ignoring the elastic modes above the third order, the longitudinal dynamic model of the elastic hypersonic vehicle is shown in Equation (1).

[0013] (1)

[0014] In the formula: These are speed, altitude, track angle, angle of attack, and pitch rate, respectively, indicated by the superscript " " represents the first derivative, with a superscript " " denotes the second derivative; For elastic modes, that is, the first three generalized coordinates of elastic vibration; and These are the damping ratio and vibration frequency of the elastic mode, respectively; These are the aircraft's mass, moment of inertia, and gravitational acceleration, respectively. and These are the generalized forces representing lift, drag, thrust, pitching moment, and elastic modes, respectively, and their specific expressions are as follows:

[0015] (2)

[0016] In the formula:

[0017] (3)

[0018] in, For dynamic pressure; This is the aerodynamic reference area; For thrust torque arm; The average aerodynamic chord length; This refers to the throttle opening. Elevator deflection; This refers to the canard deflection. , , and These are the lift coefficient, drag coefficient, thrust coefficient, and pitching moment coefficient, respectively. A vector consisting of the first three generalized coordinates and their derivatives. ; , , , , , These are the coefficients of generalized force-related aerodynamic parameters; , , , , , , , These are the thrust-related aerodynamic parameter coefficients; , , , , , These are the coefficients of aerodynamic parameters related to pitching moment; , , , , These are the lift-related aerodynamic parameter coefficients; , , , , , , , These are the drag-related aerodynamic parameter coefficients.

[0019] Consider the rate gyroscope measurement model as follows:

[0020] (4)

[0021] In the formula: The pitch angular velocity measured by the gyroscope. The pitch angle measured by the gyroscope. The pitch angular velocity produced by the rigid body. The pitch angle produced by the rigid body; , For the first Mode shapes of the first elastic modes, This indicates the installation location of the gyroscope.

[0022] Step 2: Online estimation of elastic vibration modes

[0023] For vibration frequency of Elastic mode notch filtering problem, cascaded notch filter for:

[0024] (5)

[0025] In the formula: For delay operators, This represents a delay of 2 sampling periods; For the parameters to be identified, Sampling time; To adjust the parameters.

[0026] Assuming elastic vibration frequency The identification value The notch filter obtained based on the frequency identification results is as follows:

[0027] (6)

[0028] In the formula: Cascaded notch filter The estimate; for The estimated value; This is an estimated value for the vibration frequency.

[0029] Based on equation (6), it can be deduced that the first... The first notch filter Step output for:

[0030] (7)

[0031] In the formula: ;

[0032] So, the first The first notch filter pole part Step output for:

[0033] (8)

[0034] Combining equations (7) and (8), we obtain the following formula:

[0035] (9)

[0036] To prevent frequency crossover issues during elastic vibration natural frequency estimation and to improve the accuracy and robustness of elastic vibration frequency identification, an identification performance evaluation signal is introduced. The performance evaluation signal for identifying elastic vibration frequencies is defined as follows: The output of the improved notch filter As shown below:

[0037] (10)

[0038] Define the cost function :

[0039] (11)

[0040] In the formula: Forgetting factor;

[0041] definition: , We obtain the following formula:

[0042] (12)

[0043] Will Compared to Differentiating and setting it to zero, we get the following equation:

[0044] (13)

[0045] Using the recursive least squares algorithm, equation (13) can be written in another form:

[0046] (14)

[0047] In the formula: , ;

[0048] Since we don't know whether the identification results are reasonable, we need to design a mechanism to monitor the identification performance in order to determine its quality.

[0049] (15)

[0050] The identification results of the elastic vibration frequency are as follows:

[0051] (16)

[0052] According to equation (4), the pitch angular velocity measured by the sensor is passed through a notch filter. , and Then the actual output signal of each notch filter is:

[0053] (17)

[0054] In the formula: , and These are the filtering error coefficients.

[0055] According to equation (17), we get , and The estimated value is:

[0056] (18)

[0057] Similarly, we can obtain , and The estimated value is:

[0058] (19)

[0059] In the formula: , and according to Obtain.

