A method, system, electronic equipment, and storage medium for autonomous landing output feedback control of a quadcopter unmanned aerial vehicle (UAV).

Abstract

This invention discloses a method, system, electronic equipment, and storage medium for autonomous landing output feedback control of a quadrotor unmanned surface vessel (USV), relating to the field of intelligent control technology for integrated air-sea operations. It includes: constructing dynamic and kinematic models of the USV and USV; designing a trajectory phase switching mechanism for the USV to generate a reference landing trajectory; constructing a neural network extended state observer to reconstruct the USV's unmeasurable velocity state, internal unknown nonlinearities, and external disturbances; designing an outer-loop position control law based on the reference landing trajectory and reconstructed information to obtain the USV's desired attitude angle and thrust command; constructing an inner-loop attitude control law, and achieving high-precision trajectory tracking and safe, stable landing of the quadrotor USV on the USV platform based on the thrust command output by the outer-loop position control law and the inner-loop attitude control law, thereby improving the control accuracy, safety, input smoothness, and system robustness of the autonomous landing process in complex marine environments.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for integrated air and sea operations, and in particular to a method, system, electronic device and storage medium for autonomous landing output feedback control of a quadcopter UAV. Background Technology

[0002] Quadrotor drones, due to their simple structure, maneuverability, excellent hovering performance, and strong vertical takeoff and landing capabilities, have been widely used in scenarios such as marine inspection, target search, emergency delivery, and unmanned surface vessel (USV) collaborative operations. In air-sea collaborative missions, after completing their sensing or delivery tasks, drones typically need to return and land on a moving USV platform. Therefore, autonomous landing has become a key technical issue in air-sea collaborative systems.

[0003] Compared to landing on a static platform, autonomous landing of UAVs on unmanned surface vessels (USVs) faces greater time-varying characteristics and complexity. USVs experience continuously changing platform motion due to wind, waves, currents, and their own maneuvering, requiring UAVs to not only accurately track their relative position during landing but also maintain attitude stability and smooth control inputs. In this process, difficulty in acquiring velocity status and insufficient handling of complex disturbances are two prominent issues affecting autonomous landing performance. On the one hand, in real marine environments, the velocity state of UAVs is often difficult to measure directly, continuously, and with high precision due to factors such as sensor noise, platform vibration, drift of low-cost inertial devices, visual obstruction, and sea surface reflection. Especially in the final stage near the moving deck, velocity estimation errors are easily amplified by the control loop, thus affecting position tracking accuracy and landing stability. Therefore, how to achieve state reconstruction under conditions where velocity cannot be directly and reliably measured is a key issue in output feedback control design. On the other hand, the autonomous landing process of unmanned aerial vehicles (UAVs) is simultaneously affected by multiple factors, including internal unknown nonlinearities, model parameter perturbations, actuator efficiency variations, and external wind and wave disturbances. Existing disturbance mitigation methods often employ direct compensation after unified estimation. While this can suppress the impact of disturbances to some extent, it may also cancel out disturbance components that contribute to error convergence, leading to increased control energy consumption, abrupt changes in control input, or decreased terminal landing stability. Therefore, effectively identifying and appropriately handling the effects of disturbances remains a crucial problem to be solved in current autonomous landing control. Summary of the Invention

[0004] This invention provides an autonomous landing output feedback control method for quadcopter unmanned aerial vehicles (UAVs) to overcome the above-mentioned technical problems.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows: A method for autonomous landing output feedback control of a quadcopter UAV, comprising the following steps: S1. Construct dynamic and kinematic models for unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs); S2. Based on the kinematic and dynamic models of the unmanned surface vessel (USV), a trajectory phase switching mechanism is designed to provide a safe window for determining the trajectory phase switching of the USV, thereby generating a reference landing trajectory for the USV. S3. Based on the dynamic and kinematic models of the UAV, a neural network extended state observer is constructed to reconstruct the unmeasurable velocity state, internal unknown nonlinearity, and external disturbances of the UAV online. S4. Design the position outer loop control law based on the reference landing trajectory and the reconstructed information obtained based on the neural network extended state observer; S5. Combining the aforementioned position outer loop control law, the desired attitude angle and thrust command of the UAV are obtained; S6. Construct an inner-loop attitude control law based on the desired attitude angle of the UAV, and realize the autonomous landing control of the quadcopter UAV based on the thrust command output by the outer-loop position control law and the inner-loop attitude control law.

