A Transition Segment Control Method for Aircraft Based on Adaptive Disturbance Rejection and Dynamic Allocation

Abstract

This invention relates to an aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation, comprising: acquiring the real-time state of the aircraft through flight state sensing and extended state observers, and outputting controlled state estimates, controlled state rate of change estimates, and total composite disturbance estimates; generating reference controlled state commands based on the current rotor tilt angle and airspeed; constructing attitude and altitude tracking controllers based on adaptive sliding mode control laws, and updating controller parameters online in real time according to the estimated values ​​of airspeed and disturbances to calculate virtual control commands; and employing a dynamic control allocation mechanism to smoothly map the virtual control commands to rotor speed and fixed-wing aerodynamic control surfaces according to airspeed and tilt angle, thereby achieving smooth control of the aircraft during the transition phase. This invention can compensate for composite disturbances online during the transition phase and mitigate the actuator saturation risk under fixed control allocation, thereby improving the performance of altitude holding, control allocation, and online calculation during the transition phase.

Description

Technical Field

[0001] This invention belongs to the field of aircraft flight control technology, specifically relating to an adaptive disturbance rejection control and dynamic control allocation method and system for tiltrotor aircraft during the transition between vertical takeoff and landing (VTOL) and forward flight modes. More specifically, it is an aircraft transition control method based on adaptive disturbance rejection and dynamic allocation. This method can improve the adaptability of flight control systems to complex nonlinear dynamics and is mainly applied in fields such as urban air traffic (UAM), unmanned aerial vehicle (UAV) logistics, and high-altitude long-endurance reconnaissance. Background Technology

[0002] Tiltrotor-type vertical takeoff and landing (VTOL) aircraft combine the advantages of helicopter VTOL and fixed-wing aircraft high-speed cruise, playing a crucial role in modern aviation services and management. However, during the "transition phase" from hovering mode to fixed-wing forward flight mode, the aerodynamic characteristics of the aircraft exhibit strong nonlinearity, strong coupling, and rapid time-varying features. The rotor tilting during this transition phase causes a dynamic exchange of lift between rotor thrust and wing aerodynamic lift, making traditional control methods based on small-disturbance linearization models and fixed-gain (such as conventional PID control) ineffective.

[0003] With the continuous development of urban air mobility concepts, the demand for high-precision control of aircraft in complex micro-meteorological environments is becoming increasingly urgent. Existing transition phase flight control methods mainly rely on gain scheduling strategies or simple robust control techniques, which are difficult to effectively cope with sudden gust interference and drastic changes in aerodynamic parameters caused by rotor downwash. Especially in complex scenarios such as shear winds and strong turbulence brought about by dense urban building clusters, the control accuracy and disturbance rejection performance of traditional methods are significantly reduced, which can easily lead to abnormal altitude loss, attitude divergence, or even loss of control of the aircraft, thereby seriously affecting flight safety, passenger comfort, and flight path execution accuracy.

[0004] Currently, flight control technologies for single flight modes (pure hovering or pure fixed-wing cruise) are relatively mature. However, for complex multiple-input multiple-output (MIMO) nonlinear systems in the transition phase of aircraft, existing methods still have certain limitations in terms of composite disturbance suppression, smooth allocation of control authority across multiple control surfaces, and real-time computational efficiency. Smooth and safe transition phase control is one of the key core technologies in the field of aircraft control, directly related to the aircraft's envelope extension capability, availability under complex weather conditions, and overall operational safety. This invention addresses the technical challenges of aircraft transition phase control by proposing a flight control method that combines extended state observation and adaptive sliding mode control, providing a new technical approach for the field of intelligent aircraft control. Summary of the Invention

[0005] The purpose of this invention is to provide an aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation. This invention utilizes a nonlinear extended state observer (ESO) to perform real-time lumped estimation and compensation of unmodeled internal dynamics and external wind field disturbances, serving as the feedforward input for the control law. It proposes an adaptive sliding mode controller (ASMC), designing an adaptive update law for switching gain based on Lyapunov stability theory to suppress chattering and improve the tracking capability for nonlinear dynamics. A dynamic control allocation module (DCA) is designed to dynamically adjust the control effectiveness weights of the rotor and aerodynamic control surfaces according to real-time dynamic pressure and tilt angle, achieving a smooth transition of control authority. The algorithm of this invention exhibits strong robustness and moderate computational complexity, making it suitable for airborne flight control computers with limited computing resources.

