Control torque redistribution method for reusable launch vehicles facing control surface jamming

Abstract

This invention belongs to the field of reusable launch vehicle control technology, specifically relating to a method for redistributing control torque in reusable launch vehicles facing control surface jamming. The purpose of this invention is to address the problems of poor adaptability and insufficient robustness of control torque distribution when control surfaces of reusable launch vehicles are jammed. It introduces a dynamic control distribution reconfiguration strategy considering control surface synergy to achieve torque redistribution under fault conditions. The method includes constructing a six-degree-of-freedom model of the reusable launch vehicle and a control surface jamming fault model; detecting control surface jamming faults through an online recursive least squares algorithm; and designing a dynamic control distribution reconfiguration strategy considering control surface synergy to achieve torque distribution across multiple control surfaces. This method is a fault detection and dynamic control torque redistribution method and system adapted to wide-speed-range flight characteristics and has broad application prospects.

Description

Technical Field

[0001] This invention belongs to the field of reusable launch vehicle control technology, specifically relating to a method for redistributing control torque in reusable launch vehicles facing control surface jamming. It is a fault detection and dynamic control torque redistribution method adapted to wide speed range flight characteristics under control surface jamming fault conditions. Background Technology

[0002] Compared to traditional aircraft, reusable launch vehicles have become a major research focus in the aerospace industry worldwide due to their advantages such as rapid global reach and strike capabilities, adaptability to a wide speed range, and flexible flight trajectories. However, because their models are characterized by nonlinearity, strong coupling, and strong time-varying features, these vehicles are easily affected by external interference, impacting flight control accuracy and potentially leading to aerodynamic control surface malfunctions. Control surface jamming is the most significant failure type affecting aircraft handling performance and posing the greatest threat to flight safety. Specifically, some control surfaces become fixed at a certain deflection angle and fail to respond to control commands, directly resulting in reduced control redundancy and impaired control torque output. Control torque redistribution is the most commonly used fault-tolerant control method after control surface jamming. Therefore, research on wide-speed-range dynamic control torque redistribution methods for reusable launch vehicles under control surface jamming conditions is of great significance for improving the reliability, stability, and safety of aircraft.

[0003] The patent "A Fault-Tolerant Flight Control Method for Fixed-Wing UAV Control Surface Jamming" (CN114415515B) proposes a fractional-order backstepping sliding mode controller as the core, coupled with a control mixer to achieve command redistribution after a fault, aiming to solve the control surface jamming problem of small fixed-wing UAVs. However, it has significant limitations: it does not consider the differences in aerodynamic characteristics at different speeds, and cannot meet the requirements of wide-speed-range flight; the control mixer uses fixed weight allocation logic, which cannot dynamically adjust the priority of redundant control surfaces according to the fault, and attitude fluctuations are prone to occur in the early stage of the fault.

[0004] The patent "A Method for Adjusting the Jamming of Aircraft Tail and Elevator" (CN115946843B) proposes adding a rotatable spare horizontal tail to the original vertical tail, which, when deployed, is perpendicular to the original vertical tail. The aim is to reduce the risk of accidents caused by elevator jamming through hardware redundancy. However, it relies on an additional spare control surface, increasing the weight and complexity of the aircraft structure, and does not consider the differences in aerodynamic characteristics over a wide speed range. Furthermore, it only addresses elevator jamming and cannot cover scenarios where other critical control surfaces such as the rudder, flaps, and ailerons may jam, resulting in a narrow tolerance range.

[0005] The patent "An Aircraft Attitude Fault-Tolerant Control System and Control Method for Control Surface Faults" (CN115629547B) proposes a method that uses an adaptive multi-model observer to detect faults and combines nonlinear dynamic inverse with a control surface adjustment module to achieve fault-tolerant control. However, this method has significant shortcomings: aerodynamic parameters are not dynamically updated with Mach number, and the fault detection response is lagging in special ranges such as transonic speeds. In addition, the control allocation of this method relies on offline preset fault modes, which cannot optimize torque allocation in real time, resulting in low control accuracy under complex fault conditions.

[0006] The paper "Suppression and Recovery of Spiral Motion of UAV under Limited Rudder Jamming" ([J]. Firepower and Command Control, 2011, 36(04): 53-56.) treats rudder jamming as a constant disturbance, establishes the lateral motion equation of the UAV, and derives the control surface compensation for spiral motion suppression for both all-moving and differential horizontal tails. Although the method in the paper can handle certain faults and disturbances, it depends on a specific control surface configuration, has poor versatility for conventional multi-control surface aircraft, and does not propose a dynamic torque redistribution strategy, making the control accuracy susceptible to impact after a fault.