[0006] Beneficial effects: This invention enables high-precision trajectory tracking and safe and stable landing of a quadcopter UAV onto a mobile unmanned surface vessel (USV) platform under conditions where the speed state of the UAV is difficult to measure directly and reliably, internal unknown nonlinearity and external ocean disturbances work together, and the continuity requirements of multi-stage landing switching are high. It improves the control accuracy, safety, input smoothness and system robustness of the autonomous landing process in complex marine environments. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a first flowchart of an autonomous landing output feedback control method for a quadcopter unmanned aerial vehicle (UAV) according to the present invention. Figure 2 This is a second flowchart of an autonomous landing output feedback control method for a quadcopter unmanned aerial vehicle (UAV) according to the present invention. Figure 3 This is a schematic diagram of the structure of an autonomous landing output feedback control system for a quadcopter unmanned aerial vehicle (UAV) according to an embodiment of the present invention; Figure 4 This is a three-dimensional spatial trajectory diagram of the UAV-USV in an embodiment of the present invention; Figure 5 This is a position error curve diagram in an embodiment of the present invention; Figure 6 This is a diagram of the attitude error curve in an embodiment of the present invention; Figure 7 This is a comparison diagram of the position, velocity state, and estimated state of the quadcopter UAV in an embodiment of the present invention; Figure 8 This is a comparison diagram of the attitude, angular velocity state, and estimated state of the quadcopter UAV in an embodiment of the present invention; Figure 9 This is a comparison diagram of the disturbance and the estimated disturbance of the quadcopter UAV in an embodiment of the present invention; Figure 10 This is a thrust curve diagram of a quadcopter UAV in an embodiment of the present invention; Figure 11 This is a control torque curve diagram of a quadcopter UAV in an embodiment of the present invention; Figure 12 This is a schematic diagram of the position outer ring smoothing judgment function in an embodiment of the present invention; Figure 13 This is a schematic diagram of the attitude inner loop smoothing judgment function in an embodiment of the present invention; Figure 14 This is a diagram showing the L2 norm curve of the position outer ring weight estimate in an embodiment of the present invention; Figure 15 This is a diagram of the L2 norm curve of the attitude inner loop weight estimate in an embodiment of the present invention. Figure 16 This is a schematic diagram of the overall Simulink simulation interface in an embodiment of the present invention; Figure 17 This is a schematic diagram of the construction module of the UAV model in the Simulink simulation interface in an embodiment of the present invention; Figure 18 This is a schematic diagram of the construction module of the unmanned surface vessel model in the Simulink simulation interface in an embodiment of the present invention; Figure 19 This is a schematic diagram of the UAV position loop module in the Simulink simulation interface of this invention. Figure 20 This is a schematic diagram of the Simulink simulation interface UAV attitude control module in an embodiment of the present invention; Figure 21 This is a schematic diagram of the neural network extended state observer module for the UAV position loop in the Simulink simulation interface in this embodiment of the invention; Figure 22 This is a schematic diagram of the neural network extended state observer module of the UAV attitude loop in the Simulink simulation interface in this embodiment of the invention; Figure 23 This is a schematic diagram of the drone attitude calculation module in the Simulink simulation interface of this invention. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] This embodiment provides an output feedback control method for autonomous landing of quadrotor UAVs on boats, targeting autonomous landing scenarios of quadrotor UAVs in complex marine environments and on moving decks, such as... Figure 1 and Figure 2 As shown, the specific steps include: S1. Construct dynamic and kinematic models for unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs); In a specific embodiment, the specific steps in S1 for constructing the dynamic and kinematic models of the UAV and unmanned surface vessel include: In a specific embodiment, to accurately describe the relative motion relationship between the quadcopter UAV and the unmanned surface vessel during the coordinated landing process at sea, a geodetic coordinate system is established. UAV attached coordinate system and the coordinate system of the unmanned surface vessel ; Among them, the geodetic coordinate system is fixed at the Earth reference point and is used to describe the absolute position of the UAV and the unmanned surface vessel; the UAV appendage coordinate system is fixed at the geometric center of the quadcopter and is used to describe the linear velocity, angular velocity, thrust and control torque; the unmanned surface vessel appendage coordinate system is fixed at the deck reference position or the geometric center of the unmanned surface vessel and is used to describe the attitude of the unmanned surface vessel platform and its time-varying motion.

[0011] In a specific embodiment, a six-degree-of-freedom expression is used to construct the dynamics and kinematics model of the quadcopter UAV. The expressions for the constructed dynamics and kinematics model of the UAV are as follows: (1) Among them, the position vector of the UAV in the geodetic coordinate system , These represent the UAV in the geodetic coordinate system. Position coordinates along the axis, linear velocity vector of the UAV in the attached coordinate system , These represent the components of the UAV's velocity along the three coordinate axes of the corresponding coordinate system, and the attitude vector. , These represent the UAV's roll angle, pitch angle, and heading angle, respectively, and the angular velocity vector in the UAV's attached coordinate system. , These represent the coordinates of the UAV around the attached coordinate system. The roll rate, pitch rate, and yaw rate of the shaft. For the total mass of the drone, Here is the rotational inertia matrix. For the thrust and control torque of the quadcopter drone , These represent the roll control torque, pitch control torque, and yaw control torque of the UAV, respectively, with the gravity term as the other term. , , Represents gravitational acceleration. and These represent the internal unknown nonlinear terms in the linear velocity channel and the angular velocity channel, respectively. and These represent the external disturbance terms formed by wind field, sea surface airflow disturbance, and platform relative motion induced effect, respectively. Represents the attitude rotation matrix. The angular velocity transformation matrix can be represented using Euler angles, a common expression in this field. This applies when the UAV's roll and pitch angles are within their normal operating range. As a bounded matrix, this condition usually holds within the normal control range of unmanned aerial vehicle (UAV) autonomous landing missions.

[0012] Specifically, the UAV dynamics and kinematics model constructed in this embodiment includes: UAV position sub-model. The system comprises a velocity subsystem, an attitude subsystem, and an angular velocity subsystem. The system's linear velocity is mapped to the geodetic coordinate system using an attitude rotation matrix. The system's angular velocity is mapped to the attitude angular rate of change using a transformation matrix. Internal unknown nonlinear terms and external disturbance terms are introduced into the velocity and angular velocity channels, respectively, to form a system model suitable for output feedback control design. The UAV's position rate of change is determined by the linear velocity... The rate of change of linear velocity is determined by thrust, gravity, and internal unknown nonlinearities. External disturbances The attitude change rate is jointly determined by the transformation matrix. With angular velocity The rate of change of angular velocity is determined by the control torque and internal unknown nonlinearity. and external disturbances A joint decision. In a specific embodiment, the kinematic and dynamic models of the unmanned surface vessel (USV) constitute the USV platform motion model, which describes the USV's position, attitude, and time-varying deck motion state under the combined effects of wave excitation, manipulation inputs, and external disturbances. The expressions for the USV's kinematic and dynamic models are as follows: (2) Among them, the position of the unmanned surface vessel in the geodetic coordinate system , These represent the unmanned surface vessel in the geodetic coordinate system. Position coordinates along the axis; velocity of the unmanned surface vessel in the geodetic coordinate system , These represent the unmanned surface vessel in the geodetic coordinate system. Velocity components along the axial direction; attitude of the unmanned surface vessel in the hull appendage coordinate system. , These represent the roll, pitch, and bow angles of the unmanned surface vessel (USV), respectively, and the angular velocity of the USV. , These represent the angular velocity components of the unmanned surface vessel (USV) around the three coordinate axes of the hull's appendage coordinate system. Due to the inertia and added mass effect of the unmanned surface vessel, For hydrodynamic damping, ; , and These are the control force and torque generated by the propeller, respectively. and These represent the amplitude and frequency of the ocean waves, respectively. and External disturbance; This represents the linear velocity transformation matrix from the ship's attached coordinate system to the geodetic coordinate system; This represents the angular velocity transformation matrix from the ship's attached coordinate system to the geodetic coordinate system.

[0013] Specifically, by using formulas (1) and (2), the relative position, relative attitude, and relative motion state between the UAV and the unmanned surface vessel platform during the autonomous landing process can be obtained.

