A method for integrated control of air-breathing hypersonic vehicle flight and propulsion

Abstract

The application belongs to the technical field of air-breathing hypersonic vehicle control, and relates to an air-breathing hypersonic vehicle flight-propulsion integrated control method, comprising: establishing an air-breathing hypersonic vehicle propulsion system mechanism model, mechanism model correction, control-oriented dynamic modeling and flight-propulsion integrated controller design. Through strict mathematical proof and comprehensive simulation verification, the short-period dynamic response quality and robustness of the vehicle are significantly improved, and the application prospect is wide.

Description

Technical Field

[0001] This invention belongs to the field of air-breathing hypersonic vehicle control technology, and relates to an integrated flight and thrust control method for air-breathing hypersonic vehicles. Background Technology

[0002] Due to the integrated design of the airframe and propulsion system, a significant flight-propulsion coupling effect exists between the propulsion system and aerodynamic characteristics. On the one hand, changes in flight state affect the inlet shock wave structure and flow capture characteristics, thereby altering the operating point and thrust output of the propulsion system. On the other hand, the additional pitching moment generated by the propulsion system directly impacts the aircraft's attitude dynamics, creating a strong two-way coupling. In this context, traditional separate flight / propulsion design methods based on decoupling assumptions are insufficient to meet the requirements for high-precision and robust control. Furthermore, existing mechanistic models inevitably suffer from parameter perturbations and modeling errors under wide-speed-range and large-envelope flight conditions, leading to a significant decrease in model accuracy. Therefore, there is an urgent need to develop an integrated flight / propulsion anti-saturation control strategy that integrates mechanistic models for online correction, thereby improving the adaptability and robustness of the control system through real-time model error correction.

[0003] The patent "An Altitude Control Method for Air-breathing Hypersonic Vehicles Based on Feature Models" (CN111061283A) proposes a feature modeling and adaptive control scheme for the longitudinal plane of air-breathing hypersonic vehicles. This method decomposes the longitudinal plane into an outer-loop subsystem based on altitude and track angle, and an inner-loop subsystem based on pitch angle and pitch rate. Second-order feature models are established for each subsystem, and angle-of-attack commands, canard deflection control laws, and elevator deflection control laws are designed accordingly to achieve altitude tracking under high-maneuverability flight conditions. However, this method primarily focuses on control law design and feature parameter adaptation, without systematically analyzing and correcting the vehicle's mechanistic model from a flight-thrust coupling perspective. Therefore, its ability to support the accuracy of the mechanistic model is limited in wide-speed-range flight-thrust integrated scenarios.

[0004] The patent "A Correction Method for the Mathematical Model of Shock Train Leading Edge Position in an Inward-Rotating Inlet" (CN115600372B) proposes a correction method for the empirical formula of shock train leading edge position in an inward-rotating inlet to solve the problem of inaccurate prediction by traditional empirical formulas due to the complex geometric configuration of the inward-rotating inlet. However, this method mainly focuses on engine performance and does not analyze the aircraft mechanism model from the perspective of flight / propulsion coupling.

[0005] The patent "Integrated Flight and Thrust Control Method for Hypersonic Vehicles Based on INDI" (CN121277212B) discloses an integrated flight and thrust control scheme based on incremental nonlinear dynamic inversion (INDI). This scheme utilizes INDI at the control level to jointly calculate the elevator and fuel equivalence ratio (or equivalent thrust control quantity) on the "angular acceleration / linear acceleration" scale; the trajectory loop uses traditional PI or proportional guidance to generate commands. However, its drawback lies in the lack of a switching or exit mechanism for the integrated control mode. The use of a long-term integrated control architecture is not in line with engineering practice and is prone to causing long-period speed drift and unnecessary increases in fuel consumption.

[0006] The paper "Model Correction and Control of High-Speed ​​Aircraft Oriented to Mode Transition [J]. Journal of Astronautics, 2026, 47(6):1525-1535" proposes a method that integrates flight / propulsion mechanism model correction and robust control. By inputting the shock train length, the flight-propulsion integrated mechanism model is corrected to ensure the attitude stability of the aircraft during the mode transition process of the propulsion system. However, this paper only considers flight / propulsion integration at the mechanism model correction level, and still adopts the traditional decoupled control strategy at the dynamic modeling and control levels, without making full use of the flight-propulsion coupling characteristics.