[0007] In existing technologies, the lack of a dynamic adaptive control torque redistribution mechanism after a control surface jamming failure makes it difficult to quickly adjust the control weights of redundant control surfaces, easily leading to control torque overshoot or undershoot. This can cause instability in the control system, especially under extreme flight conditions or unforeseen circumstances. Furthermore, existing control methods typically ignore the coupling effects of control surface dynamics and wide-speed-range disturbances under fault conditions, and the allocation strategies do not take into account both control surface physical constraints (deflection amount, deflection rate) and wide-speed-range aerodynamic uncertainties, resulting in insufficient robustness. Some allocation methods also rely on accurate aircraft dynamics models. Therefore, there is an urgent need to design a dynamic control torque redistribution method that adapts to wide-speed-range flight characteristics and can quickly respond to control surface jamming failures, improving the system's fault tolerance and robustness while ensuring control accuracy. Summary of the Invention

[0008] The purpose of this invention is to provide a control torque redistribution method for reusable launch vehicles facing control surface jamming, in order to solve the problems of poor adaptability and insufficient robustness of control torque distribution when the control surface of a wide-speed-range aircraft is jammed. The invention introduces a dynamic control distribution reconfiguration strategy that considers the synergistic effect of the control surfaces to achieve torque redistribution under fault conditions.

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

[0010] A method for redistributing control torque in reusable launch vehicles facing control surface jamming includes: constructing a six-degree-of-freedom model of the reusable launch vehicle and a control surface jamming fault model; detecting control surface jamming faults using an online recursive least squares algorithm; and designing a dynamic control allocation and reconfiguration strategy considering control surface synergy to achieve multi-control surface torque distribution. Details are as follows:

[0011] Step 1: Construct a six-degree-of-freedom model of the reusable launch vehicle.

[0012] The reusable launch vehicle adopts a lifting body configuration and an integrated airframe / engine design. Its six-degree-of-freedom rigid body model can fully reflect the typical operating conditions of wide-range flight. During the modeling process, the elastic deformation effect of the aircraft structure is ignored, and the sloshing effect of fuel and propellant is neglected. Only the rigid body motion characteristics are considered.

[0013] The six-degree-of-freedom rigid body model contains 12 state variables. With 3 control inputs ;in, and These are speed, track inclination angle, and track yaw angle, respectively. and These represent the displacements of the launch vehicle in the ground coordinate system; and The aircraft orbits the fuselage. and Angular velocity of the axis; and These are pitch angle, yaw angle, and roll angle, respectively. and These correspond to the left elevator deflection angle, right elevator deflection angle, and rudder deflection angle, respectively. The flight status of the aircraft can be controlled by adjusting the control input.

[0014] The dynamic equations for the motion of the center of mass are as follows:

[0015] (1)

[0016] In the formula, the superscript " " represents the first derivative; For engine thrust, For flight drag, For lift, It is a lateral force; For the mass of the aircraft, It is the gravitational constant. The distance from the spacecraft to the Earth's center; and These are the angle of attack, sideslip angle, and roll angle, respectively.

[0017] The specific kinematic equations for the motion of the center of mass are as follows:

[0018] (2)

[0019] The specific equations of motion about the center of mass are as follows:

[0020] (3)

[0021] In the formula, and These are the rolling moment, yaw moment, and pitch moment, respectively. and The moment of inertia of the aircraft's three axes. and This is the derivative of the moment of inertia with respect to time.

[0022] The specific kinematic equations for motion about the center of mass are as follows:

[0023] (4)

[0024] In the formula, and These are pitch angle, yaw angle, and roll angle, used to describe the aircraft's attitude state.

[0025] The calculation relationships between some physical quantities and the conversion logic between state quantities are as follows:

[0026] (5)

[0027] In the formula, For the Earth's radius, The altitude at which the aircraft flies; For the engine's specific impulse. This refers to the rate of change in quality. and These are the initial moment of inertia and the initial mass, respectively, used to describe the relationship between the moment of inertia and the mass.

[0028] Aerodynamics (lift) ,resistance Lateral force ) and aerodynamic torque (rolling torque) , yaw moment Pitch moment The calculation expression for ) is as follows:

[0029] (6)

[0030] in, For dynamic pressure, The atmospheric density at the flight altitude. For the aircraft reference area, and These are the horizontal and vertical reference lengths, respectively. , , These are the lift coefficient, drag coefficient, and lateral force coefficient, respectively. and These are the roll moment coefficient, yaw moment coefficient, and pitch moment coefficient, respectively. All of these aerodynamic coefficients are multivariable high-order polynomials relating to flight state parameters.

[0031] Step 2: Construction of the control surface jamming fault model

[0032] Control surface jamming refers to a situation where one or more control surfaces in the flight control system fail to move normally according to control commands or remain completely stationary due to mechanical failure, hydraulic system failure, or other reasons, resulting in a "jammed" state. This causes the aircraft to lose effective control in the corresponding direction, thereby seriously affecting flight safety and performance. The control surface jamming fault model is represented as follows:

[0033] (7)

[0034] in, Indicates in Time of the first The amount of control surface deflection output by each control surface feedback. The value for the control surface jamming is a constant.