[0006] The technical solution adopted by this invention to achieve the above objectives is: a flight transition control method based on adaptive disturbance rejection and dynamic allocation, comprising the following steps:

[0007] The flight state sensing unit acquires the real-time state of the aircraft, and the extended state observer is used to estimate the total composite disturbance of the aircraft's flight control system in the transition phase in real time, thus obtaining the controlled state estimate. Estimated rate of change of controlled state and the estimated total combined disturbance ;

[0008] Based on the rotor tilt angle obtained through the tilt mechanism encoder Compared with the flight airspeed obtained through the airspeed measurement channel The trajectory planner generates a dynamic reference flight trajectory and attitude control state commands for the transition segment. ;

[0009] An adaptive sliding mode controller is used, combined with attitude control state commands. Controlled state estimate Estimated rate of change of controlled state Combined total disturbance estimate The virtual control command vector is obtained. And adaptively update the control gain;

[0010] Using a dynamic control assigner, virtual control command vectors are... Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surfaces to achieve transition control of the aircraft.

[0011] The method of using an extended state observer to estimate the total composite disturbance of the aircraft flight control system in the transition phase in real time includes the following steps:

[0012] The flight status perception module acquires the aircraft's three-axis attitude angles, angular rates, position, and body velocity components, forming a sensor fusion measurement vector. And extract the controlled state vector actually used by the controller from it. ; These represent altitude, roll angle, pitch angle, and yaw angle, respectively. These are the aircraft's roll, pitch, and yaw angular velocities, respectively. For position coordinates, For the velocity components of the machine system;

[0013] Extended state observer will measure the controlled state value Compared with the controlled state estimate The difference between them is defined as the output estimation error. Aerodynamic parameter perturbations, external gust disturbances, and inter-channel coupling are all uniformly equivalent to a total composite disturbance. Its estimated value is And set the corresponding expansion state, i.e. ;

[0014] The extended state observer estimates the error based on the output. Correct the observed state and output the estimated value of the controlled state. Estimated rate of change of controlled state and the estimated total combined disturbance and will , and Output to the adaptive sliding mode controller.

[0015] The system employs an adaptive sliding mode controller, combined with attitude control state commands. Combined total disturbance estimate This yields virtual control command vectors for attitude and altitude correction. And adaptively update the control gain, including the following steps:

[0016] Based on the controlled state estimate output by the extended state observer The reference controlled state vector output by the trajectory planner Calculate the tracking error: ;

[0017] And based on the estimated rate of change of the controlled state output by the extended state observer Rate of change of the reference controlled state vector Calculate the error derivative: ;

[0018] Constructing the integral sliding surface: ;

[0019] The sliding mode control law is expressed as:

[0020]

[0021] in, , This is a virtual command for total thrust. , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively. The nominal nonlinear dynamic function is known. This is the estimated total composite disturbance. To control the input gain coefficient; , These are the parameters of the sliding surface; For adaptively updated switching gain; function; This represents the boundary layer thickness.

[0022] The adaptive update control gain, its switching gain The adaptive regulation law is:

[0023]

[0024] Additional constraints: When season , These are positive constants to ensure that the switching gain is always positive; among them, The positive learning rate constant, This is a forgetting factor used to prevent excessive gain expansion. The absolute value of the sliding surface.

[0025] The method utilizes a dynamic control allocator to distribute virtual control command vectors. Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surface actuators to achieve transition control of the aircraft, specifically including:

[0026] Determining the rotor control effectiveness matrix based on the rotor thrust model The aerodynamic control surface performance matrix is ​​determined based on the aerodynamic surface aerodynamic model. ;

[0027] According to the rotor tilt angle Calculate the transition weight function And according to the flight airspeed Calculate the dynamic pressure efficiency factor ,Will and The fusion yields the global control performance matrix:

[0028]

[0029] Receive the virtual control command vector output by the adaptive sliding mode controller: Where T is the virtual command for total thrust. , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively.

[0030] The vector of the actual control quantity to be determined in the dynamic control distributor is defined as:

[0031]

[0032] in, For the first The actual control quantity of each rotor channel For the first The actual control quantity of each aerodynamic control surface channel. The number of rotor channels to be allocated. The number of aerodynamic control surface channels to be allocated;

[0033] First according to and Calculate the initial value of the bias-weighted pseudo-inverse assignment. : ,in, express The bias-weighted pseudo-inverse matrix is ​​then determined, and the initial value of the bias-weighted pseudo-inverse is assigned. Whether to trigger actuator amplitude and rate constraints;

[0034] If not triggered, And output; if triggered, with As the initial iteration value or reference initial value for solving the quadratic programming problem, the following constrained quadratic optimization problem is solved:

[0035]

[0036] in, This is the channel priority weight matrix. and Limiting the physical amplitude of the actuator. Limit the maximum speed of the actuator. This represents the actual control quantity from the previous control cycle. To control the cycle;

[0037] The solution obtained This is the optimal solution for the combination of rotor speed and aerodynamic control surface deflection angle, and its components include the optimal execution commands for each rotor channel. Optimal execution commands for each aerodynamic control surface channel The output is then sent from the dynamic control distributor to the rotor motor and aerodynamic control surface actuator.