[0014] S2. Based on the kinematic and dynamic models of the unmanned surface vessel (USV), a trajectory phase switching mechanism is designed to provide a safe window for determining the trajectory phase switching of the USV, thereby generating a reference landing trajectory for the USV. In a specific embodiment, in S2, during the autonomous landing mission, the continuity of the reference trajectory directly affects the smoothness of the control input and the safety of the terminal landing. Therefore, this embodiment divides the autonomous landing mission into an approach phase and a landing phase. The approach phase primarily achieves rapid approach of the UAV to the UAV platform; the landing phase primarily achieves high-precision stable docking and safe landing of the UAV. To ensure landing reliability, this embodiment designs a UAV trajectory phase switching mechanism based on the UAV's kinematic and dynamic models to provide a safety window for trajectory phase switching, which is expressed as follows: (3) in, The threshold for unmanned surface vessel attitude determination is denoted as , and ; Specifically, the unmanned surface vessel (USV) deck is considered to have entered a safe landing window only when the heave, roll, and pitch motions of the USV deck simultaneously satisfy the criteria in equation (3). This safe landing moment is recorded as the safe landing switching moment for the USV. ,when At that time, the UAV's expected trajectory changes from the approach phase trajectory to the landing phase trajectory in order to achieve a rapid and stable approach and landing on the UAV deck.

[0015] Specifically, the method proposed in this embodiment is applicable to autonomous approach, relative alignment, terminal glide and contact landing missions of UAVs on unmanned surface vessels, manned ships or other time-varying motion platforms. It is also applicable to various air-sea collaborative operation scenarios where the speed state is not easy to measure directly, the model uncertainty is significant, and there are continuous complex disturbances such as wind, waves and currents.

[0016] In the design of UAV landing reference trajectories, a smooth reference trajectory is not only a prerequisite for ensuring landing safety, but also directly affects the robustness and real-time performance of the controller. This embodiment aims to avoid abrupt changes in the UAV's reference landing trajectory during phase transitions, while also reducing the complexity of the trajectory generation process. It uses a trajectory phase transition mechanism to obtain the safe landing transition time. This leads to the design of the desired trajectory for the UAV to smoothly transition from a safe offset position to the deck landing point, i.e., the reference landing trajectory. , represented as: (4) in, , For safety redundancy height, For smoothing parameters; Represents the three-dimensional spatial trajectory of the unmanned surface vessel; For the current moment, It is the hyperbolic tangent function.

[0017] S3. Based on the dynamic and kinematic models of the UAV, a neural network extended state observer is constructed to reconstruct the unmeasurable velocity state, internal unknown nonlinearity, and external disturbances of the UAV online. Specifically, such as Figure 2As shown, the Extended State Observer (ESO) is an observation method that expands unknown dynamics and external disturbances into additional states and performs online estimation. This embodiment introduces a neural network approximator into the ESO structure, forming a Neural Network Extended State Observer (NN-ESO). In this specific embodiment, considering that the velocity state of a UAV in a marine environment is often difficult to obtain directly with high precision, and that the system contains unmodeled aerodynamic effects, parameter uncertainties, and actuator efficiency variations, this embodiment constructs a neural network extended state observer. Its core idea is: to use a radial basis function neural network to approximate the unknown nonlinearities within the UAV online, obtaining estimates of these internal nonlinearities; to expand external disturbances and the neural network approximation residuals into observation states; and to form observation error feedback terms based on position and attitude measurement information, obtaining estimates of the UAV's linear velocity, angular velocity, position channel composite disturbance, and attitude channel composite disturbance, thereby achieving synchronous reconstruction of states and uncertainties under a unified structure. In S3, the specific steps for constructing a neural network extended state observer based on the dynamics and kinematics models of the UAV include: S31. Construct neural network approximators for the linear velocity channel and the angular velocity channel respectively to approximate the internal unknown nonlinear terms in the linear velocity channel and angular velocity channel of the UAV-based dynamics and kinematics models. The expressions are as follows: (5) in, These are the ideal weight matrices for the linear velocity channel and the angular velocity channel neural network approximation, respectively. These are the basis function vectors for the linear velocity channel and the angular velocity channel neural network approximator, respectively. Let represent the estimation error vector, and ,in and , where are positive constants, representing the upper bounds of the two types of approximation errors respectively; S32. Based on the dynamic model, kinematic model and formula (5) of the UAV, the position state of the UAV is defined. Unmeasurable speed state of drones Position-loop related perturbation state UAV attitude status Unmeasurable angular velocity state Attitude loop related perturbation state , , ; The error between the actual position and the observed position of the UAV is defined as: The error between the actual attitude state and the observed attitude state is Define the position loop weight error Attitude loop weight error , This is the ideal weight estimation matrix; Based on the above definition, a neural network extended state observer is designed, with the following expression: (6) in, , , , , and All are observer gains; These correspond to the estimated values ​​of the UAV's position, linear velocity, and position loop perturbation, respectively. These correspond to the estimated values ​​of the UAV's attitude, angular velocity, and attitude loop perturbation, respectively. It is the outer loop control law for the position to be designed; The adaptive weight update law for the neural network extended state observer is designed as follows: (7) In a neural network extended state observer, some parameters satisfy the following inequality, where the observation error... and and neural network weight estimation error and It is bounded, that is: (8) in, and All are positive definite matrices. Let the upper bound of the norm of the rotation matrix be , and The forgetting factor is used to effectively suppress weight drift and improve the robustness of online learning.

[0018] Specifically, the weight adaptive update law corrects the weights online based on the observation error and the basis function vector. In this embodiment, a forgetting term is introduced into the update law to suppress weight drift and ensure that the observation error system is consistent and eventually bounded.

[0019] Specifically, the input of the neural network extended state observer designed in this embodiment consists of UAV position measurement values, attitude measurement values, and their filtering values. Compared with traditional observers that only estimate the system state, the neural network extended state observer can perform integrated online reconstruction of UAV linear velocity, angular velocity, internal unknown nonlinearity, and external complex disturbances, and can output UAV linear velocity estimates, angular velocity estimates, position channel complex disturbance estimates, and attitude channel complex disturbance estimates.