[0035] Step 3: Surface deadlock fault detection based on online recursive least squares algorithm

[0036] Recursive least squares is not fundamentally different from ordinary least squares; they can be understood as different manifestations of the same algorithm. Recursive least squares does not require recalculating the information matrix for each data update; it only needs an initial value to achieve online recursion, reducing computational load and improving computational efficiency. Furthermore, the introduction of a "forgetting factor" in the recursive formula allows for the estimation of time-varying parameters, which is unnecessary for fixed parameters.

[0037] For parameter estimation, the relationship between the measured output and the model parameters is more important. If the output is given by the following equation, the model is called a linear parametric model, expressed as follows:

[0038] (8)

[0039] In the formula, It is the output of the model measurement. These are the basis vectors of the model to be identified. Vector to be identified.

[0040] When using the least squares (LS) form of the model, no information about Any uncertainty, i.e., no information about The probability expression. The least squares estimate can be obtained through the following corollary: for a given... , The “best” estimate is obtained by minimizing the weighted sum of the squared errors between the measured output and the model output:

[0041] (9)

[0042] In the formula, It is a weighted sum of the squared errors.

[0043] Generally speaking, as the amount of observational information increases, the accuracy of the estimate will become higher and higher and tend to stabilize, which is also one of the means to verify whether the estimate is accurate. However, as the amount of observational information increases, the computational workload also increases, and as can be seen from the following analysis, since each calculation requires all the information, the preceding calculation process is repeated continuously. To overcome this shortcoming, the recursive least squares (RLS) method is introduced.

[0044] To use recursive least squares, you must first calculate the initial value required for a recursion using the known observations and outputs.

[0045] Represents the time value, in the known time period Record all observations and outputs prior to time point [time]. Then the first Parameters required for identification at any time for:

[0046] (10)

[0047] In the formula, for The basis vectors of the model to be identified at time step. for The inverse matrix, for The transpose of . for The model's measured output at time t.

[0048] If the variance of the measured noise is , Indicates by OK The set of real matrices of columns; then the covariance matrix of the estimated values ​​is:

[0049] (11)

[0050] In the formula, These are the estimated values ​​of the parameters that need to be identified.

[0051] According to formula (10), firstly through the first The data from before time point provides initial values. Represents a positive natural number at a given moment. The information matrix is... Recursively starting from the first The moment begins:

[0052] (12)

[0053] (13)

[0054] (14)

[0055] In the formula, It is the intermediate matrix of the recursion. It is the identity matrix. For the first The estimated values ​​of the parameters to be identified at each time step. For the first The estimated values ​​of the parameters to be identified at each time step. for The time-based model measures the output.

[0056] Formulas (12) to (14) represent the recursive least squares process. In practical applications, as information increases, the positive definiteness of the information matrix continuously decreases, and its improvement effect on new information gradually becomes zero; this phenomenon is called data saturation. To solve this problem, a recursive algorithm incorporating a forgetting factor is proposed. To represent the forgetting factor, replace formulas (12) and (14) with:

[0057] (15)

[0058] (16)

[0059] The aircraft control surface jamming fault model can be described in the following form:

[0060] (17)

[0061] in, Indicates the discrete time step. This is the actual output of the control surface deflection. It is the desired rudder deflection. This is the output gain (normally set to 1 by default). This is a constant deviation (normally 0 by default). When the control surface jams, The value will be 0. This will be the rudder deflection value when the rudder surface is stuck.

[0062] In order to estimate output gain online Deviation from Hang Seng Index The value is identified in real time using the online recursive least squares method. The specific process is as follows:

[0063] when Initialize model parameters:

[0064] (18)

[0065] in, These are model parameters. These are the initial values ​​for output gain and constant deviation.

[0066] Initialize the covariance matrix:

[0067] (19)

[0068] in, It is an information matrix. These are the initialization coefficients of the information matrix.

[0069] In this invention This is the input vector at the current moment. By updating online according to the method in step 3, it is possible to identify in real time whether the control surface has jammed and to estimate the control surface jamming deflection value.

[0070] Step 4: Dynamic control allocation reconfiguration strategy considering rudder surface synergy.

[0071] After detecting the fault based on the method in step 3, a dynamic control allocation and reconfiguration strategy is designed to achieve fault-tolerant control of this fault.