[0038] A high-priority protection mechanism is set up between the trajectory planner and the dynamic control allocator, specifically including:

[0039] Based on the height estimate output by the extended state observer Reference height output by the trajectory planner Calculate the height channel error: ;

[0040] when When the aircraft is found to have exceeded a preset threshold, it is determined that the aircraft has exceeded the threshold. High negative deviation; trajectory planner generates and references the controlled state vector Same-dimensional height subsidence compensation vector: ;in, It is the height compensation amount, and is determined based on the portion of the negative height deviation that exceeds the threshold.

[0041] The trajectory planner superimposes the height sink compensation vector onto the reference controlled state vector to obtain the corrected reference controlled state vector:

[0042]

[0043] in, For smooth transition interpolation factor, This represents the baseline controlled state vector for the hovering mode. Represents the baseline controlled state vector for forward flight mode;

[0044] Simultaneously, the dynamic control allocator increases the weights corresponding to the total thrust channels in the channel priority weight matrix W, so that the weight matrix satisfies:

[0045]

[0046] in, For the total thrust channel weight, , , These are the weights for the roll, pitch, and yaw moment channels, respectively; the dynamic control distributor solves for the optimal actual control quantity. The updated version is used in the aforementioned quadratic optimization problem. .

[0047] An aircraft transition phase control system based on adaptive disturbance rejection and dynamic allocation includes:

[0048] An extended state observer is used to estimate the total composite disturbance of the aircraft's flight control system during the transition phase based on the real-time aircraft state acquired through the flight state sensing unit, thereby obtaining the controlled state estimate. Estimated rate of change of controlled state and the estimated total combined disturbance ;

[0049] A trajectory planner is used to calculate the rotor tilt angle based on the rotor tilt angle obtained through the tilt mechanism encoder. Compared with the flight airspeed obtained through the airspeed measurement channel The trajectory planner generates a dynamic reference flight trajectory and attitude control state commands for the transition segment. ;

[0050] Adaptive sliding mode controller, used to combine attitude control state commands Controlled state estimate Estimated rate of change of controlled state Combined total disturbance estimate The virtual control command vector is obtained. And adaptively update the control gain;

[0051] Dynamic control assigner, used to assign virtual control command vectors Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surfaces to achieve transition control of the aircraft.

[0052] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation.

[0053] The present invention has the following beneficial effects and advantages:

[0054] 1. This invention uses an extended state observer to estimate the total composite disturbance in real time and incorporates the estimation result into an adaptive sliding mode control law. This can mitigate altitude drop and attitude deviation when aerodynamic characteristics change during the transition phase and external disturbances are present. Based on the current semi-physical simulation embodiment, the complete control link exhibits a smaller altitude drop trend compared to disabling the ESO or using a conventional PID baseline.

[0055] 2. The dynamic control allocation mechanism designed in this invention is based on the rotor tilt angle. and flight airspeed Dynamically integrating rotor and aerodynamic control surface performance can alleviate actuator saturation and control authority conflicts that easily occur during the transition phase in fixed allocation methods. In the current semi-physical simulation embodiment, the complete link using DCA exhibits a lower actuator saturation rate compared to the control group without dynamic allocation;

[0056] 3. This invention constructs a hierarchical control architecture based on adaptive sliding mode, extended state observation, and weighted control allocation. The computational flow is clear and suitable for periodic online solutions. In the current semi-physical simulation embodiment using the Python prototype, the average solution time for the complete link is approximately 0.781 ms, less than a 10 ms control cycle.

[0057] 4. This invention reduces the heavy reliance on high-precision wind tunnel aerodynamic data by observing and compensating for lumped nonlinear disturbances in real time through ESO, and provides a highly robust and universal solution for the control of eVTOL transition sections with different aerodynamic configurations. Attached Figure Description

[0058] Figure 1 A schematic diagram of the overall process of the aircraft transition control method;

[0059] Figure 2 The diagram shows the internal structure of the Dynamic Control Allocator (DCA) and the dual-path solution process of this invention. Detailed Implementation

[0060] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0061] Currently, among the methods for transition control of tiltrotor aircraft, techniques based on linear parameter variation (LPV) theory combined with gain scheduling are dominant. However, such methods heavily rely on accurate wind tunnel aerodynamic parameter matrices, and their ability to characterize and suppress unmodeled dynamics such as sudden gusts and rotor downwash interference is limited. Especially in complex urban airflow environments at low altitudes, it is difficult to balance attitude stability and trajectory tracking accuracy. Therefore, it is particularly important to explore a robust control method that can adaptively estimate unmodeled dynamics and achieve smooth allocation of control authority. This invention focuses on the transition control challenges of aircraft from vertical takeoff and landing to forward flight, and proposes a flight control system based on an extended state observer and adaptive sliding mode control. Considering the characteristics of aerodynamic parameter drastic changes and multi-control surface redundancy in the transition phase, traditional control architectures are prone to actuator saturation or attitude oscillations in complex environments, while existing pure nonlinear optimization algorithms suffer from excessively long solution times. This invention analyzes the force coupling characteristics of the transition section and discovers that internal and external disturbances can be aggregated into an expanded state for independent observation and feedforward compensation. Furthermore, by introducing a smooth attenuation function based on the tilt angle, a seamless transition between rotor thrust and wing lift can be achieved. Based on this, this invention designs a dedicated transition section control architecture, aiming to improve the anti-disturbance capability and tracking accuracy under strong disturbance conditions, while maintaining lightweight algorithm computation to meet the real-time requirements of airborne flight control systems.