[0020] In this embodiment, the extended state observer of the neural network is analyzed by constructing a Lyapunov function, as shown in formula (9): (9) Combining formula (9), we get: (10) (11) in, and These are the bias terms of the neural network basis functions for the linear velocity channel and the angular velocity channel, respectively. , , and These are the upper bound constants for the bias terms of the neural network basis functions in the linear velocity and angular velocity channels, respectively. , , and These are the ideal weight matrices. upper bound of norm and ideal weight matrix The upper bound of the norm; and They are basis function vectors and The bounded variation coefficients; the above parameters are used to transform the neural network approximation error term into a bounded constant term, thereby facilitating subsequent Lyapunov stability analysis.

[0021] According to formulas (9), (10), and (11), we can obtain: (12) in, This represents the lower bound of the convergence coefficients of the observer error system. This represents the bounded constant term composed of neural network weights, bounded perturbations, and approximation errors, where: , The largest eigenvalue of the matrix. ; This is the upper bound of the perturbation derivative; Solving formula (12), we get: (13) in, yes The initial value is selected from the preset design parameters. and And ensure that formula (8) holds, so as to minimize the state observation error. and It meets the preset observation performance requirements and converges to the preset range.

[0022] S4. Design the position outer loop control law based on the reference landing trajectory and the reconstructed information obtained based on the neural network extended state observer; In a specific embodiment, S4 includes the following steps for designing the outer-loop control law for the position based on the reference landing trajectory and the reconstructed information obtained from the neural network extended state observer: Within the outer position loop, the UAV position error is constructed based on the reference landing trajectory. for: (14) in, Indicates the reference landing trajectory; Differentiating the position error, we obtain the dynamic equation for the position error: (15) To ensure position error convergence, a virtual control law for the position loop is designed. for: (16) Among them, positive definite diagonal matrix ; Traditional backstepping design requires Performing successive differentiations can easily lead to increased complexity and computational burden. This embodiment introduces a first-order dynamic surface filter to filter the virtual control law of the position loop, thus avoiding repeated differentiation of the virtual control law. This includes: Let the output of the dynamic surface filter be The filtering process is as follows: (17) in, The filtering time constant; Based on formula (17), the position loop filtering error is defined as follows: The derivative of the position loop filtering error is obtained. ; Therefore, the position outer-loop control law can be achieved using only the filtered virtual control law. and its first derivative Participate in control design to avoid the problem of higher-order derivative expansion; Combining the reconstruction information obtained from the neural network extended state observer, formula (16), and position loop filtering error Design the outer loop control law for the location, including: The velocity error is constructed based on formula (17). for: (18) Differentiating the velocity error yields: (19) The Lyapunov preselection function is constructed to perform position loop performance analysis on the UAV, as shown in formula (20): (20) Differentiating from formulas (15), (19), and (20) yields: (twenty one) in, It is a composite disturbance of the drone's position channel. .

[0023] Existing control laws based on extended state observers typically employ a full compensation strategy, treating all disturbances uniformly as detrimental factors and suppressing them. However, this method neglects the varying effects of disturbances under different conditions. Therefore, this embodiment utilizes UAV position error... Speed ​​error The position loop filtering error and formula (21) are used to design the position outer loop control law, which is expressed as: (twenty two) Among them, positive definite diagonal matrix , To estimate the velocity error of the position loop obtained by reconstructing the velocity based on the neural network extended state observer, , It is the position loop conditional perturbation suppression smoothing judgment function matrix after being processed by the smoothing factor. It is an estimate of the composite disturbance of the UAV's position channel. , and It is the reconstructed information obtained by the neural network extended state observer; Based on Lyapunov energy analysis, a position loop condition perturbation suppression mechanism is designed, including: Substituting equation (22) into equation (21) yields:

[0024] Based on the above formula, perform energy analysis and select... As an evaluation metric, to determine whether a composite disturbance is beneficial, it includes: Determine whether If yes, it indicates that the composite disturbance has an adverse effect on the system error convergence, and it is necessary to control the corresponding channel to perform disturbance suppression or compensation, that is, let ,in It is the identity matrix; otherwise, i.e. When this occurs, it indicates that the composite disturbance promotes energy dissipation in the system, requiring control of the corresponding channel to retain the disturbance effect or reduce the compensation intensity, i.e., let .

[0025] Specifically, the composite disturbance estimation result obtained from the reconstruction result of the neural network extended state observer described in this embodiment, combined with the velocity error, is used to construct a composite disturbance discrimination term. Based on the comparison between the composite disturbance discrimination term and a preset threshold, a conditional disturbance suppression non-smoothing judgment function is obtained. A smoothing factor is used to process the conditional disturbance suppression non-smoothing judgment function to generate a position loop conditional disturbance suppression smoothing judgment function and an attitude loop conditional disturbance suppression smoothing judgment function, so as to realize shockless switching of disturbance processing state between different control stages and avoid sudden changes in control input.

[0026] Specifically, in this embodiment, since the non-smooth judgment function for suppressing conditional perturbations is prone to causing control signal chattering, a smoothing factor is used to smooth it. The position loop conditional perturbation suppression smooth judgment function matrix after smoothing factor processing is obtained. The first in j Each component Represented as: (twenty three) in, It is a smoothing factor and a positive constant.

[0027] This embodiment reconstructs internal and external disturbances, enabling the identification of their combined effects, and allowing for the selective utilization and suppression of disturbances based on the identification results.

[0028] Specifically, this embodiment addresses the problem of multiple sources of disturbances acting simultaneously during autonomous landing, including internal unknown nonlinearities, parameter perturbations, actuator efficiency changes, and external wind and waves. After obtaining the comprehensive disturbance estimation result, instead of employing a uniform full compensation approach, it further constructs a conditional disturbance suppression mechanism to discriminate and process the effects of disturbances. This approach avoids canceling out disturbance components that contribute to system error convergence and energy dissipation, thus overcoming the problems of increased control input spikes, increased energy consumption, and response lag that traditional full compensation strategies easily lead to, achieving a balance between control accuracy and input smoothness. Especially in the terminal phase of autonomous landing, this mechanism helps reduce control abrupt changes and attitude jitter, improving the stability and safety of the UAV during contact with the moving deck.