[0007] The paper "Flight-Propulsion Integration Dynamic Analysis and Adaptive Control of the Hypersonic Vehicle at Wide-Range Mach Numbers [J]. IEEEAccess, 2022, 41:878-886" establishes a control-oriented mathematical model for the integrated flight and propulsion of a hypersonic vehicle and proposes a dynamic characteristic analysis method considering flight-propulsion coupling. Adaptive pole placement is used to achieve integrated flight-propulsion control over a wide speed range. Although this paper establishes the basic paradigm of integrated flight-propulsion control, it does not specifically discuss the compensation mechanism for model errors in wide-speed-range flight and their impact on control, thus lacking strong practical engineering value.

[0008] In summary, the existing research on integrated flight and propulsion control still has the following common shortcomings: (1) Model correction mostly adopts mathematical methods and fails to combine data and physical information, resulting in insufficient physical interpretability of mechanism model correction for mechanical calculation; (2) Modeling and control are separated, and mechanism model correction and integrated dynamic model for control have not yet been integrated to form a complete closed loop of "online model correction - integrated control"; (3) Current integrated flight and propulsion control methods introduce physical quantities that are difficult to measure in engineering, making it difficult to fully utilize coupling characteristics while taking into account realistic constraints and engineering feasibility. Summary of the Invention

[0009] The purpose of this invention is to design a flight / propulsion coordinated control strategy that integrates intelligent online correction of the mechanism model, integrated anti-saturation control, and mode switching mechanism, so as to systematically solve the problems of model mismatch and control quality degradation under wide speed range flight conditions, and provide an integrated flight and propulsion control method for air-breathing hypersonic vehicles.

[0010] The technical solution of the present invention is as follows: A method for integrated flight and thrust control of an air-breathing hypersonic vehicle, comprising the following specific steps: Step 1: Establish a mechanism model of the propulsion system of an air-breathing hypersonic vehicle; By establishing a mechanism model of the propulsion system of an air-breathing hypersonic vehicle, the Mach number of the airflow can be solved. ,pressure With temperature Spacecraft body axis coordinates x By studying the variation pattern of the shock wave train in the isolation section, the length of the shock wave train in the isolation section is determined, and finally the airflow state at the tail nozzle is obtained. The thrust of the engine is then determined using Newton's laws.

[0011] Mach number of airflow in the isolation section and combustion chamber ,pressure With temperature The governing equations are: (1) (2) (3) Among them, cross-sectional area parameters , The cross-sectional area of ​​the internal flow channel. For the body axis coordinates of the aircraft; , Combustion chamber expansion ratio; Total temperature variation coefficient , Total airflow temperature; fuel flow rate variation coefficient , Fuel flow rate; coefficient of friction of the internal flow channel , The specific heat ratio of the airflow. The friction coefficient of the internal flow channel. The hydraulic diameter of the aircraft's body axis; intermediate variable .

[0012] Among them, the total temperature of the airflow Described using empirical formulas: (4) in, The total temperature at the intake duct outlet. It is the empirical exothermic constant; For dimensionless position, , This is the location of the fuel injection point. This refers to the location of the combustion chamber outlet. This refers to the heating ratio.

[0013] The theoretical length of the shock train is calculated as follows: The theoretical length of the shock train in the supercombustion mode for: (5) in, The Reynolds number is... The height or diameter of the isolation section. For boundary layer thickness, This refers to the intake outlet pressure. This refers to the Mach number of the airflow at the intake outlet. To reduce pressure at the outlet of the isolation section.

[0014] The theoretical length of the shock train in sub-combustion mode for: (6) in, The Mach number of the airflow at the outlet of the isolation section. This represents the Mach number of the gas flow at the combustion chamber exit.

[0015] At the tailpipe, the expansion state of the high-temperature, high-pressure airflow is calculated using a plume model to obtain the outlet airflow velocity of the tailpipe. and tail nozzle outlet pressure Then, determine the engine thrust according to Newton's laws. : (7) in, , , , , These are air mass flow rate, fuel equivalence ratio, appropriate air-fuel ratio, inlet airflow velocity, and inlet static pressure, respectively. The tailpipe cross-sectional area is... This refers to the cross-sectional area of ​​the air intake.