[0062] This invention addresses the problem of unstable attitude and altitude control for tiltrotor aircraft (eVTOL) during the transition between vertical takeoff and landing and forward flight modes (transition phase) due to drastic changes in aerodynamic characteristics and external wind disturbances. It proposes an aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation.

[0063] An aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation includes the following steps: acquiring the real-time state of the aircraft through a flight state sensing unit and estimating the total composite disturbance of the system in real time using an extended state observer; generating a dynamic reference flight trajectory and attitude reference commands for the transition phase using a trajectory planner based on the rotor tilt angle and airspeed; employing an adaptive sliding mode controller, combining real-time state feedback and the estimated total composite disturbance, calculating virtual control commands for attitude and altitude correction, and adaptively updating the control gain; and using a dynamic control allocator to map the virtual control commands in real time to the target rotational speed of each rotor and the target deflection angle of each aerodynamic control surface, thereby achieving aircraft transition phase control.

[0064] The method utilizes an extended state observer to estimate the total composite disturbance of the system (an aircraft flight control system consisting of a flight state sensing unit, an onboard control computing unit, and actuators) in real time. The core idea is to treat aerodynamic parameter perturbations, external gust disturbances, and inter-channel coupling as new extended state variables. The estimated error is output to drive asymptotic convergence, achieving real-time tracking and feedforward compensation of the total composite disturbance. Detailed implementation methods are provided for the specific observer structure and parameter design.

[0065] The system employs an adaptive sliding mode controller to calculate virtual control commands for attitude and altitude correction, utilizing an integral sliding surface and an adaptive reaching law. The integral sliding surface consists of the proportional, integral, and derivative terms of the tracking error, used to eliminate steady-state error. The control law is based on equivalent control, introducing the disturbance estimate of the ESO output as a feedforward compensation term, and replacing the traditional sign function with a saturation function to reduce chattering. The switching gain is adjusted online via an adaptive law, without requiring prior knowledge of the upper limit of the disturbance. Detailed control law expressions, sliding surface definitions, and parameter descriptions are provided in the embodiments.

[0066] The adaptive update control gain dynamically adjusts its switching gain based on the degree of deviation from the sliding surface: when the system moves far from the sliding surface, the gain is automatically increased to accelerate the approach; when it approaches the sliding surface, the gain is automatically attenuated through a forgetting mechanism to avoid over-control; simultaneously, a lower bound is set to ensure that the control effect does not disappear. For detailed expressions and parameter descriptions of the adaptive law, please refer to the embodiments.

[0067] The method utilizes a dynamic control distributor to construct a global control efficiency matrix that integrates rotor thrust and aerodynamic control surface forces, and introduces a control based on rotor tilt angle. Smooth transition function and flight airspeed The relevant dynamic pressure efficiency factor enables seamless handover of control authority between the rotor and aerodynamic control surfaces; a weighted pseudo-inverse allocation is used to provide a fast analytical solution for unconstrained operating conditions, and a quadratic programming correction is introduced when the actuator is saturated to ensure the physical realizability of the solution. For details on the efficiency matrix fusion formula and optimization problem modeling, please refer to the implementation example.

[0068] An aircraft transition phase control system based on adaptive disturbance rejection and dynamic allocation includes: a flight state sensing unit, an airborne control computing unit, and an actuation mechanism;

[0069] The flight status sensing unit is a hardware measurement component used to acquire the real-time status of the aircraft and rotor tilt angle. and flight airspeed It also outputs sensor fusion measurement information to the airborne control computing unit;

[0070] The airborne control computing unit is the control algorithm execution part, including a trajectory planner, an extended state observer, an adaptive sliding mode controller, and a dynamic control distributor;

[0071] The trajectory planner is used to determine the rotor tilt angle. and flight airspeed Generate reference controlled state vector And when the negative height deviation exceeds a preset threshold, a height subsidence compensation vector is generated. ;

[0072] The extended state observer is used to output a controlled state estimate based on the real-time state of the aircraft obtained by the flight state sensing unit. Estimated rate of change of controlled state and the estimated total combined disturbance ;

[0073] The adaptive sliding mode controller is used to... , , and Calculate the virtual control command vector ;

[0074] The dynamic control allocator is used to... Rotor tilt angle Flight airspeed The actual control quantity of the previous control cycle In addition to the physical constraints of the actuator, the optimal actual control quantity is solved. ;

[0075] The actuation mechanism is a hardware execution part, including a rotor motor, a tilting mechanism, and a pneumatic control surface actuator, used to receive... The corresponding rotor channel executes commands and the aerodynamic control surface channel executes commands, driving the aircraft to complete the transition control.