[0029] S5. Combining the aforementioned position outer loop control law, the desired attitude angle and thrust command of the UAV are obtained; Specifically, considering the relative attitude coordination requirements between the UAV and the unmanned surface vessel during the landing process, the expected bow angle of the UAV is set as follows: Setting it to the current heading angle of the unmanned surface vessel helps reduce the adverse effects of relative attitude mismatch, simplifies control design, and improves the stability and safety of the landing process.

[0030] In a specific embodiment, S5, the specific steps for obtaining the desired attitude angle and thrust command of the UAV by combining the outer loop control law include: The thrust command for the UAV is obtained based on the position outer loop control law. Based on the longitudinal, lateral, and vertical components of the outer-loop control law, and combined with the current desired heading angle of the UAV, the desired roll angle and desired pitch angle of the UAV are calculated using the following formulas: (twenty four) in, For the expected roll angle of the drone, For the desired pitch angle of the drone, The desired heading angle for the drone; Specifically, this embodiment generates a virtual control law based on the position outer-loop control law, and further combines disturbance compensation to construct thrust commands and desired attitude angles, forming a dual-closed-loop output feedback control structure from position control to attitude control. Based on this, by continuously generating attitude reference signals, the UAV can maintain a smooth transition of reference values ​​between stages such as approach, alignment, descent, and terminal landing, avoiding problems such as control shocks, amplified position errors, and unstable terminal contact caused by abrupt changes in the reference trajectory or control law during multi-stage switching.

[0031] S6. Construct an inner-loop attitude control law based on the desired attitude angle of the UAV, and realize the autonomous landing control of the quadcopter UAV based on the thrust command output by the outer-loop position control law and the inner-loop attitude control law.

[0032] In a specific embodiment, S6 includes the following steps for constructing the attitude inner-loop control law based on the desired attitude angle of the UAV: Within the attitude inner loop, the desired attitude of the UAV is obtained according to formula (24). Thus, attitude error is constructed. , represented as: (25) Taking the derivative of the attitude error, we obtain the dynamic equation of the attitude error: (26) To ensure the convergence of attitude error, a virtual control law for the attitude loop is designed. for: (27) Among them, positive definite diagonal matrix ; Traditional backstepping design requires Performing successive differentiation can easily lead to increased complexity in expression and computational burden. Therefore, this embodiment introduces a first-order dynamic surface filter to filter the virtual control law of the attitude loop, including: Let the output of the dynamic surface filter be The filtering process is as follows: (28) in, The filtering time constant; Based on formula (28), the attitude loop filtering error is defined as follows: Taking the derivative of the attitude loop filtering error, we get ; Therefore, the attitude inner-loop virtual control law can be obtained using only the filtered attitude inner-loop virtual control law. and its first derivative parameter This is achieved through control design, thereby avoiding the problem of higher-order derivative expansion.

[0033] Construct the angular velocity error based on formula (28) for: (29) in, The angular velocity of the drone; Differentiating the angular velocity error yields: (30) The Lyapunov preselected function is constructed to perform attitude loop performance analysis on the UAV, as shown in Equation (31): (31) Differentiating equations (30) and (31) yields: (32) in, It is a composite disturbance in the attitude channel of the UAV. ; Based on attitude error Angular velocity error Attitude loop filtering error Formulas (31) and (32) are used to design the attitude inner-loop control law, and the expression is: (33) Among them, positive definite diagonal matrix , To estimate the angular velocity error of the attitude loop obtained by reconstructing the angular velocity based on the neural network extended state observer, , It is the attitude loop conditional perturbation suppression smoothing judgment function after being processed by the smoothing factor. It is an estimate of the composite perturbation in the attitude channel of the UAV. , and It is the reconstructed information obtained by the neural network extended state observer; A mechanism for suppressing attitude loop conditional perturbations based on Lyapunov energy analysis is designed, including: Substituting formula (33) into formula (32) yields: ; Based on the above formula, and based on energy analysis, and selecting... As evaluation indicators to determine whether a disturbance is beneficial, the following are included: Determine whether If yes, it indicates that the composite disturbance has an adverse effect on the convergence of system error, and therefore needs to be addressed. To control the corresponding channel to perform disturbance suppression or compensation, that is, to... ,in It is the identity matrix; otherwise, i.e. When this occurs, it indicates that the composite disturbance promotes energy dissipation in the system, requiring control of the corresponding channel to retain the disturbance effect or reduce the compensation intensity, i.e., let .

[0034] Specifically, the attitude loop conditional perturbation suppression smoothing judgment function matrix after smoothing factor processing. The first in j Each component Represented as: (34) in, It is a smoothing factor and a positive constant.

[0035] Specifically, the position outer loop and attitude inner loop together form a complete dual-closed-loop control structure for autonomous landing of UAVs.

[0036] Specifically, the attitude inner-loop control law further selectively compensates for the composite disturbance estimation results through the attitude channel conditional disturbance suppression factor, and uses the Lyapunov function to jointly analyze the attitude error, angular velocity error and attitude loop filtering error to ensure that all signals of the closed-loop system are bounded.

[0037] Specifically, in this embodiment, a Lyapunov preselection function is constructed for performance analysis, expressed as follows: (35) Differentiating formula (35) yields: (36) make , According to Young's inequality , , , , and All of them are positive constants, and thus the following can be derived from formula (36): (37) in, This represents the lower bound of the convergence coefficient in the stability analysis of the closed-loop controller. , This is the bounded constant term in the stability analysis of the closed-loop controller. ; Furthermore, in the stability proof, selecting appropriate design parameters makes... , , , , , , , ; Integrating equation (36) yields: (38) in, Consistent and ultimately bounded, which can be further obtained , , , , and Both are bounded. , , and This indicates that the drone's position, velocity, attitude, and angular velocity are all bounded; at the same time and Bounded, control input can be obtained and Bounded. In summary, all signals in the closed-loop system are bounded.