[0060] Step 3: Decomposition of the rigid-elastic coupling model

[0061] Structural elasticity mainly affects the short-period modes of an elastic hypersonic vehicle. Neglecting the relatively minor effects of long-period modes such as velocity, altitude, and track angle, the attitude and elastic vibration equations shown in equation (1) can be written as:

[0062] (20)

[0063] In the formula: the coefficient of the generalized coordinate derivative term is 0. The relevant functions and expressions are as follows:

[0064]

[0065] definition , and Equation (20) can be written as follows:

[0066] (twenty one)

[0067] In the formula: The variables are introduced to facilitate derivation; ;make We obtain the following formula:

[0068] (twenty two)

[0069] In the formula: For intermediate variables of the rigid body subsystem, The rudder deflection generated by the rigid body subsystem.

[0070] The coupled model shown in equation (21) is decomposed into a rigid body subsystem and an elastic subsystem. The rigid body subsystem is as follows:

[0071] (twenty three)

[0072] in:

[0073]

[0074]

[0075] In the formula: Angle of attack for the rigid body subsystem.

[0076] make , Rapidly changing time scales And the elevator deflection of the elastic subsystem Then the elastic subsystem can be written as:

[0077] (twenty four)

[0078] The elastic subsystem can be further obtained as follows:

[0079] (25)

[0080] In the formula: ,in The expression is:

[0081] (26)

[0082] Step 4: Design of an attitude active disturbance rejection controller for a flexible hypersonic vehicle

[0083] To avoid the influence of non-minimum phase, the canard deflection angle is designed. Ignoring long-period states such as speed, altitude, and track angle, where elasticity has a relatively small impact, the attitude and elastic system equations shown in equation (1) can be written as:

[0084] (27)

[0085] In the formula: For the steady-state control gain of the model; Let be the total disturbance of the pitch channel, including the effects of thrust and elasticity on the pitch moment. , and The expression is:

[0086] (28)

[0087] Design an extended state observer for equation (27):

[0088] (29)

[0089] in: As an auxiliary variable; This is the estimated total disturbance.

[0090] After obtaining the total disturbance estimate, the design incorporates a comprehensive control input that includes both disturbance compensation and error feedback:

[0091] (30)

[0092] in: For pitch angle command, and The pitch angle and pitch rate are measured by the sensor. and These are the proportional and derivative control gains, respectively.

[0093] Substituting (30) into (27), we get the following equation:

[0094] (31)

[0095] in, The term represents the total disturbance observation error. When the ESO bandwidth is greater than the bandwidth of the total disturbance, Able to track quickly The changes made If the value approaches zero, then classic PD control can achieve good control over the compensated object.

[0096] make and These represent the pitch angular velocity and pitch angle outputs of the measured signal after passing through the adaptive notch filter, respectively:

[0097] (32)

[0098] (33)

[0099] Equation (33) is the final designed elevator deflection, used for the control of the rigid body subsystem in the overall control framework.

[0100] Step 5: Active-passive combined vibration suppression control for hypersonic vehicles with rigid-elastic coupling

[0101] Equation (25), based on the RBF neural network, is rewritten as follows:

[0102] (34)

[0103] Take the maximum control input value , , ,in, For the rudder deflection of the elastic subsystem, To approximate the rudder deflection of the elastic subsystem, the control input is constrained. Represented as:

[0104] (35)

[0105] The input / output algorithm for the RBF network is as follows:

[0106] (36)

[0107] in: The input to the network consists of elastic generalized coordinates and their derivatives. For the hidden layer of the network One network input; For the first The center of each kernel function, For the first The width parameter of each kernel function; This is the output of the Gaussian function; The ideal weights for the network; Approximating the ideal neural network The error, , This represents the maximum error.

[0108] If the network input is taken as elastic generalized coordinates and their derivatives, then the network output is:

[0109] (37)

[0110] in: For network output, These are the estimated weights for the neural network.

[0111] Pick ,but .

[0112] For the second-order nonlinear system shown in equation (34), the control objective is taken as... That is, approaching , For elastic generalized coordinates. The angular error is defined as... ,but The sliding mode function is , If the parameter is greater than 0, then:

[0113] (38)

[0114] The elevator deflection of the elastic subsystem is as follows:

[0115] (39)

[0116] in, .