[0076] Example:

[0077] like Figure 1 As shown, in its specific implementation, this invention first acquires one-dimensional and three-dimensional state data through multi-source sensor fusion to solve the key problem of high-frequency vibration and noise interference. A nonlinear extended state observer (ESO) is used to reconstruct internal and external disturbances of the system. Its core is to improve the observation gain with small errors through a non-smooth feedback function (fal function), achieving high-precision observation of lumped disturbances including wind shear and aerodynamic perturbations.

[0078]

[0079] fal function definition:

[0080]

[0081] in:

[0082]

[0083]

[0084] By adjusting parameters to balance the observation response speed and noise resistance, high-precision interference feedforward compensation was effectively obtained. The sensor fusion measurement vector is denoted as... The controlled state vector actually used by the controller is denoted as... ESO output As the controlled state feedback of ASMC, the output Used to construct the error derivative, output As a feedforward compensation for the composite total disturbance, this feedforward observation and planning process not only eliminates steady-state error but also lays the foundation for attitude correction of the adaptive sliding mode controller by compensating for high-frequency disturbances.

[0085] The trajectory planner is based on the current flight airspeed With rotor tilt angle (0° = hovering, 90° = level flight), a smooth transition reference curve from hovering to level flight is generated using a cosine smoothing dynamic interpolation algorithm. Rotor tilt angle. The information comes from the tilt mechanism encoder and is not related to the aircraft's three-axis attitude angles; airspeed. From the pitot tube measurement channel.

[0086] Specifically, we define the normalized transition variable and the interpolation factor:

[0087]

[0088] The reference controlled state vector is defined as:

[0089]

[0090] in:

[0091]

[0092] It is a compensation item for high-level subsidence and is linked to the high-priority protection mechanism.

[0093] The control system is divided into two stages: virtual command computation and control allocation. A lightweight computation architecture is constructed based on a hierarchical structure. The virtual command computation stage gradually converges the tracking error by introducing an adaptive law and a sliding mode surface.

[0094] 1. Attitude and altitude error calculation: The reference controlled state vector output from stage two is used to calculate the attitude and altitude error. The controlled state estimate output by the ESO in stage one Subtraction yields the tracking error with consistent dimensions. The error derivative is constructed from the estimated rate of change of the ESO output and the reference rate of change. This avoids directly differentiating the noisy state numerically and simultaneously calculates the error integral term.

[0095] 2. Interference feedforward compensation: This involves adjusting the estimated total composite interference value observed in real-time by the ESO. By directly introducing the control equations, the feedforward channel is used to offset most of the nonlinearity of the system and the effects of external gusts.

[0096] 3. Adaptive Sliding Mode Control (ASMC): Construct an integral sliding surface and design an adaptive reaching law to calculate the three-axis virtual torque and total thrust command. The sliding mode control law is as follows:

[0097]

[0098] The integral sliding surface is:

[0099]

[0100] Switching gain Updated online according to the adaptive law:

[0101]

[0102] in, , This is a virtual command for total thrust. , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively. , which is a known nominal nonlinear dynamic function obtained based on the standard aerodynamic equations and rigid body motion equations, used to represent the mapping relationship between the controlled state, the rate of change of the controlled state and the nominal acceleration term when the total combined disturbance is not considered; This is the estimated total composite disturbance. To control the input gain coefficient; , These are the parameters of the sliding surface; For adaptively updated switching gain; function; This represents the boundary layer thickness.

[0103] The gain is automatically reduced when the system approaches the sliding surface, which fundamentally weakens the high-frequency chattering phenomenon inherent in traditional sliding mode control.

[0104] Control allocation phase (e.g.) Figure 2 (As shown) This module aims to solve the command mapping problem for redundant control surfaces (such as the four tilt rotors, ailerons, elevator, and rudder). The Dynamic Control Allocation (DCA) module is designed for transition characteristics, integrating commands based on rotor tilt angle. The efficiency reconstruction and weighted pseudo-inverse solution matrix resolve the permission conflicts caused by traditional direct allocation.

[0105] Determining the rotor control effectiveness matrix based on the rotor thrust model The rotor control performance matrix Used to represent the mapping relationship between the actual control quantities of the rotor channel and the total thrust and three-axis torque generated by the rotor;

[0106] Determining the control effectiveness matrix of aerodynamic control surfaces based on aerodynamic surface model The aerodynamic control surface efficiency matrix It is used to represent the mapping relationship between the actual control quantity of the aerodynamic control surface channel, the airspeed, and the aerodynamic force and torque generated by the aerodynamic control surface.