[0072] Control surface synergy refers to the interaction and cooperation among multiple control surfaces in controlling an aircraft's attitude. Higher synergy indicates enhanced control effects between control surfaces, reducing redundancy or conflict and improving aircraft handling performance. Therefore, an index is defined to measure the control surface synergy of reusable launch vehicles. Its expression is as follows:

[0073] (20)

[0074] in, and Indicates the first The and the first The control efficiency vector of each control surface is related to the control surface's ability to control the aircraft's attitude (pitch, roll, yaw). The aircraft has a total of Each rudder surface. In equation (20), the dot product... The dot product reflects the projection relationship between the two control surfaces in the control direction, indicating the intensity of their action regardless of whether the directions are consistent. If the control directions of the two control surfaces are consistent, the dot product is positive and large; if the directions are opposite, the dot product is negative, but the absolute value may still be large, indicating potential redundancy or cancellation effect. This indicates that the dot product is normalized, restricting the result to a certain range. The range is then determined. Finally, the synergistic effects of all pairwise combinations of control surfaces are summed up. The larger the final value, the stronger the overall synergy of the control surface system.

[0075] As mentioned above, the control surface synergy index evaluates the synergy efficiency of the control surface system by quantifying the consistency of control directions between control surfaces. A higher value indicates that the control effects between control surfaces are mutually reinforcing. In the design of control surface allocation for reusable launch vehicles, maximizing this synergy dynamically adjusts control surface deflection to ensure efficient control surface synergy.

[0076] To enable more precise control distribution of the control surfaces after a malfunction, a nonlinear optimization method is employed to model the torque distribution problem of reusable launch vehicles as a constrained optimization problem. This involves constructing a suitable objective function, defining the optimal criterion for torque distribution, and introducing physical constraints to calculate the optimal actuator result. The nonlinear optimization method used in this invention is sequential quadratic programming (SQP), an iterative algorithm that decomposes a nonlinear optimization problem into a series of quadratic programming problems. SQP is primarily based on the Karush-Kuhn-Tucker (KKT) conditions of nonlinear programming, combined with Newton's method. By linearizing the objective function and constraints and applying quadratic approximations, it gradually approximates the global optimum. The form of SQP can be expressed as follows:

[0077] (twenty one)

[0078] in, It is the objective function. and These are equality and inequality constraints, respectively, meaning the final solution is obtained by solving the above optimization problem.

[0079] When designing the control allocation for a reusable launch vehicle according to the standard form of nonlinear optimization using formulas (20) and (21), the first step is to construct an optimization problem based on the physical background. The input to the control allocation is the output of the control law, i.e., the three-axis torque command, and the output of the control allocation is the deflection angle of each control surface. Next, the optimization objective is determined. Considering the characteristics of the reusable launch vehicle studied in this invention, the optimization objective includes a torque allocation error index and a control surface synergy effect index.

[0080] The final optimization objective is as follows:

[0081] (twenty two)

[0082] In the formula, It is the actual output torque. It is the expected torque. It is the rudder surface coordination index coefficient. It is the maximum value of the set rudder surface coordination index. It is the torque tracking error of the aircraft, and the torque can be obtained by reconstructing and reusing the aerodynamic model of the launch vehicle; It is a symmetric of rudder surface synergy, achieved by minimizing... To maximize .

[0083] Finally, considering that reusable vehicles also have upper and lower bound constraints, the variables to be optimized are constrained as follows:

[0084] (twenty three)

[0085] In the formula, and These are the minimum and maximum deflection constraint values ​​for the left elevator, respectively. and These are the minimum and maximum deflection constraint values ​​for the right elevator, respectively. and These are the minimum and maximum rudder deflection constraint values, respectively.

[0086] In summary, a nonlinear optimization problem for torque control distribution in reusable launch vehicles was established. By optimizing the solution based on the designed objective function and constraints, the control surface deflection command that satisfies the current torque command can be quickly obtained.

[0087] The beneficial effects of this invention are:

[0088] This invention addresses the control challenges posed by control surface jamming in reusable launch vehicles by proposing a control surface jamming fault detection based on the recursive least squares method and a dynamic control allocation and reconfiguration strategy that considers control surface synergy, effectively overcoming the shortcomings of existing technologies.

[0089] This invention is adaptable to a wide speed range of flight characteristics. It constructs a six-degree-of-freedom model using aerodynamic data, fully considering the differences in aerodynamic characteristics across the wide speed range. This avoids control mismatch caused by neglecting the influence of the speed range in traditional methods, ensuring control stability across the entire speed range. It achieves rapid and accurate detection of control surface jamming faults, employing a recursive least squares algorithm with a forgetting factor. It does not rely on offline preset fault modes and can identify faulty control surfaces and jammed deflection values ​​online in real time. This solves the data saturation problem of ordinary least squares methods and improves fault response speed, gaining critical reaction time for subsequent torque reallocation and preventing the fault from escalating.