[0117] then:

[0118] (40)

[0119] Equation (39) is the elastic subsystem rudder deflection approximated by the RBF neural network, which, together with the rigid body subsystem rudder deflection designed in step 4, controls the aircraft.

[0120] The beneficial effects of this invention are:

[0121] This invention addresses the problem of poor attenuation of elastic vibrations in hypersonic vehicles under excitation, proposing a combined active and passive vibration suppression control method for rigid-elastic coupling hypersonic vehicles under excitation. First, based on singular perturbation theory, the rigid-elastic coupling model is decomposed into a rigid body subsystem and an elastic subsystem. For the elastic subsystem, a composite adaptive parameter identification method and a filtering estimation method are proposed. Using the estimated high-precision elastic state variables, an active elastic feedback controller is designed to ensure the stable convergence of the elastic modes under excitation. For the rigid body subsystem, a passive active disturbance rejection control (ADC) for the hypersonic vehicle is designed. The elastic modes are treated as disturbances to the rigid body, and linear ADC is used to quickly estimate and compensate for elastic disturbances from within the model, ensuring the rigid body control performance of the hypersonic vehicle under rigid-elastic coupling characteristics.

[0122] This invention introduces an RBF neural network, which can approximate any nonlinear function with a compact size and arbitrary precision. In control design, due to the limitations of the actuator, the problem of constrained control input is easily generated, making it difficult to achieve excessively large control law values, thus affecting the stability and control performance of the control system. Utilizing the powerful approximation capability of the RBF neural network, limited compensation for constrained control input is achieved. In this system, the control rudder deflection of the elastic subsystem is compensated, enabling the control input to accurately and effectively suppress the elastic mode, thereby achieving a better active suppression effect. In other words, the active-passive integrated control method for hypersonic vehicles with rigid-elastic coupling proposed in this invention exhibits higher accuracy and adaptability in the control of elastic hypersonic vehicles, especially under the complex flight environment and high uncertainty of elastic hypersonic vehicles, providing a more efficient solution. Attached Figure Description

[0123] Figure 1 This is a block diagram of the integrated active and passive control system for a hypersonic vehicle.

[0124] Figure 2 This is a speed tracking diagram of a hypersonic vehicle;

[0125] Figure 3 This is a speed error diagram for hypersonic vehicles;

[0126] Figure 4 This is an altitude tracking diagram of a hypersonic vehicle;

[0127] Figure 5 This is a diagram showing the altitude error of a hypersonic vehicle.

[0128] Figure 6 This is a tracking diagram of the trajectory angle of a hypersonic vehicle;

[0129] Figure 7 This is a tracking diagram of the pitch angular velocity of a hypersonic vehicle;

[0130] Figure 8 This is a pitch angle tracking diagram of a hypersonic vehicle;

[0131] Figure 9 This is a simulation diagram of the first-order generalized coordinates of a hypersonic vehicle;

[0132] Figure 10 This is a simulation diagram of the first-order generalized coordinate derivative of a hypersonic vehicle;

[0133] Figure 11 This is a simulation diagram of the second-order generalized coordinates of a hypersonic vehicle;

[0134] Figure 12 This is a simulation diagram of the second-order generalized coordinate derivative of a hypersonic vehicle;

[0135] Figure 13 This is a simulation diagram of the third-order generalized coordinates of a hypersonic vehicle;

[0136] Figure 14 This is a simulation diagram of the third-order generalized coordinate derivative of a hypersonic vehicle;

[0137] Figure 15 This is a simulation diagram of the throttle opening of a hypersonic aircraft;

[0138] Figure 16 This is a simulation diagram of the elevator deflection angle of a hypersonic aircraft;

[0139] Figure 17 This is a simulation diagram of the canard deflection angle of a hypersonic aircraft. Detailed Implementation

[0140] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings and technical solutions.

[0141] The integrated active and passive control method for hypersonic vehicles with rigid-elastic coupling includes the following:

[0142] Step 1: Construct a longitudinal dynamic model of the elastic hypersonic vehicle. The control model used in this invention is the Lisa model, which significantly reduces the model complexity while preserving the elastic effect of the hypersonic vehicle, and can meet the requirements of control system design.