[0016] Step 2: Mechanism model correction; First, potential error sources introduced during the modeling process in step 1 are analyzed by substituting these error sources into the modeling formula, and parameters for correction are selected. Second, taking the consistency of the shock train length ratio with the isolation section as the correction target, a physical information neural network is designed to correct the mechanism model.

[0017] Air-breathing hypersonic vehicles are equipped with dual-mode scramjet engines. When calculating their thrust and induced pitch moment, the airflow pressure is closely related to the velocity. Therefore, formulas (1) and (2) determine the basic characteristics of the thrust and thrust-induced pitch moment. After considering the sources of error and introducing compensation terms, the airflow Mach number... With pressure Represented as: (8) (9) in, and To account for the Mach number and air pressure of the airflow after error compensation, This represents the change in the coefficient of friction. This represents the change in cross-sectional area parameters. This represents the change in the total temperature coefficient. The coefficient for change in fuel flow rate. This represents the change in the friction coefficient of the internal flow channel.

[0018] Combining (1) and (2), equations (8) and (9) can be expressed as: (10) (11) Equations (10) and (11) show that the effects of the cross-sectional area of ​​the inner flow channel, the total temperature change error, the fuel injection error, and the friction coefficient error are linearly correlated and can compensate for each other. Therefore, according to equations (10) and (11), the selection of the appropriate parameters... Make corrections, because only The coefficients in equations (10) and (11) are the same. Based on equations (10) and (11), the outlet pressure of the isolation section can be obtained. Mach number of the airflow at the outlet of the isolation section Mach number of the airflow at the combustion chamber exit , to support the calculation of shock train length.

[0019] In model correction, accurate thrust data typically comes from scaled-down wind tunnel tests. Due to significant size differences and limitations in measurement technology, the Mach numbers at key nodes are often difficult to obtain. M ,pressure P Thrust and modeling results cannot usually be directly compared to verify model accuracy. In engineering, the accuracy is typically verified by comparing the calculated relative length of the shock train within the isolation section with experimental data. If they match, the model is considered to have achieved sufficient accuracy. In experiments, the actual length of the shock train is measured by sensors.

[0020] A physical information neural network is used to correct the mechanistic model, with the goal of ensuring that the relative lengths of the theoretical shock train and the isolation section calculated by the model are consistent with the actual relative lengths of the shock train and the isolation section measured by the sensors. The input to the physical information neural network is height. h ,Mach number Angle of attack Elevator deflection Fuel equivalence ratio The output is ,Right now The estimated value is expressed as: (12) in, For output variables, The number of interneurons. n The number of input quantities. Indicates the first One input variable. Weights and These represent the connections from the input layer to the hidden layer and from the hidden layer to the output layer, respectively. Accordingly, and These represent the bias terms of neurons in the hidden layer and the output layer, respectively. The nonlinear transformations within the hidden and output layers are determined by the activation functions. and Decide.

[0021] Output variables It is used in the calculation of the mechanism model, and the loss function is constructed based on the shock train length of the isolation section predicted by the mechanism model.

[0022] The relative positions of the shock train leading edges predicted by the mechanistic model are denoted as follows: The high-precision data of the relative positions of the leading edges of the shock train in the isolation section are denoted as ,in, and L 1 represents the position of the shock train leading edge and the length of the isolation section in the mechanistic model. x 2 and L 2 represents the position of the leading-edge shock train and the length of the isolation section in the experiment. Correspondingly, the loss function of the physical information neural network is defined as: (13) in, Let be the loss function of the neural network. This represents the number of data sets used for training. The neural network is trained using backpropagation of the error.

[0023] Step 3: Dynamic modeling for control; Neglecting airframe elasticity, establish the longitudinal dynamic equations for an air-breathing hypersonic vehicle: (14) in," " is the first derivative; , , , , , and These are velocity, trajectory angle, angle of attack, pitch angular velocity, gravitational acceleration, mass, and pitch moment of inertia. , , and These are thrust, lift, drag, and pitch moment, which can be obtained through interpolation tables, as follows: (15) in, and These are altitude and elevator, respectively. For thrust interpolation function, For the lift interpolation function, For resistance interpolation function, This is the pitch moment interpolation function.