[0107] Control effectiveness matrix It consists of two dynamically weighted fusion parts:

[0108]

[0109] Cosine transition weight function:

[0110]

[0111] Dynamic pressure efficiency factor:

[0112]

[0113] in:

[0114]

[0115] Weighting function Compared with flight airspeed dynamic pressure factor Together, they achieve a seamless handover of control authority between the rotor and aerodynamic control surfaces.

[0116] Receive the virtual control command vector output by the adaptive sliding mode controller:

[0117] Where T is the virtual command for total thrust. , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively.

[0118] The vector of the actual control quantity to be determined in the dynamic control distributor is defined as:

[0119]

[0120] in, For the first The actual control quantity of each rotor channel For the first The actual control quantity of each aerodynamic control surface channel. The number of rotor channels to be allocated. The number of aerodynamic control surface channels to be allocated;

[0121] First according to and Calculate the initial value of the bias-weighted pseudo-inverse assignment. ,

[0122]

[0123] in, express The bias-weighted pseudo-inverse matrix is ​​obtained. The initial value of the bias-weighted pseudo-inverse is then determined. Whether the actuator amplitude and rate constraints are triggered, i.e., the constraint expression in the quadratic optimization problem. ;

[0124] when The actuator amplitude and rate constraints were not triggered, i.e. If the amplitude is within the allowable range of the actuator and the change in adjacent control cycles does not exceed the actuator rate limit, then it indicates that neither the actuator amplitude constraint nor the rate constraint has been activated. And output; If any rotor channel or aerodynamic control surface channel exceeds or reaches a preset amplitude constraint boundary, or if its change in actual control quantity relative to the previous control cycle exceeds or reaches a preset rate constraint boundary, it indicates that when any rotor channel or aerodynamic control surface channel touches the actuator amplitude constraint or rate constraint, it will trigger an automatic control. As initial iteration values ​​or reference initial values ​​for solving the quadratic programming problem, this is achieved by solving a quadratic optimization problem with physical magnitude and rate saturation constraints:

[0125]

[0126] in:

[0127]

[0128] The solution obtained This is the optimal solution for the combination of rotor speed and aerodynamic control surface deflection angle, and its components include the optimal execution commands for each rotor channel. Optimal execution commands for each aerodynamic control surface channel The output is then sent from the dynamic control distributor to the rotor motor and aerodynamic control surface actuator.

[0129] In the aforementioned control allocation mechanism, the weighted pseudo-inverse and quadratic programming constitute two progressive levels within the same allocation framework:

[0130] When constraints are not activated: the optimization problem degenerates into a weighted pseudo-inverse analytical solution;

[0131] When the constraint is activated: the optimal feasible solution that satisfies the saturation limit is obtained by solving the quadratic programming numerical solution.

[0132] The two work together to ensure both rapid calculation under normal operating conditions and optimal allocation within the safety boundary under saturated operating conditions.

[0133] To address the issue of aircraft easily losing altitude due to wind shear, this invention employs altitude drop compensation logic at the trajectory planning layer: when... When the highly negative deviation of the representation exceeds a preset threshold, the system... Altitude compensation is added and the weight of the total thrust control channel in the weight matrix W is dynamically increased, forcing the DCA module to prioritize thrust allocation in the altitude dimension, thereby improving flight safety in complex and harsh environments.

[0134] A high-priority protection mechanism is set up between the trajectory planner and the dynamic control allocator, specifically including:

[0135] Based on the height estimate output by the extended state observer Reference height output by the trajectory planner Calculate the height channel error: ;

[0136] when When the aircraft is found to have exceeded a preset threshold, it is determined that the aircraft has exceeded the threshold. High negative deviation; trajectory planner generates and references the controlled state vector Same-dimensional height subsidence compensation vector: ,in, It is the height compensation amount, and is determined based on the portion of the negative height deviation that exceeds the threshold.

[0137] The trajectory planner superimposes the height sink compensation vector onto the reference controlled state vector to obtain the corrected reference controlled state vector:

[0138]

[0139] Simultaneously, the dynamic control allocator increases the weights corresponding to the total thrust channels in the channel priority weight matrix W, so that the weight matrix satisfies:

[0140]

[0141] in, For the total thrust channel weight, , , The weights for roll, pitch, and yaw moment channels are respectively used; the dynamic control allocator uses the updated weights in the quadratic optimization problem. This prioritizes reducing the distribution error of the total thrust channels, thereby ensuring that the thrust required for altitude maintenance is met within the actuator constraints. The actuators include the actual execution channels such as the rotor motor and aerodynamic control surface actuators.