[0143] Step 2: Online estimation of elastic vibration modes. The elastic vibration frequency is adaptively identified by a cascaded adaptive filter. An elastic mode filtering estimation method is designed to provide high-precision and low-cost elastic modal state quantities for the subsequent active feedback controller design.

[0144] Step 3: Decomposition of the rigid-elastic coupling model;

[0145] Step 4: Design of attitude active disturbance rejection controller for elastic hypersonic vehicle. The elastic mode is regarded as a disturbance to the rigid body. An extended state observer is designed, and linear active disturbance rejection control is used to quickly estimate and compensate for elastic disturbances from inside the model, thereby ensuring the rigid body control performance of the hypersonic vehicle under the rigid-elastic coupling characteristics.

[0146] Step 5: Active and passive vibration suppression control for hypersonic vehicles with rigid-elastic coupling. For the rigid body subsystem, the hypersonic vehicle self-disturbance rejection method designed in step 4 is adopted to ensure rigid body control performance. For the elastic subsystem, an RBF neural network is introduced to compensate for the elastic generalized coordinates. Combined with sliding mode control, the control rudder deflection of the elastic subsystem is designed to achieve effective and rapid suppression of elastic modes.

[0147] The specific implementation method is as follows:

[0148] Step 1: Construct a longitudinal dynamic model of the elastic hypersonic vehicle. Here are the definitions of the variables in the formula: Where: These are speed, altitude, track angle, angle of attack, and pitch rate, respectively. For elastic modes and their derivatives; and These are the damping ratio and vibration frequency of the elastic mode, respectively; These are the aircraft's mass, moment of inertia, and gravitational acceleration, respectively. and These are the generalized forces representing lift, drag, thrust, pitching moment, and elastic modes, respectively. The model is as follows:

[0149] (41)

[0150] Step 2: Online estimation of elastic vibration modes. The designed cascaded adaptive filter can achieve adaptive identification of elastic vibration frequencies. Compared with traditional identification algorithms, it introduces an evaluation signal. A composite adaptive parameter identification method was designed based on the detection mechanism (Equation (15)) to improve the accuracy and robustness of the elastic vibration frequency identification process.

[0151] The elastic vibration frequency identification value obtained by equation (16) is used for the design of cascaded adaptive filters to ensure passive notch suppression of elastic modes; the elastic mode estimation values ​​obtained by equations (18) and (19) are used for the design of the fifth step elastic mode active suppression controller to ensure the rapid stabilization of elastic modes under control force excitation.

[0152] Step 3: Decomposition of the rigid-elastic coupling model. Using singular perturbation theory, the rigid-elastic coupling model is decomposed into a rigid body subsystem and an elastic subsystem. Subsequently, controllers are designed for the rigid body subsystem and the elastic subsystem respectively, so as to satisfy the tracking control of the rigid body attitude command while rapidly suppressing the elastic multimodal behavior.

[0153] Step 4: Design of an attitude self-disturbance rejection controller for the elastic hypersonic vehicle. To eliminate non-minimum phase effects, a canard deflection angle was designed. Based on equation (27), an extended state observer was designed. After obtaining the estimated value of the total disturbance, a comprehensive control input that simultaneously includes disturbance compensation and error feedback was designed:

[0154] (42)

[0155] The above-mentioned active disturbance rejection control design ensures the rigid body control performance of hypersonic aircraft.

[0156] Step 5: Active-passive combined vibration suppression control for rigid-elastic coupled hypersonic vehicles. Combining active disturbance rejection control (ADRC) with the hypersonic vehicle's active-passive combined vibration suppression control, the overall control block diagram for rigid-elastic coupled hypersonic vehicles is shown below. Figure 1 As shown.

[0157] The simulation lasts for a total of 100 seconds, with a controller step size of 0.01 seconds. After 50 seconds, the rigid body controller parameters are changed from... Become This is used to simulate changes in internal control excitation. A second-order filter is used to smooth the speed and altitude commands.