[0038] like Figure 3As shown, this embodiment also provides an autonomous landing output feedback control system for a quadcopter UAV, used to realize the autonomous landing output feedback control method for a quadcopter UAV, including: airborne sensors, shipborne sensors, a cooperative control unit, a quadcopter actuator, and a wireless transmission network; The airborne sensor is used to collect motion status information of the quadcopter drone, and the shipborne sensor is used to collect motion status information of the unmanned surface vessel platform. The motion status information of the quadcopter drone and the unmanned surface vessel platform are transmitted to the collaborative control unit through the wireless transmission network. Specifically, the collected motion state information includes: the UAV's position, attitude, linear velocity, and angular velocity information measured by the airborne sensor module, and the unmanned surface vessel's position, attitude, velocity, and heading information measured by the shipborne sensor module. The airborne sensor module may include a GPS positioning unit, an IMU inertial measurement unit, a barometer, a visual sensor, or a laser rangefinder, while the shipborne sensor module may include a GPS positioning unit, an IMU inertial measurement unit, a heading sensor, and a deck attitude detection sensor.

[0039] The collaborative control unit processes the collected motion status information of the quadcopter UAV and the unmanned surface vessel platform, and completes the autonomous landing output feedback control of the quadcopter UAV on the mobile unmanned surface vessel platform in complex marine environments. Specifically, the collaborative control unit includes: a construction module, a reconstruction module, a position loop module, a discrimination module, an attitude calculation module, an attitude control module, and a driving module; The construction module is used to construct the dynamic and kinematic models of UAVs and unmanned surface vessels, and to obtain measurable output information. It is also used to design a UAV trajectory stage switching mechanism based on the kinematic and dynamic models of the unmanned surface vessels, thereby generating a reference landing trajectory for the UAV. The reconstruction module is used to construct a neural network extended state observer based on the dynamic and kinematic models of the UAV, thereby reconstructing the unmeasurable velocity state, internal unknown nonlinearity, and external disturbances of the UAV online. The position loop module is used to design the position outer loop control law based on the reference landing trajectory and the reconstructed information obtained based on the neural network extended state observer; The discrimination module is used to design a conditional disturbance suppression mechanism to discriminate and compensate for composite disturbances under different operating conditions, so as to reduce the impact of environmental disturbances and model uncertainties on control performance. The attitude calculation module is used to combine the position outer loop control law to obtain the desired attitude angle and thrust command of the UAV. The attitude control module is used to construct an inner-loop attitude control law based on the desired attitude angle of the UAV. The drive module is used to convert the thrust command output by the outer loop control law of position and the control torque command output by the inner loop control law of attitude into target drive parameters corresponding to each rotor, so as to drive the quadcopter UAV to achieve autonomous landing. The quadcopter actuator is used to execute control commands sent by the controller, specifically including rotation speed adjustment commands, thrust commands, and roll, pitch, and yaw control torque commands for each rotor, so as to achieve stable tracking of the reference trajectory of the mobile unmanned surface vessel platform by the quadcopter UAV and complete autonomous landing.

[0040] This embodiment also provides an electronic device, including a processor and a memory. The memory stores a computer program, which, when called by the processor, is used to execute a quadcopter unmanned aerial vehicle (UAV) autonomous landing output feedback control method.

[0041] This embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement an autonomous landing output feedback control method for a quadcopter unmanned aerial vehicle.

[0042] In this embodiment, an autonomous landing scenario is simulated, in which a drone approaches and lands on a moving vessel under adverse sea conditions. and The radial basis function neural networks all contain 20 nodes, with their centers evenly distributed in... Within the range, and all widths are 1. The initial conditions for the drone are: , , , The initial conditions for the unmanned surface vessel are: , , , Design parameters , , , . Weight update law parameters , , , Disturbance judgment condition parameters , The initial weight value is... and .

[0043] like Figure 4 As shown, the quadcopter UAV can effectively track the three-dimensional motion trajectory of the unmanned surface vessel (USV) and maintain it within approximately 0.2 meters above the designated landing area, demonstrating that the proposed method can achieve stable tracking and precise landing of the mobile USV. Figure 5 As shown, Conditional Disturbance Suppression (CDN) represents a control mechanism that selectively retains, weakens, or compensates for disturbances based on their actual impact on the system's motion state. Full Disturbance Compensation (FDC) represents a control strategy that treats all disturbances as adverse factors and compensates for them. Smooth Conditional Disturbance Suppression (S-CDN) represents a conditional disturbance suppression mechanism that uses a continuous smooth function to achieve state switching during disturbance handling. Non-Smooth Conditional Disturbance Suppression (NS-CDN) represents a conditional disturbance suppression mechanism that does not use a continuous smooth function for switching. Figure 5 It can be seen that, compared to FDC, the CDN control scheme enables faster convergence of position errors and achieves higher convergence accuracy; compared to NS-CDN, the position error variation is smoother after adopting S-CDN. Figure 6 As shown, compared to FDC, using CDN results in faster tracking speed and smaller attitude error for the quadcopter UAV in the desired attitude; compared to NS-CDN, the attitude error changes are smoother after using S-CDN. Combined with... Figures 7 to 9 It can be seen that the neural network extended state observer can accurately reconstruct the unmeasurable velocity state, attitude-related state, and combined perturbations of the quadcopter UAV, indicating that the integrated online reconstruction method for state and uncertainty constructed in this embodiment has good observation performance. Figure 10 and Figure 11 It can be seen that during control switching, compared to NS-CDN, the changes in thrust and control torque are more stable after adopting S-CDN, which can effectively reduce chattering caused by switching. Figure 12 and Figure 13 It can be seen that the smooth decision function can achieve a relatively smooth switching effect in both the outer position loop and the inner attitude loop, thereby improving the continuity of control input. From Figure 14 and Figure 15 It can be seen that the L2 norm of the weight estimates in both the position outer loop and the attitude inner loop remains bounded.

[0044] In practice, the final control commands are obtained from the position loop controller and attitude loop controller. These commands include thrust, roll control torque, pitch control torque, and yaw control torque. The control commands are parsed to extract the control parameters for each execution channel of the quadcopter UAV. Specifically, the thrust and three-axis control torque commands are converted into target speed commands, drive voltage commands, current commands, or pulse width modulation control laws for each rotor according to the quadcopter UAV's control distribution relationship. These control parameters are then transmitted to the corresponding actuator controllers via the onboard flight control bus. The actuator controllers include an electronic speed controller, a rotor drive motor controller, and an attitude execution control unit. Finally, differential adjustment is performed on the four rotors based on the control parameters corresponding to different actuators, enabling the quadcopter UAV to generate the required lift and attitude control torque. This drives the quadcopter UAV to track the reference trajectory and gradually approach the target landing area of ​​the unmanned surface vessel (USV) platform.