[0090] This invention achieves significant results in dynamically optimizing control torque allocation. By defining a control surface synergy effect index to quantify the consistency of control direction between control surfaces, and combining it with a sequential quadratic programming method to construct a nonlinear optimization objective, it maximizes control surface synergy efficiency while minimizing torque tracking error. It also takes into account physical constraints such as control surface deflection and deflection rate, effectively avoiding control torque overshoot or undershoot. This solves the problem that traditional fixed weight allocation cannot be dynamically adjusted, and provides a reliable guarantee for safe and stable flight of wide-speed-range reusable launch vehicles under control surface jamming failure. Attached Figure Description

[0091] Figure 1 This is a technical block diagram of a reusable launch vehicle control torque redistribution method for navigating control surface jamming.

[0092] Figure 2 This is a block diagram of the fault detection technology based on recursive least squares of the present invention;

[0093] Figure 3 It is a speed-time curve of a reusable launch vehicle;

[0094] Figure 4 It is a graph of the altitude-time curve of a reusable launch vehicle;

[0095] Figure 5 It is a pitch angle-time curve of a reusable launch vehicle;

[0096] Figure 6 It is a graph of the angle-of-attack time of a reusable launch vehicle;

[0097] Figure 7 It is a roll angle-time curve of a reusable launch vehicle;

[0098] Figure 8 It is a side-slip angle-time curve of a reusable launch vehicle;

[0099] Figure 9 This is a time curve of the left elevator deflection angle of a reusable launch vehicle.

[0100] Figure 10 This is a time curve of the right elevator deflection angle of a reusable launch vehicle;

[0101] Figure 11 It is a time curve of the rudder deflection angle of a reusable launch vehicle; Detailed Implementation

[0102] The embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0103] like Figure 1 As shown, the present invention provides a method for redistributing control torque in a reusable launch vehicle facing rudder surface jamming, comprising:

[0104] Step 1: Construct a six-degree-of-freedom model of the reusable launch vehicle, which can fully reflect the typical operating conditions of wide-range flight.

[0105] Step 2: Constructing the control surface jamming fault model to obtain a mathematical model of the control surface jamming state;

[0106] Step 3: Based on the online least squares recursive algorithm, a "forgetting factor" is introduced into the recursive formula to estimate time-varying parameters, identify in real time whether the control surface is stuck, and estimate the control surface stuck deflection value.

[0107] Step 4: Design a dynamic control allocation reconfiguration strategy that considers the synergistic effect of the control surfaces to achieve rapid redistribution of control torque.

[0108] Specifically, the embodiments of the present invention are described as follows:

[0109] Step 1: Construct a six-DOF model of the reusable vehicle. This six-DOF rigid body model contains 12 state variables. With 3 control inputs The definitions of the main variables are given below: and These are speed, track inclination angle, and track yaw angle, respectively. For engine thrust, For flight drag, For lift, It is a lateral force; For the mass of the aircraft, It is the gravitational constant. The distance from the spacecraft to the Earth's center; and These are the angle of attack, sideslip angle, and roll angle, respectively. and These represent the displacements of the launch vehicle in the ground coordinate system. and The aircraft orbits the fuselage. and Angular velocity of the axis; and These are the rolling moment, yaw moment, and pitch moment, respectively. and The moment of inertia of the aircraft's three axes. and This is the derivative of the moment of inertia with respect to time. and These are pitch angle, yaw angle, and roll angle, respectively. For the Earth's radius, The altitude at which the aircraft flies; For the engine's specific impulse. This refers to the rate of change in quality. and These are the initial moment of inertia and the initial mass, respectively. For dynamic pressure, The atmospheric density at the flight altitude. For the aircraft reference area, and These are the horizontal and vertical reference lengths, respectively. , , These are the lift coefficient, drag coefficient, and lateral force coefficient, respectively. and These are the roll moment coefficient, yaw moment coefficient, and pitch moment coefficient, respectively. Combining dynamics and kinematics, the six-degree-of-freedom model of the reusable launch vehicle is represented as follows:

[0110] (twenty four)

[0111] (25)

[0112] (26)

[0113] (27)

[0114] (28)

[0115] (29)

[0116] Step 2: Construction of the Control Surface Jamming Fault Model. Control surface jamming refers to a situation where one or more control surfaces in the flight control system fail to move normally according to control commands or remain completely stationary due to mechanical failure, hydraulic system failure, or other reasons, resulting in a "jammed" state. This causes the aircraft to lose effective control in the corresponding direction, severely impacting flight safety and performance. In its mathematical model, Indicates in Time of the first The control surface deflection output by each control surface feedback. The value at which the control surface is stuck is a constant. The mathematical model is expressed as follows:

[0117] (30)

[0118] Step 3: Surface jamming fault detection based on online recursive algorithm to achieve rapid detection of control surface jamming faults. Recursive least squares does not require recalculating the information matrix for each data update; only an initial value is needed for online recursion, reducing computational load and improving efficiency. Furthermore, introducing a "forgetting factor" into the recursive formula allows for estimation of time-varying parameters. Specifically: For parameter estimation, the relationship between the measured output and model parameters is more important. If the output is given by the following equation, the model is called a linear parametric model:

[0119] (31)

[0120] In the formula, It is the output of the model measurement. These are the basis vectors of the model to be identified. Vector to be identified.