[0158] Initial flight status , , , , , All other states are 0. The simulation comparison scheme of the hypersonic vehicle's active disturbance rejection passive control designed in step 4 is denoted as "adaptive filter-active disturbance rejection control," while the "active-passive combined vibration suppression control for hypersonic vehicles oriented towards rigid-elastic coupling" designed in step 5 is denoted as "active-passive combined vibration suppression control." Simulation results are as follows... Figures 2 to 17 . Figures 2 to 5 The speed, altitude tracking, and error diagrams demonstrate that the present invention can achieve rapid tracking of speed and altitude commands. Figures 6 to 8 Simulation graphs for track angle, pitch angular velocity, and pitch angle are shown respectively, all demonstrating excellent control performance. Figures 9 to 14 The simulation diagrams of the first three generalized coordinates and their derivative terms clearly show that the active-passive integrated control method can effectively and quickly suppress elasticity. Under elastic disturbances, it can quickly suppress vibration and has a fast convergence speed. Figures 16 to 17 Simulation diagrams for throttle opening, elevator deflection angle, and canard deflection angle are shown. No divergence was observed in the control surface and fuel equivalence ratio responses under both control methods, indicating that both methods have good stability and feasibility in terms of control input. However, this invention can quickly adjust the control input to a smooth state, ensuring system stability and control accuracy. In summary, this invention can significantly accelerate the convergence speed of the elastic mode and reduce the elastic amplitude, enabling hypersonic vehicles to achieve satisfactory tracking performance.