[0045] Specifically, the thrust command is mainly used to adjust the vertical rise and fall motion of the quadcopter UAV so that the UAV maintains the required altitude change trend during autonomous landing; the roll control torque command and pitch control torque command are mainly used to adjust the attitude changes of the quadcopter UAV in the lateral and longitudinal channels so that the UAV can accurately track the generated desired roll angle and desired pitch angle; the yaw control torque command is mainly used to adjust the yaw attitude of the quadcopter UAV so that the actual heading angle of the UAV is consistent with the heading angle of the UAV or meets the preset relative yaw angle constraint.

[0046] Furthermore, during the autonomous landing mission of the quadcopter UAV, the actuator controller continuously adjusts the rotational speed of the four rotors according to the real-time updated thrust and control torque commands. This enables the UAV to maintain stable tracking of the reference trajectory under the combined effects of the motion of the unmanned surface vessel platform, external disturbances in the marine environment, and unknown nonlinearities within the system, and continuously reduces the relative position and attitude errors between the UAV and the unmanned surface vessel platform.

[0047] Furthermore, when the quadcopter UAV enters the area above the preset landing zone and meets the relative position constraints, attitude constraints, and speed constraints of the landing phase, the control system continues to output thrust commands and control torque commands that match the current flight state to drive the quadcopter UAV to descend smoothly along the preset glide trajectory and finally land in the designated landing zone of the unmanned surface vessel platform, thereby completing the autonomous landing process of the quadcopter UAV on the unmanned surface vessel platform.

[0048] like Figures 16 to 23The diagram shown illustrates the simulation implementation of the quadcopter UAV autonomous landing output feedback control system provided in this embodiment under the MATLAB / Simulink environment. This embodiment uses this simulation system for verification to illustrate the implementation methods and workflows of the control method, system, and functional modules proposed in this invention. It should be noted that the simulation interface, module division, and specific implementation methods shown in this embodiment are only one feasible implementation example, used to illustrate the technical solution proposed in this embodiment, and are not the only limitation on the system implementation method. In practical engineering applications, those skilled in the art can adaptively adjust and optimize the module division, algorithm implementation methods, and parameter configurations according to the specific hardware platform, software environment, and control requirements.

[0049] Beneficial effects of this embodiment: (1) By unifying the position loop control results and attitude loop control results into the target drive parameters that can be executed by each rotor of the quadcopter UAV, an effective mapping from the high-level control law to the low-level actuator control law is realized, so that the controller output can directly act on the UAV execution system, thus enhancing the engineering feasibility. (2) In the process of outputting control commands, not only the lift and attitude adjustment requirements of the quadcopter UAV itself are considered, but also the motion state of the unmanned surface vessel platform and the relative position and attitude constraints during the landing phase are considered, so that the UAV can continuously correct its flight state under dynamic ship deck conditions and improve the trajectory tracking accuracy and end contact stability during the autonomous landing process. (3) To address the problem that the speed state is difficult to measure directly, continuously and with high precision during autonomous landing, the output process of thrust command and control torque command is combined with the online reconstruction results of speed state, internal unknown nonlinearity and external disturbance by the neural network extended state observer to estimate the speed state and comprehensive disturbance online. This avoids the strong dependence of traditional state feedback control on complete state information, so that even under adverse conditions such as sensor noise, platform vibration, visual obstruction and low-cost inertial device drift, the state information that meets the control requirements can still be obtained, improving the control feasibility and tracking accuracy of the system under conditions where speed cannot be directly and reliably measured. (4) By introducing the neural network extended state observer and conditional disturbance suppression mechanism into the generation and execution of thrust and control torque commands, the system can suppress harmful disturbances and selectively utilize beneficial disturbances when facing wind, waves, platform relative motion and internal unknown dynamic coupling effects in complex marine environments. This ensures that the quadcopter UAV can smoothly and accurately track the reference trajectory and complete autonomous landing under different working conditions. In complex marine environments, it can obtain higher trajectory tracking accuracy, better attitude stability and smoother control input and output characteristics, thereby improving the robustness, stability and intelligence of the overall control system.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for autonomous landing output feedback control of a quadcopter unmanned aerial vehicle, characterized in that, The specific steps include: S1. Construct dynamic and kinematic models for unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs); S2. Based on the kinematic and dynamic models of the unmanned surface vessel (USV), a trajectory phase switching mechanism is designed to provide a safe window for determining the trajectory phase switching of the USV, thereby generating a reference landing trajectory for the USV. S3. Based on the dynamic and kinematic models of the UAV, a neural network extended state observer is constructed to reconstruct the unmeasurable velocity state, internal unknown nonlinearity, and external disturbances of the UAV online. S4. Design the position outer loop control law based on the reference landing trajectory and the reconstructed information obtained based on the neural network extended state observer; S5. Combining the aforementioned position outer loop control law, the desired attitude angle and thrust command of the UAV are obtained; S6. Construct an inner-loop attitude control law based on the desired attitude angle of the UAV, and realize the autonomous landing control of the quadcopter UAV based on the thrust command output by the outer-loop position control law and the inner-loop attitude control law.