[0121] Use recursive least squares (e.g.) Figure 2 As shown in the figure, the first step is to use the known observations and outputs to calculate an initial value required for the recursion. Represents the time value, in the known time period Record all observations and outputs prior to time point [time]. Then the first Parameters required for identification at any time for:

[0122] (32)

[0123] In the formula, for The basis vectors of the model to be identified at time step. for The inverse matrix, for The transpose of . for The model's measured output at time t.

[0124] If the variance of the measured noise is , Indicates by OK The set of real matrices of columns; then the covariance matrix of the estimated values ​​is:

[0125] (33)

[0126] In the formula, These are the estimated values ​​of the parameters that need to be identified.

[0127] According to formula (32), firstly through the first The data from before time point provides initial values. Represents a positive natural number at a given moment. The information matrix is... recursively from Starting from a given time, a recursive algorithm is introduced, incorporating the forgetting factor:

[0128] (34)

[0129] (35)

[0130] (36)

[0131] The aircraft control surface jamming fault model can be described in the following form:

[0132] (37)

[0133] in, Indicates the discrete time step. This is the actual output of the control surface deflection. It is the desired rudder deflection value, where, This is the output gain (normally set to 1 by default). This is a constant deviation (normally 0 by default). When a control surface jams, The value will be 0. This will be the rudder deflection value when the rudder surface is stuck.

[0134] In order to estimate output gain online Deviation from Hang Seng Index The value is identified in real time using an online recursive least squares method. The specific process is as follows:

[0135] when Initialize model parameters:

[0136] (38)

[0137] in, These are model parameters. These are the initial values ​​for output gain and constant deviation.

[0138] Next, initialize the covariance matrix.

[0139] (39)

[0140] in, It is the covariance matrix. These are the initialization coefficients of the information matrix, set to 10. 4 ; It is an identity matrix.

[0141] In this invention This is the input vector at the current moment. By updating online using the method described above, it is possible to identify in real time whether the control surface has jammed and to estimate the jammed deflection value.

[0142] Step 4: Consider the dynamic control allocation and reconfiguration strategy based on the control surface synergy effect. The control surface synergy effect refers to the interaction and cooperation capability of multiple control surfaces when controlling the attitude of an aircraft. Higher synergy indicates that the control effects between control surfaces are mutually enhanced, reducing redundancy or conflict and improving the aircraft's handling performance. Therefore, an index is defined to measure the control surface synergy effect of reusable launch vehicles, and its expression is as follows:

[0143] (40)

[0144] In the formula: and Indicates the first The and the first The control efficiency vector of each control surface, the aircraft has a total of Each control surface. Its control capability is related to the control surface's ability to control the aircraft's attitude (pitch, roll, yaw). In equation (39) above, the dot product... The dot product reflects the projection relationship between the two control surfaces in the control direction, indicating the intensity of their action regardless of whether the directions are consistent. If the control directions of the two control surfaces are consistent, the dot product is positive and large; if the directions are opposite, the dot product is negative, but the absolute value may still be large, indicating potential redundancy or cancellation effect. This indicates that the dot product is normalized, restricting the result to a certain range. The range is then determined. Finally, the synergistic effect of each pair of control surfaces is summed; the larger the final value, the stronger the overall synergy of the control surface system. In the design of control surface allocation for reusable launch vehicles, maximizing this value dynamically adjusts control surface deflection, ensuring efficient control surface synergy.

[0145] To enable more precise control distribution of the control surfaces after a malfunction, a nonlinear optimization method is employed to model the torque distribution problem of reusable launch vehicles as a constrained optimization problem. By constructing a suitable objective function, the optimal criterion for torque distribution is defined. The nonlinear optimization method used in this invention is sequential quadratic programming (SQP), an iterative algorithm that decomposes a nonlinear optimization problem into a series of quadratic programming problems. SQP is primarily based on the Karush-Kuhn-Tucker (KKT) conditions of nonlinear programming, combined with the idea of ​​Newton's method. By linearizing and quadratically approximating the objective function and constraints, it gradually approximates the global optimum. The SQP form can be expressed as follows:

[0146] (41)

[0147] In the formula: It is the objective function. and These are equality and inequality constraints, respectively, meaning the final solution is obtained by solving the above optimization problem.