[0024] Based on the small perturbation assumption, the longitudinal dynamic equation is linearized by decomposing the current state into the equilibrium point and the small perturbation quantity and ignoring higher-order terms. Under the small perturbation assumption, the nonlinear dynamic model of equation (14) is transformed into the linear form shown in equation (16), which is achieved by expressing the state variables as the sum of their balancing values ​​and incremental deviations, and ignoring second-order and higher-order components.

[0025] (16) in, a , b These represent the state correlation coefficient and the control correlation coefficient, respectively. , , , , , , , , , , , , , , , , , , , . For the partial derivative of thrust with respect to velocity, For the partial derivative of thrust with respect to angle of attack, For the thrust-fuel equivalence ratio partial derivative, For the drag partial derivative with respect to velocity, For the drag partial derivative with respect to the angle of attack, For the deflection of the elevator by the drag, For the drag partial derivative with respect to fuel equivalence ratio, For the partial derivative of lift with respect to velocity, For the lift derivative with respect to the angle of attack, To guide the elevator for lift, For the lift-to-fuel equivalence ratio partial derivative, For the partial derivative of pitch moment with respect to velocity, For the partial derivative of pitch moment with respect to angle of attack, For the partial derivative of pitch moment with respect to pitch angular velocity, To guide the pitch moment on the elevator deflection, For the partial derivative of pitching moment with respect to fuel equivalence ratio, For pitch rotation inertia, For the rate of change of velocity, The rate of change of the trajectory angle. For the rate of change of angle of attack, For the rate of change of the elevator, The rate of change of fuel equivalence ratio. This represents the rate of change of pitch angular velocity.

[0026] According to the longitudinal dynamic equations shown in equations (14) and (15), and Both can directly affect the pitch moment characteristics to achieve attitude control. For long-term speed control, it directly affects thrust, but it can also indirectly affect the aircraft speed by changing the angle of attack, thus affecting thrust. Therefore, although the first line of formula (16) describes The relationship between the effect on speed, but It is usually not used as a control variable for speed regulation. Ignore. The effect on speed, the extracted state-space equation for short-cycle attitude control is as follows: : (17) in, For the first k The derivative of the state vector at step +1. State matrix , For the first k Step state vector, , , The first kTrack angle, angle of attack, pitch rate, For control matrix , For the output matrix , For the first k The control input vector of the step, , , The first k Elevator, fuel equivalence ratio For the first k The output matrix of the step.

[0027] Step 4: Design of the integrated flight and propulsion controller; Using the kinematic equations established in step 3 as the controlled object, and the dynamic characteristic data obtained from the modified mechanism model in step 2, the integrated control design of flight propulsion is completed.

[0028] The flight-thrust attitude dynamics of an air-breathing hypersonic vehicle constitute a typical constrained multiple-input multiple-output (MIMO) system. For constrained MIMO systems, model predictive control is a highly promising candidate method because it provides a systematic approach to handling time-varying model characteristics.

[0029] For the discrete-time linear model shown in equation (17), a model predictive controller is designed. To ensure local stability, the model predictive controller is considered as a regulation problem with an infinite time-domain cost function, as shown below: (18) in, and To adjust the weight matrix, For the first k Step cost function. For from the first k Step start, then... i The state vector of the step, For from the first k Step start, then... i The control vector for the first step. In the... k Step 1: The parameters of the model predictive controller are obtained by solving the optimization problem shown in equation (19): (19) And it satisfies the following constraints: (20) in, and This represents the minimum and maximum control constraints of the control quantity. This indicates the control quantity from the previous step. and Represents the minimum and maximum constraints of state variables. This represents the state variables from the previous step in balancing.

[0030] The beneficial effects of this invention are: This invention addresses the model error problem of wide-speed-range aircraft by providing an in-depth analysis of the error sources and propagation mechanisms caused by flow channel adjustments. Based on this, a correction framework based on a physical information neural network is constructed. Real-time model correction is performed using the mechanism model correction method described in equations (12)-(13) and sensor measurement data, significantly improving the fidelity of the dynamic model under variable flow channel conditions. This overcomes the shortcomings of traditional pure numerical simulation methods in engineering applications due to insufficient reliability, providing a practical and feasible technical means for high-precision modeling.