[0142] To verify the control effect of the control link under transition disturbance conditions, this embodiment uses a semi-physical simulation experimental platform built with Python for closed-loop verification. The verification experiment uses altitude and pitch angle as the main controlled objects, with a control cycle set to 10ms. The rotor tilt angle smoothly transitions from 0° to 60°, and 3m / s horizontal disturbance and 1m / s vertical disturbance windows are set to examine altitude maintenance, attitude tracking, actuator allocation, and calculation time under transition disturbance conditions. The verification process includes:

[0143] Environment initialization: Configure the wind field model (including turbulence and discrete gusts), set the eVTOL physical model and initial hovering state.

[0144] Command generation and closed-loop: Configure a 50kg-class eVTOL dynamic model, rotor tilting process, horizontal and vertical disturbance windows, and set the initial hovering state.

[0145] Command generation and closed-loop control: The trajectory planner generates altitude and pitch reference commands, the ESO estimates the controlled state and complex disturbances, and the ASMC calculates virtual control commands for total thrust and pitch moment every 10ms.

[0146] Distribution and actuator response: DCA distributes virtual control commands to the two rotor thrusters and elevator; control groups are configured with ESO off, DCA off, or PID baseline control.

[0147] The experiment included four control groups: complete control link, ESO disabled, DCA disabled, and PID baseline, to evaluate the contributions of disturbance observation compensation, dynamic control assignment, and complete control link to the transient control performance.

[0148]

[0149] The above results show that in the semi-physical simulation experiment, the maximum height drop of the complete control link is about 0.25m, which is about 97.7% lower than the baseline of the conventional proportional-integral-derivative control, and significantly less than about 4.29m when the extended state observer is turned off; the actuator saturation ratio of the complete control link is 0%, which is significantly lower than about 33.70% when the dynamic control distributor is turned off; the average solution time of the complete control link is about 0.781ms, which is less than the 10ms control cycle and meets the online solution requirements within the control cycle set in this experiment.

[0150] Experimental results demonstrate that this invention improves the performance of transition height maintenance, control allocation, and online solution by compensating for composite disturbances with an extended state observer, enhancing tracking robustness through adaptive sliding mode, and mitigating actuator saturation risk through dynamic control allocation.

Claims

1. A flight transition control method based on adaptive disturbance rejection and dynamic allocation, characterized in that, Includes the following steps: The flight state sensing unit acquires the real-time state of the aircraft, and the extended state observer is used to estimate the total composite disturbance of the aircraft's flight control system in the transition phase in real time, thus obtaining the controlled state estimate. Estimated rate of change of controlled state and the estimated total combined disturbance ; Based on the rotor tilt angle obtained through the tilt mechanism encoder Compared with the flight airspeed obtained through the airspeed measurement channel The trajectory planner generates a dynamic reference flight trajectory and attitude control state commands for the transition segment. ; An adaptive sliding mode controller is used, combined with attitude control state commands. Controlled state estimate Estimated rate of change of controlled state Combined total disturbance estimate The virtual control command vector is obtained. And adaptively update the control gain; Using a dynamic control assigner, virtual control command vectors are... Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surfaces to achieve transition control of the aircraft.

2. The aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation according to claim 1, characterized in that, The method of using an extended state observer to estimate the total composite disturbance of the aircraft flight control system in the transition phase in real time includes the following steps: The flight status perception module acquires the aircraft's three-axis attitude angles, angular rates, position, and body velocity components, forming a sensor fusion measurement vector. And extract the controlled state vector actually used by the controller from it. ; These represent altitude, roll angle, pitch angle, and yaw angle, respectively. These are the aircraft's roll, pitch, and yaw angular velocities, respectively. For position coordinates, For the velocity components of the machine system; Extended state observer will measure the controlled state value Compared with the controlled state estimate The difference between them is defined as the output estimation error. Aerodynamic parameter perturbations, external gust disturbances, and inter-channel coupling are unified and equivalently represented as a total composite disturbance. Its estimated value is And set the corresponding expansion state, i.e. ; The extended state observer estimates the error based on the output. Correct the observed state and output the estimated value of the controlled state. Estimated rate of change of controlled state and the estimated total combined disturbance and will , and Output to the adaptive sliding mode controller.

3. The aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation according to claim 1, characterized in that, The system employs an adaptive sliding mode controller, combined with attitude control state commands. Combined total disturbance estimate This yields virtual control command vectors for attitude and altitude correction. And adaptively update the control gain, including the following steps: Based on the controlled state estimate output by the extended state observer The reference controlled state vector output by the trajectory planner Calculate the tracking error: ; And based on the estimated rate of change of the controlled state output by the extended state observer Rate of change of the reference controlled state vector Calculate the error derivative: ; Constructing the integral sliding surface: ; The sliding mode control law is expressed as: ； in, , For total thrust virtual command, , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively. The nominal nonlinear dynamic function is known. This is the estimated total composite disturbance. To control the input gain coefficient; , These are the parameters of the sliding surface; For adaptively updated switching gain; function; This represents the boundary layer thickness.