[0148] The final optimization objective is as follows:

[0149] (42)

[0150] In the formula, It is the actual output torque. It is the expected torque. This is the rudder surface coordination index coefficient, which is set to 1.5; This is the maximum value of the set rudder surface coordination index, which is set to 2. (First item) This refers to the torque tracking error of the aircraft. The torque is obtained by reconstructing and reusing the aerodynamic model of the launch vehicle. (The second term...) It is a rudder surface synergy index, therefore the second term represents the effect achieved by minimizing... To maximize .

[0151] Finally, considering that reusable vehicles also have upper and lower bound constraints, the variables to be optimized are constrained as follows:

[0152] (43)

[0153] In the formula: and These represent the minimum and maximum deflection constraints of the left elevator, respectively, and are set to -20° and 20°. and These represent the minimum and maximum deflection constraints of the right elevator, respectively, and are set to -20° and 20°. , These represent the minimum and maximum rudder deflection constraints, respectively, and are set to -20° and 20°.

[0154] To verify the feasibility of the proposed control torque redistribution method for reusable launch vehicles under control surface jamming conditions, the following verification simulation was conducted. In the simulation scenario, the reference command consisted of two basic maneuvering processes: at 2 seconds, a left elevator control surface jamming fault was set; at 10 seconds, the roll angle command was set to 10°. The initial state and expected values ​​of the aircraft in the simulation experiment are given in Table 1.

[0155] Table 1 Initial and Expected Values ​​of Flight Status

[0156]

[0157] The relevant simulation results are as follows Figures 3 to 9 As shown. Figure 5 and Figure 7 This demonstrates that the reusable launch vehicle control torque redistribution control method designed in this invention under control surface jamming can achieve rapid tracking of pitch and roll angle reference commands, and exhibits small tracking errors. Furthermore, Figure 3 , Figure 4 , Figure 6 and Figure 8 The curves showing the changes in speed, altitude, angle of attack, and sideslip angle over time demonstrate that the attitude angles of the reusable launch vehicle exhibit a smooth effect and minimal vibration. Figures 9 to 11 The deflection angle-time curves of the left and right elevators and rudders of the reusable launch vehicle are presented, showing the distribution of other control surfaces after a left elevator surface jamming. In summary, under the control torque redistribution control method proposed in this invention, the reusable launch vehicle can achieve satisfactory flight performance even when control surfaces are jammed.

Claims

1. A method for control moment redistribution for a reusable vehicle in the event of a stuck control surface, the method comprising: Comprise: Step 1, build a six degree of freedom model of reusable launch vehicle; Step 2, build a model of stuck actuator failure; Step 3, stuck actuator failure detection based on online recursive least squares algorithm; Step 4, dynamic control allocation reconfiguration strategy considering actuator synergy effect; After the method of step 3 detects the failure, a dynamic control allocation reconfiguration strategy is designed to achieve fault-tolerant control of the failure; Step 4 is as follows: Define an index for measuring the synergistic effect of reusable vehicle control surfaces The expression is as follows: (20) ； where, and denote the control effectiveness vectors of the th and the th control surface, respectively, and the aircraft has control surfaces; in equation (20), the dot product reflects the projection relationship of the two control surfaces in the control direction, which indicates the strength of the action of the two; denotes the normalization of the dot product, which limits the result to the interval; finally, the synergistic effects of all the control surfaces in pairs are accumulated, and the larger the final value is, the stronger the overall synergism of the control surface system is. A nonlinear optimization method is used to model the torque allocation problem of reusable launch vehicle as a constrained optimization problem, and by reasonably constructing the objective function, the optimal criteria of torque allocation are defined, and physical constraints are introduced to solve the optimal actuator results; The nonlinear optimization method used is sequential quadratic programming SQP, and the form of SQP is as follows: (21) ； wherein, is the objective function, and are equality and inequality constraints, respectively; When the standard form of nonlinear optimization according to formulas (20) and (21) is used to design the control allocation of reusable launch vehicle, first, the optimization problem needs to be constructed according to the physical background, the input of control allocation is the output of control law, that is, three-axis torque command, and the output of control allocation is the deflection angle of each actuator; Then determine the optimization objective, when designing the optimization objective, it contains torque allocation error index and actuator synergy effect index; The final optimization objective is as follows: (22) ； wherein, is the actual output moment, is the desired moment, is the rudder coordination index coefficient, is the set maximum value of the rudder coordination index; is the moment tracking error of the aircraft, wherein the moment is obtained by reconstructing the aerodynamic model of the reusable launch vehicle; is the rudder coordination effect index, which is maximized by minimizing ; and ; Finally, since there are upper and lower limits for the reusable launch vehicle, the optimization variables are constrained as follows: (23) ； wherein and are the left elevator deflection minimum and maximum constraint values, respectively; and are the right elevator deflection minimum and maximum constraint values, respectively; and are the rudder deflection minimum and maximum constraint values, respectively.