[0031] A flight-propulsion integrated attitude dynamics control strategy that meets the requirements of practical engineering is proposed, overcoming the inherent limitations of traditional decoupled control. This strategy uses elevator deflection and fuel equivalence ratio as joint inputs to construct a fuselage-propulsion coupled dynamics model, and combines this with model predictive control to achieve integrated flight-propulsion attitude control. This design effectively suppresses strong aerodynamic / propulsion coupling effects while significantly improving the short-cycle dynamic response quality of the aircraft.

[0032] This invention demonstrates outstanding robustness and safety when the actuator faces saturation risks. By coordinating the elevator and fuel equivalence ratio, this strategy can automatically limit the amplitude of control variables within physical constraints, effectively avoiding instability caused by control surface or engine saturation, and theoretically guaranteeing the local stability of the closed-loop system. This invention provides a novel technical path for high-precision, high-reliability flight of wide-speed-range aircraft under complex constraints, possessing significant engineering practical value and promising prospects for widespread application. Attached Figure Description

[0033] Figure 1 This is a diagram showing the effect of model correction in the embodiment; Figure 2 These are simulation curves of the integrated flight and propulsion control during the accelerated climb phase; among them, (a) is the altitude curve, (b) is the velocity curve, (c) is the angle of attack curve, (d) is the trajectory angle curve, (e) is the pitch angle curve, (f) is the pitch angular velocity curve, (g) is the elevator curve, (h) is the fuel equivalence ratio curve, (i) is the angle of attack error curve, (j) is the trajectory angle error curve, (k) is the altitude error curve, and (l) is the velocity error curve. Figure 3These are simulation curves of the integrated control of angle of attack step flight thrust; among them, (a) is the altitude curve, (b) is the velocity curve, (c) is the angle of attack curve, (d) is the trajectory angle curve, (e) is the pitch angle curve, (f) is the pitch angular velocity curve, (g) is the elevator curve, (h) is the fuel equivalence ratio curve, (i) is the angle of attack error curve, (j) is the trajectory angle error curve, (k) is the altitude error curve, and (l) is the velocity error curve. Detailed Implementation

[0034] The embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0035] A method for integrated flight and propulsion control of an air-breathing hypersonic vehicle includes: Step 1: Establish a mechanism model of the propulsion system of an air-breathing hypersonic vehicle; In terms of the propulsion system of air-breathing hypersonic vehicles, the mechanical characteristics of the propulsion system are calculated based on formulas (1) to (5).

[0036] Step 2: Mechanism model correction; This invention designs a neural network that incorporates physical information to correct the propulsion system mechanism model. The specific network form and loss function are shown in formulas (12)-(13). In this embodiment, the model correction effect is as follows: Figure 1 As shown.

[0037] Step 3: Dynamic modeling for control; Neglecting airframe elasticity, the longitudinal dynamic equations of the air-breathing hypersonic vehicle are established as shown in equation (14). The state-space equation extracted for short-cycle attitude control, which is related to the effect on speed, is shown in Equation (17).

[0038] Step 4: Design of the integrated flight and propulsion controller; The optimization problem for control is constructed as shown in equations (18)-(20). Solve the optimization problem to obtain the controller parameters.

[0039] Figure 2 and Figure 3 The effectiveness of the proposed method was demonstrated from two perspectives: accelerated climb simulation and angle-of-attack step simulation. The accelerated climb flight mission was set as follows: accelerating from an altitude of 16,984 m and a speed of 1,180 m / s to an altitude of 35,889 m and a speed of 2,503 m / s. The initial simulation conditions were: aircraft mass 16,400 kg, reference area 26 m². 2The reference length is 1.5 m. Adaptive pole placement and trajectory linearization (TLC) control is selected as the benchmark. The control simulation frequency is set to 100 Hz, and a 20% deviation is applied to lift, drag, and pitch moment. The initial state for the angle-of-attack step simulation is set to an altitude of 23,200 m and an initial velocity of 1,910 m / s. The initial mass of the aircraft in both simulations is 40,000 kg, the characteristic length is 1.5 m, and the engine capture area is 1 m². 2 The reference area is 21 m². 2 The pitch moment of inertia is taken as 22 times its mass value.