Claims

1. A method for integrated active and passive control of a hypersonic vehicle with rigid-elastic coupling, characterized in that, The specific steps are as follows: Step 1, constructing the longitudinal dynamics model of elastic hypersonic vehicle Neglecting the elastic modes above the third order, the longitudinal dynamics model of elastic hypersonic vehicle is shown in equation (1); (1) ； where are the velocity, altitude, flight-path angle, angle of attack and pitch rate, respectively, and denotes the first derivative, and denotes the second derivative; are the elastic modes, i.e. the first three generalized coordinates of elastic vibration; and are the damping ratios and vibration frequencies of the elastic modes, respectively; are the mass, moment of inertia and gravitational acceleration of the aircraft, respectively; and are the lift, drag, thrust, pitching moment and generalized forces of the elastic modes, respectively. Considering the rate gyroscope measurement model as: (4) ； wherein: is the measured pitch rate of the gyroscope, is the measured pitch angle of the gyroscope, is the pitch rate of the rigid body, is the pitch angle of the rigid body; , is the mode shape of the elastic mode of order , is the mounting position of the gyroscope; Step 2, online estimation of elastic vibration modes For the elastic modal wave trap problem with a vibration frequency of ω0= 1000 Hz, the cascade type wave trap is ​ (5) ； wherein: is a delay operator, represents a delay of 2 sample periods; is a parameter to be identified, is a sampling time; is a tuning parameter; The elastic vibration frequency is identified as , and the notch filter obtained according to the frequency identification result is: (6) ； In the formula: is an estimate of the cascade type notch filter is an estimate of the cascade type notch filter is an estimate of the cascade type notch filter is an estimate of the cascade type notch filter is an estimate of the cascade type notch filter Based on equation (6), it can be deduced that the first... The first notch filter Step output for: (7) ； In the formulae: ; So, the first step output of the notch pole section is : (8) ； Equations (7) and (8) are combined to obtain the following formula: (9) ； Introducing a recognition performance evaluation signal the output of the improved notch filter as follows: (10) ； Defining a cost function : (11) ； In the formula: is a forgetting factor; Definitions: , , to give the following formula: (12) ； The with respect to differentiating and setting to zero gives the following: (13) ； Using the recursive least squares algorithm, equation (13) is written in another form: (14) ； In the formulae: , ; According to equation (4), the sensor measured pitch rate is passed through a notch filter , and The actual output signal of each notch filter is then: (17) ； wherein: , and are filter error coefficients; According to equation (17), the estimate of , and is (18) ； By analogy, the estimated values of , and are: (19) ； In the formulae: , and According to acquisition; Step 3, decomposition of rigid-elastic coupled model Neglecting the velocity, altitude and track angle, the attitude and elastic vibration equations shown in equation (1) are written as: (20) ； wherein the coefficient of the generalized coordinate derivative term is zero, The correlation function and expression are as follows: ； Definitions , and , formula (20) is written as follows: (21) ； In the formulae: are introduced variables; ; Let , to give the following formula: (22) ； wherein: is an intermediate variable in the rigid body subsystem, is the rudder deflection generated by the rigid body subsystem; The coupled model shown in equation (21) is decomposed into a rigid body subsystem and an elastic subsystem, and the rigid body subsystem is as follows: (23) ； Wherein: ； ； In the formulae: is the angle of attack of the rigid subsystem; Let , , the fast time scale , and the elastic subsystem elevators deflection , the elastic subsystem is written as: (24) ； Further, the elastic subsystem is as follows: (25) ； In the formula: wherein The expression is: (26) ； Step 4, design of active disturbance rejection controller for elastic hypersonic vehicle attitude Design of canard deflection angle Neglecting velocity, altitude, and flight-path angle, the attitude and elastic system equations are written as: (27) ； where: is the model steady-state control gain; is the total disturbance to the pitch channel, including the effect of thrust, elastic effects on the pitch moment; let , and the expression is: (28) ； An extended state observer is designed for equation (27): (29) ； wherein: is an auxiliary variable; is a total disturbance estimate; After obtaining the total disturbance estimate, a comprehensive control input is designed, which contains both disturbance compensation and error feedback: (30) ； wherein: is a pitch angle command, and are the pitch angle and pitch angular velocity measured by the sensor, and are the proportional and derivative control gains, respectively. Substituting equation (30) into equation (27), the following equation is obtained: (31) ； wherein, the term is the total disturbance observation error term; Let and denote the pitch angle velocity and the pitch angle output of the measurement signal after the adaptive notch filter, respectively, i.e.: (32) ； (33) ； Equation (33) is the final design of elevator deflection, which is used for rigid body subsystem control in the overall control framework; Step 5, passive vibration suppression control for rigid-elastic coupled hypersonic vehicle Equation (25) is rewritten based on RBF neural network as follows: (34) ； Taking the maximum control input value as , , where, is the elastic subsystem rudder deflection, is the approximated elastic subsystem rudder deflection, the control input limited function is expressed as: (35) ； The input and output algorithm of RBF network is as follows: (36) ； wherein: is the network input, the input quantity is the elastic generalized coordinate and its derivative; is the network hidden layer the th network input; is the center of the th kernel function, is the width parameter of the th kernel function; is the output of the Gaussian basis function; is the ideal weight of the network; is the error of the ideal neural network approximation , , is the maximum error; The network input is the elastic generalized coordinates and their derivatives, and the network output is as follows: (37) wherein: is the network output, is the estimated weight of the neural network; Take Then ; For the second-order nonlinear system shown in equation (34), the control objective is taken as... That is, approaching , For elastic generalized coordinates; define the angle error as... ,but The sliding mode function is , If the parameter is greater than 0, then: (38) ； The elevator deflection of the elastic subsystem is as follows: (39) ； wherein ; Thus: (40) ； Equation (39) is the RBF neural network approximation of the elastic subsystem deflection, which is combined with the rigid body subsystem deflection designed in step 4 to control the vehicle.

2. The hypersonic vehicle active-passive integrated control method for rigidly coupled according to claim 1, characterized in that, In formula (1), and The specific expression of the above is as follows: (2) ； In the formula: (3) ； wherein, is the dynamic pressure; is the aerodynamic reference area; is the thrust moment arm; is the average aerodynamic chord; is the throttle setting; is the elevator deflection; is the canard deflection; , , and are the lift, drag, thrust and pitching moment coefficients, respectively; is the vector of the first three generalized coordinates and their derivatives ; , , , , , are the generalized force related aerodynamic parameter coefficients; , , , , , , , are the thrust related aerodynamic parameter coefficients; , , , , , are the pitching moment related aerodynamic parameter coefficients; , , , , are the lift related aerodynamic parameter coefficients; , , , , , , , are the drag related aerodynamic parameter coefficients.

3. The hypersonic vehicle active-passive integrated control method for rigidly coupled according to claim 1, characterized in that, In step 2, a mechanism is designed to monitor the identification performance to determine whether the identification performance is good or bad: (15) ； The identification result of the elastic vibration frequency is as follows: (16)。

Images