2. The reusable vehicle control-moment-reddistri bution method for rudder surface jamming with the characteristics of claim 1, wherein, Step 1 is as follows: Six degrees of freedom rigid body model contains 12 state variables with 3 control inputs ; wherein and are velocity, path tilt angle and path yaw angle, respectively; and are the displacements of the vehicle in the ground coordinate system, respectively; and are the angular velocities of the vehicle about the body and axes, respectively; and are the pitch, yaw and roll angles, respectively; and correspond to the left elevator deflection, right elevator deflection and rudder deflection, respectively; the control of the flight state of the vehicle can be achieved by adjusting the control inputs; The dynamic equation of the center of mass motion is as follows: (1) ； where the superscripts denote first derivatives; is the engine thrust, is the flight drag, is the lift, is the side force; is the aircraft mass, is the gravitational constant, is the distance of the aircraft to the earth center; and are the angle of attack, the sideslip angle and the roll angle, respectively; The kinematic equation of the center of mass motion is as follows: (2) ； The dynamic equation of the center of mass motion is as follows: (3) ； wherein and are the roll, yaw and pitch moments, respectively; and are the three-axis rotational inertia of the aircraft, and are the time derivatives of the rotational inertia. The kinematic equation of the center of mass motion is as follows: (4) ； wherein and are the pitch, yaw and roll angles, respectively, used to describe the body attitude state; The calculation relationship of some physical quantities and the conversion logic between state variables are as follows: (5) ； wherein, is the radius of the earth, is the flight altitude of the aircraft; is the specific impulse of the engine, is the mass change rate; and are the initial moment of inertia and the initial mass, respectively, for describing the relationship between the moment of inertia and the mass. The calculation expressions of aerodynamic forces including lift , drag , side force and aerodynamic moments including roll moment , yaw moment , pitch moment are as follows: (6) ； wherein, is the dynamic pressure, is the atmospheric density at the flight altitude, is the reference area of the aircraft, and are the lateral and longitudinal reference lengths, respectively; , , are the lift, drag and side force coefficients, respectively, and are the roll, yaw and pitch moment coefficients, respectively.

3. The reusable vehicle control-moment-reconfiguration method for rudder- surface jamming of claim 1, wherein, Step 2 is as follows: The stuck actuator failure model is represented as follows: (7) ； wherein, represents the time at which the moment the deflection of the control surface, is the value at which the control surface is stuck.

4. The reusable vehicle control-moment-reddistri bution method for rudder surface jamming with the method as claimed in claim 1, characterized in that, Step 3 is as follows: For parameter estimation, if the output is given by the following equation, the model is called a linear parameter model, which is represented as follows: (8) ； wherein is a model measurement output quantity, is a model measurement basis vector to be identified, is a vector to be identified; The least squares estimate is obtained by the following inference: for a given , The estimate of is obtained by minimizing the weighted sum of the squared errors between the measured output and the model output: (9) ； wherein is a weighted sum of error squares; Using recursive least squares, first, use the known observations and output quantities to calculate the initial value required for recursion; representing the time value, all observations and outputs up to a known time are recorded The first identified parameters required at time are: (10) ； wherein is the model measurement basis vector to be recognized at the time point, is the inverse matrix of is the transpose matrix of is the model measurement output at the time point If the variance of the measurement noise is , denotes the set of real matrices consisting of rows columns; then the covariance matrix of the estimates is (11) ； In the formula, is an estimated value of the parameter to be identified; According to formula (10), first, the initial value is provided by the first data before the time point, represents a positive natural number at a time point; the information matrix is , and the recursion starts from the first time point. (12) ； (13) ； (14) ； In the formula, It is the intermediate matrix of the recursion. It is the identity matrix. For the first The estimated values ​​of the parameters to be identified at each time step. For the first The estimated values ​​of the parameters to be identified at each time step. for The time-of-flight model measures the output. A recursive algorithm with a forgetting factor is proposed, For the forgetting factor, replace formula (12) and formula (14) with formula (15) and formula (16), respectively. (15) ； (16) ； The stuck actuator failure model of the aircraft is described as follows: (17) ； wherein, denotes a discrete time step, is the actual output rudder deflection, is the desired rudder deflection, is the output gain, is a constant bias when a rudder has a stuck fault, the value of will be 0, will be the rudder deflection value when the rudder is stuck; In order to be able to estimate the output gain and the constant bias online, real-time identification is performed using an online recursive least squares method, as follows: When the model parameters are initialized: (18) ； wherein, are model parameters, are initial values of the output gain and constant bias; Initialize the covariance matrix: (19) ； wherein is an information matrix, is an information matrix initialization coefficient; is the input vector at the current time; online updating according to the method of step 3, that is, real-time identification of whether the rudder surface has a jamming failure and estimation of the rudder surface jamming deflection value.

Images