[0040] Figure 2 Figures (a) and (b) show the altitude and velocity curves, respectively, demonstrating that the three control methods—integrated flight-thrust control (the method of this invention), adaptive pole placement, and trajectory linearization control—can all enable air-breathing hypersonic vehicles to complete basic acceleration and climb tasks. Figure 2 (c) shows the angle-of-attack curve, which underwent two flow channel adjustments around 62 s and 120 s, causing significant overshoot. The integrated fly-thrust control exhibited the smallest overshoot during the transient adjustment process, with a second overshoot of only 18.6%, and stabilized near the target value more quickly, demonstrating excellent dynamic response characteristics. In contrast, the trajectory linearization control showed larger fluctuations, with a second overshoot as high as 66.9%, and a longer settling time. Figure 2 In the middle (d) and (e), the curves of the flight path inclination and pitch angles are respectively. The overshoot is not obvious in the first adjustment, but it is more significant in the second adjustment. Among the three methods, the flight-thrust integrated control has the smallest overshoot, followed by the trajectory linearization control, and the adaptive pole configuration has the largest overshoot. Figure 2 In the middle (f), the pitch angular velocity curves have the following maximum values: 0.76 ° / s for integrated flight-thrust control, 1.41 ° / s for adaptive pole placement, and 4.48 ° / s for trajectory linearization control. Figure 2 In the middle (g) and (h) curves, respectively, the elevator deflection angle and the engine fuel equivalence ratio are shown. It can be seen that the trajectory linearization control and adaptive pole configuration exhibit severe chattering during the flow channel adjustment process, while the control output of the integrated flight thrust control is smoother. Figure 2 In the table, (i) and (j) represent the angle of attack error and the track inclination error, respectively. The maximum angle of attack errors of the integrated flight-thrust control, adaptive pole placement and trajectory linearization control are -0.48°, -0.65° and -0.86°, respectively, and the maximum track inclination errors are 0.28°, 0.33° and 0.78°, respectively. Figure 2In the equations (k) and (l), altitude and velocity errors are respectively. The final altitude errors of the integrated flight-thrust control, adaptive pole placement, and trajectory linearization control are 290.84 m, -976.25 m, and 68.10 m, respectively, and the final velocity errors are 3.56 m / s, 3.53 m / s, and 2.74 m / s, respectively. Although the trajectory linearization control has a smaller trajectory tracking error, the integrated flight-thrust control performs best in terms of short-period errors, which are more critical in disturbed environments. Furthermore, in hypersonic flight environments, its trajectory error remains within an acceptable range.

[0041] Figure 3 Figures (a) and (b) show the altitude and velocity curves, respectively, confirming that all three control methods can enable the flight trajectory to track the target during the step change of the angle of attack command of the air-breathing hypersonic vehicle. Figure 3 Figure (c) shows the angle-of-attack tracking curve. For the first step angle, the settling times for the integrated flight-thrust control, adaptive pole placement, and trajectory linearization control to reach 98% of the nominal value are 0.37 s, 1.73 s, and 1.23 s, respectively; for the second step angle, they are 0.56 s, 3.34 s, and 1.85 s, respectively. It is evident that the proposed integrated flight-thrust control method significantly shortens the settling time for angle-of-attack step control tasks. Figure 3 In the middle (d) and (e) curves, the trajectory inclination and pitch angle curves are respectively. No significant errors or oscillations were observed in any of the three control methods. Figure 3 In Figure (f), the pitch angular velocity curves are shown. The maximum angular velocities of the first and second step responses are 19.08 ° / s and -15.13 ° / s for integrated flight-thrust control, 15.86 ° / s and -11.43 ° / s for adaptive pole configuration, and 10.43 ° / s and -7.14 ° / s for trajectory linearization control, respectively. Figure 3 In the figure, (g) and (h) are the elevator deflection angle and engine fuel equivalence ratio curves, respectively. The maximum control quantity of the integrated flight-thrust control in the step response is significantly greater than that of the adaptive pole placement and trajectory linearization control, especially in terms of fuel equivalence ratio. This reflects the control mechanism of the integrated flight-thrust control that utilizes the additional pitching moment generated by the propulsion system to participate in attitude angle adjustment. Figure 3 In the figure, (i) and (j) are the angle of attack error and track inclination error curves, respectively. The error under the integrated flight and thrust control converges to zero faster than the other two methods. Figure 3In the figures (k) and (l), the altitude and velocity error curves are respectively. The final altitude error is smallest for the integrated flight-thrust control (-30.72 m), followed by the adaptive pole configuration (-65.55 m) and trajectory linearization control (-40.03 m). Because the integrated flight-thrust control method incorporates the fuel equivalence ratio into attitude angle adjustment, it generates a significant velocity error in the step response. However, under hypersonic flight conditions, this error is negligible.