4. The aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation according to claim 3, characterized in that, The adaptive update control gain, its switching gain The adaptive regulation law is: ； Additional constraints: When season , These are positive constants to ensure that the switching gain is always positive; among them, The positive learning rate constant, This is a forgetting factor used to prevent excessive gain inflation. The absolute value of the sliding surface.

5. The aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation according to claim 1, characterized in that, The method utilizes a dynamic control allocator to distribute virtual control command vectors. Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surface actuators to achieve transition control of the aircraft, specifically including: Determining the rotor control effectiveness matrix based on the rotor thrust model The aerodynamic control surface performance matrix is ​​determined based on the aerodynamic surface aerodynamic model. ; According to the rotor tilt angle Calculate the transition weight function And according to the airspeed Calculate the dynamic pressure efficiency factor ,Will and The fusion yields the global control performance matrix: ； Receive the virtual control command vector output by the adaptive sliding mode controller: Where T is the virtual command for total thrust. , , These are virtual torque commands for the roll, pitch, and yaw channels, respectively. The vector of the actual control quantity to be determined in the dynamic control distributor is defined as: ； in, For the first The actual control quantity of each rotor channel For the first The actual control quantity of each aerodynamic control surface channel. The number of rotor channels to be allocated. The number of aerodynamic control surface channels to be allocated; First according to and Calculate the initial value of the bias-weighted pseudo-inverse assignment. : ,in, express The bias-weighted pseudo-inverse matrix is ​​then determined, and the initial value of the bias-weighted pseudo-inverse is assigned. Whether to trigger actuator amplitude and rate constraints; If not triggered, And output; if triggered, with As the initial iteration value or reference initial value for solving the quadratic programming problem, the following constrained quadratic optimization problem is solved: ； in, This is the channel priority weight matrix. and Limiting the physical amplitude of the actuator. Limiting the maximum speed of the actuator. This represents the actual control quantity from the previous control cycle. To control the cycle; The solution obtained This is the optimal solution for the combination of rotor speed and aerodynamic control surface deflection angle, and its components include the optimal execution commands for each rotor channel. Optimal execution commands for each aerodynamic control surface channel The output is then sent by the dynamic control distributor to the rotor motor and aerodynamic control surface actuator.

6. The aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation according to claim 1, characterized in that, A high-priority protection mechanism is set up between the trajectory planner and the dynamic control allocator, specifically including: Based on the height estimate output by the extended state observer Reference height output by the trajectory planner Calculate the height channel error: ; when When the aircraft is found to have exceeded a preset threshold, it is determined that the aircraft has exceeded the threshold. High negative deviation; trajectory planner generates and references the controlled state vector Same-dimensional height subsidence compensation vector: ;in, It is the height compensation amount, and is determined based on the portion of the negative height deviation that exceeds the threshold. The trajectory planner superimposes the height sink compensation vector onto the reference controlled state vector to obtain the corrected reference controlled state vector: ； in, For smooth transition interpolation factor, This represents the baseline controlled state vector for the hovering mode. Represents the baseline controlled state vector for forward flight mode; Simultaneously, the dynamic control allocator increases the weights corresponding to the total thrust channels in the channel priority weight matrix W, so that the weight matrix satisfies: ； in, For the total thrust channel weight, , , These are the weights for the roll, pitch, and yaw moment channels, respectively; the dynamic control distributor solves for the optimal actual control quantity. The updated version is used in the aforementioned quadratic optimization problem. .

7. A transition phase control system for an aircraft based on adaptive disturbance rejection and dynamic allocation, characterized in that, include: An extended state observer is used to estimate the total composite disturbance of the aircraft's flight control system during the transition phase based on the real-time aircraft state acquired through the flight state sensing unit, thereby obtaining the controlled state estimate. Estimated rate of change of controlled state and the estimated total combined disturbance ; A trajectory planner is used to calculate the rotor tilt angle based on the rotor tilt angle obtained through the tilt mechanism encoder. Compared with the flight airspeed obtained through the airspeed measurement channel The trajectory planner generates a dynamic reference flight trajectory and attitude control state commands for the transition segment. ; Adaptive sliding mode controller, used to combine attitude control state commands Controlled state estimate Estimated rate of change of controlled state Combined total disturbance estimate The virtual control command vector is obtained. And adaptively update the control gain; Dynamic control assigner, used to assign virtual control command vectors Rotor tilt angle Flight airspeed Given the physical constraints of the actuator, solve for the optimal actual control quantity. and will The output is sent to the rotor motor and aerodynamic control surfaces to achieve transition control of the aircraft.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements an aircraft transition phase control method based on adaptive disturbance rejection and dynamic allocation as described in any one of claims 1-6.

Images