Claims

1. A method for integrated flight and thrust control of an air-breathing hypersonic vehicle, characterized in that, The specific steps are as follows: Step 1: Establish a mechanism model of the propulsion system for an air-breathing hypersonic vehicle; By establishing a mechanism model of the propulsion system of an air-breathing hypersonic vehicle, the Mach number of the airflow can be solved. ,pressure With temperature Spacecraft body axis coordinates x The variation law is used to solve for the length of the shock train in the isolation section, and finally the airflow state at the tail nozzle is obtained. The thrust of the engine is then determined by Newton's laws. Step 2: Mechanism model correction; First, the error sources that may be introduced during the modeling process in step 1 are analyzed by substituting the error sources into the modeling formula, and the parameters to be corrected are selected. Second, taking the consistency of the length ratio of the shock train to the isolation section as the correction target, a physical information neural network is designed to complete the correction of the mechanism model. Step 3: Dynamic modeling for control; Step 4: Design of the integrated flight and propulsion controller; Using the kinematic equations established in step 3 as the controlled object, and the dynamic characteristic data obtained from the modified mechanism model in step 2, the integrated control design of flight propulsion is completed.

2. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 1, characterized in that, Step 1 is as follows: Mach number of airflow in the isolation section and combustion chamber ,pressure With temperature The governing equations are: (1) (2) (3) Among them, cross-sectional area parameter , The cross-sectional area of ​​the internal flow channel. For the body axis coordinates of the aircraft; , Combustion chamber expansion ratio; Total temperature variation coefficient , Total airflow temperature; fuel flow rate variation coefficient , Fuel flow rate; coefficient of friction of the internal flow channel , The specific heat ratio of the airflow. The friction coefficient of the internal flow channel. The hydraulic diameter of the aircraft's body axis; intermediate variable ; Among them, the total temperature of the airflow Described using empirical formulas: (4) in, The total temperature at the intake duct outlet. It is the empirical exothermic constant; For dimensionless position, , This is the location of the fuel injection point. This refers to the location of the combustion chamber outlet. The heating ratio; The theoretical length of the shock train is calculated as follows: The theoretical length of the shock train in the supercombustion mode for: (5) in, Let Reynolds number be 1. The height or diameter of the isolation section. Boundary layer thickness, This refers to the intake outlet pressure. This refers to the Mach number of the airflow at the intake outlet. To reduce pressure at the outlet of the isolation section; The theoretical length of the shock train in sub-combustion mode for: (6) in, The Mach number of the airflow at the outlet of the isolation section. This refers to the Mach number of the gas flow at the combustion chamber exit. At the tailpipe, the expansion state of the high-temperature, high-pressure airflow is calculated using a plume model to obtain the outlet airflow velocity of the tailpipe. and tail nozzle outlet pressure Then, determine the engine thrust according to Newton's laws. : (7) in, , , , , These are air mass flow rate, fuel equivalence ratio, appropriate air-fuel ratio, inlet airflow velocity, and inlet static pressure, respectively. The tailpipe cross-sectional area is... This refers to the cross-sectional area of ​​the air intake.

3. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 1, characterized in that, Step 2 is as follows: After considering the sources of error and introducing compensation terms, the airflow Mach number With pressure Represented as: (8) (9) in, and To account for the Mach number and air pressure of the airflow after error compensation, This represents the change in the coefficient of friction. This represents the change in cross-sectional area parameters. This represents the change in the total temperature coefficient. The coefficient for change in fuel flow rate. This represents the change in the friction coefficient of the internal flow channel; Combining (1) and (2), equations (8) and (9) can be expressed as: (10) (11) Based on equations (10) and (11), select the appropriate pair. Corrections are made; in model correction, the relative length of the shock train in the isolation section calculated by the model is compared with the experimental data. If they are consistent, the modeling is considered to have reached sufficient accuracy. The actual length of the shock train in the experiment is measured by the sensor. A physical information neural network is used to correct the mechanism model. The correction goal is to make the relative length of the theoretical shock train and the isolation section calculated by the model consistent with the actual relative length of the shock train and the isolation section measured by the sensor. The input of the physical information neural network is height. h ,Mach number Angle of attack Elevator deflection Fuel equivalence ratio The output is ,Right now The estimated value is expressed as: (12) in, For output variables, The number of interneurons. n The number of input quantities. Indicates the first One input variable; weight and These represent the connections from the input layer to the hidden layer and from the hidden layer to the output layer, respectively; correspondingly, and These represent the bias terms of neurons in the hidden layer and the output layer, respectively; the nonlinear transformations within the hidden and output layers are respectively determined by the activation functions. and Decide.

4. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 3, characterized in that, The relative positions of the shock train leading edges predicted by the mechanistic model are denoted as follows: The high-precision data of the relative position of the leading edge of the shock train in the isolation section is denoted as ,in, and L 1 represents the position of the shock train leading edge and the length of the isolation section in the mechanistic model. x 2 and L 2 represents the position of the leading-edge shock train and the length of the isolation section in the experiment; the loss function of the physical information neural network is defined as: (13) in, Let be the loss function of the neural network. This represents the number of data sets used for training.

5. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 1, characterized in that, Step 3 is as follows: Neglecting airframe elasticity, establish the longitudinal dynamic equations for an air-breathing hypersonic vehicle: (14) in," " is the first derivative; , , , , , and These are velocity, trajectory angle, angle of attack, pitch angular velocity, gravitational acceleration, mass, and pitch moment of inertia. Under the assumption of small perturbation, the nonlinear dynamic model of equation (14) is transformed into the linear form shown in equation (16); (16) in, a , b These represent the state correlation coefficient and the control correlation coefficient, respectively. , , , , , , , , , , , , , , , , , , , ; For the partial derivative of thrust with respect to velocity, For the partial derivative of thrust with respect to angle of attack, For the thrust-fuel equivalence ratio partial derivative, For the drag partial derivative with respect to velocity, For the drag partial derivative with respect to the angle of attack, For the deflection of the elevator by the drag, For the drag partial derivative with respect to fuel equivalence ratio, For the partial derivative of lift with respect to velocity, For the lift derivative with respect to the angle of attack, To guide the elevator for lift, For the lift-to-fuel equivalence ratio partial derivative, For the partial derivative of pitch moment with respect to velocity, For the partial derivative of pitch moment with respect to angle of attack, For the partial derivative of pitch moment with respect to pitch angular velocity, For example, the pitch moment deflects the elevator. For the partial derivative of pitching moment with respect to fuel equivalence ratio, For pitch rotation inertia, For the rate of change of velocity, The rate of change of the trajectory angle. For the rate of change of angle of attack, For the rate of change of the elevator, The rate of change of fuel equivalence ratio. The rate of change of pitch angular velocity; neglect The effect on speed, the extracted state-space equation for short-cycle attitude control is as follows: : (17) in, For the first k The derivative of the state vector at step +1. State matrix , For the first k Step state vector, , , The first k Track angle, angle of attack, pitch rate, For control matrix , For the output matrix , For the first k The control input vector of the step, , , The first k Elevator, fuel equivalence ratio For the first k The output matrix of the step.

6. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 5, characterized in that, , , and These are thrust, lift, drag, and pitch moment, obtained through interpolation tables, as follows: (15) in, and These are altitude and elevator, respectively. For thrust interpolation function, For the lift interpolation function, For resistance interpolation function, This is the pitch moment interpolation function.

7. The integrated flight and thrust control method for an air-breathing hypersonic vehicle according to claim 1, characterized in that, Step 4 is as follows: For the discrete-time linear model shown in equation (17), a model predictive controller is designed. To ensure local stability, the model predictive controller is considered as a regulation problem with an infinite time-domain cost function, as shown below: (18) in, and To adjust the weight matrix, For the first k Step cost function; For from the first k Step start, then... i The state vector of the step, For from the first k Step start, then... i The control vector of the first step; in the second step k Step 1: The parameters of the model predictive controller are obtained by solving the optimization problem shown in equation (19): (19) And it satisfies the following constraints: (20) in, and This represents the minimum and maximum control constraints of the control quantity. This indicates the control quantity from the previous step. and Represents the minimum and maximum constraints of state variables. This represents the state variables from the previous step in balancing.

Images