2. The autonomous landing output feedback control method for a quadcopter UAV according to claim 1, characterized in that, In S1, the specific steps for constructing the dynamic and kinematic models of UAVs and unmanned surface vessels include: Establishing a geodetic coordinate system UAV attached coordinate system and the coordinate system of the unmanned surface vessel ; The expressions for the constructed dynamic and kinematic models of the UAV are as follows: (1) Among them, the position vector of the UAV in the geodetic coordinate system , These represent the UAV in the geodetic coordinate system. Position coordinates along the axis, linear velocity vector of the UAV in the attached coordinate system , These represent the components of the UAV's velocity along the three coordinate axes of the corresponding coordinate system, and the attitude vector. , These represent the UAV's roll angle, pitch angle, and heading angle, respectively, and the angular velocity vector in the UAV's attached coordinate system. , These represent the coordinates of the UAV around the attached coordinate system. The roll rate, pitch rate, and yaw rate of the shaft. The total mass of the drone, Here is the rotational inertia matrix. For the thrust and control torque of the quadcopter drone , These represent the roll control torque, pitch control torque, and yaw control torque of the UAV, respectively, with the gravity term as the other term. , , Represents gravitational acceleration. and These represent the internal unknown nonlinear terms in the linear velocity channel and the angular velocity channel, respectively. and These represent the external disturbance terms formed by wind field, sea surface airflow disturbance, and platform relative motion induced effect, respectively. Represents the attitude rotation matrix. Represents the angular velocity transformation matrix. It is a bounded matrix; The expressions for the kinematic and dynamic models of the unmanned surface vessel are as follows: (2) Among them, the position of the unmanned surface vessel in the geodetic coordinate system , These represent the unmanned surface vessel in the geodetic coordinate system. Position coordinates along the axis; velocity of the unmanned surface vessel in the geodetic coordinate system , These represent the unmanned surface vessel in the geodetic coordinate system. Velocity components along the axial direction; attitude of the unmanned surface vessel in the hull appendage coordinate system. , These represent the roll, pitch, and bow angles of the unmanned surface vessel (USV), respectively, and the angular velocity of the USV. , These represent the angular velocity components of the unmanned surface vessel (USV) around the three coordinate axes of the hull's appendage coordinate system. Due to the inertia and added mass effect of the unmanned surface vessel, For hydrodynamic damping, ; , and These are the control force and torque generated by the propeller, respectively. and These represent the amplitude and frequency of the ocean waves, respectively. and External disturbance; This represents the linear velocity transformation matrix from the ship's attached coordinate system to the geodetic coordinate system; This represents the angular velocity transformation matrix from the ship's attached coordinate system to the geodetic coordinate system.

3. The autonomous landing output feedback control method for a quadcopter UAV according to claim 2, characterized in that, In S2, the specific steps for designing a trajectory phase switching mechanism based on the kinematic and dynamic models of the unmanned surface vessel (USV) to provide a safety window for trajectory phase switching and generate a reference landing trajectory for the USV include: The trajectory phase switching mechanism, designed based on the kinematic and dynamic models of the unmanned surface vessel (USV), to provide a safety window for determining trajectory phase switching, is expressed as follows: (3) in, The threshold for unmanned surface vessel attitude determination is denoted as , and ; When the heave, roll, and pitch motions of the unmanned surface vessel (USV) deck simultaneously satisfy the criteria in equation (3), it indicates that the USV deck has entered a safe landing window. This safe landing moment is recorded as the safe landing switching moment for the USV. This leads to the design of the desired trajectory for the UAV to smoothly transition from a safe offset position to the deck landing point, i.e., the reference landing trajectory. , is represented as: (4) in, , For safety redundancy height, For smoothing parameters; Represents the three-dimensional spatial trajectory of the unmanned surface vessel; For the current moment, It is the hyperbolic tangent function.

4. The autonomous landing output feedback control method for a quadcopter UAV according to claim 3, characterized in that, In S3, the specific steps for constructing a neural network extended state observer based on the dynamics and kinematics models of the UAV include: S31. Construct neural network approximators for the linear velocity channel and the angular velocity channel respectively to approximate the internal unknown nonlinear terms in the linear velocity channel and angular velocity channel of the UAV-based dynamics and kinematics models. The expressions are as follows: (5) in, These are the ideal weight matrices for the linear velocity channel and the angular velocity channel neural network approximation, respectively. These are the basis function vectors for the linear velocity channel and the angular velocity channel neural network approximator, respectively. Let represent the estimation error vector, and ,in and , where are positive constants, representing the upper bounds of the two types of approximation errors respectively; S32. Based on the dynamic model, kinematic model and formula (5) of the UAV, the position state of the UAV is defined. Unmeasurable speed state of drones Position-related perturbation state UAV attitude status Unmeasurable angular velocity state Attitude loop related perturbation state , , ; The error between the actual position and the observed position of the UAV is defined as: The error between the actual attitude state and the observed attitude state is Define the position loop weight error Attitude loop weight error , This is the ideal weight estimation matrix; Based on the above definition, a neural network extended state observer is designed, with the following expression: (6) in, , , , , and All are observer gains; These correspond to the estimated values ​​of the UAV's position, linear velocity, and position loop perturbation, respectively. These correspond to the estimated values ​​of the UAV's attitude, angular velocity, and attitude loop perturbation, respectively. It is the outer loop control law for the position to be designed; The adaptive weight update law for the neural network extended state observer is designed as follows: (7) In designing a neural network extended state observer, some parameters satisfy the following inequalities: (8) in, and All are positive definite matrices. Let the upper bound of the norm of the rotation matrix be , and This represents the forgetting coefficient.

5. The autonomous landing output feedback control method for a quadcopter UAV according to claim 4, characterized in that, In S4, the specific steps for designing the outer-loop control law for the position based on the reference landing trajectory and the reconstructed information obtained from the neural network extended state observer include: Within the outer position loop, the UAV position error is constructed based on the reference landing trajectory. for: (14) in, Indicates the reference landing trajectory; Differentiating the position error, we obtain the dynamic equation for the position error: (15) To ensure position error convergence, a virtual control law for the position loop is designed. for: (16) Among them, positive definite diagonal matrix ; A first-order dynamic surface filter is introduced to filter the virtual control law of the position loop, including: Let the output of the dynamic surface filter be The filtering process is as follows: (17) in, The filtering time constant; Based on formula (17), the position loop filtering error is defined as follows: The derivative of the position loop filtering error is obtained. ; Combining the reconstruction information obtained from the neural network extended state observer, formula (17), and position loop filtering error Design the outer loop control law for the location, including: The velocity error is constructed based on formula (17). for: (18) Differentiating the velocity error yields: (19) The Lyapunov preselection function is constructed to perform position loop performance analysis on the UAV, as shown in formula (20): (20) Differentiating from formulas (15), (19), and (20) yields: (21) in, It is a composite disturbance of the drone's position channel. ; Based on UAV position error Speed ​​error The position loop filtering error and formula (21) are used to design the position outer loop control law, which is expressed as: (22) Among them, positive definite diagonal matrix , To estimate the velocity error of the position loop obtained by reconstructing the velocity based on the neural network extended state observer, , It is the position loop conditional perturbation suppression smoothing judgment function matrix after being processed by the smoothing factor. It is an estimate of the composite disturbance of the UAV's position channel. , and It is the reconstructed information obtained by the neural network extended state observer; Based on Lyapunov energy analysis, a position loop condition perturbation suppression mechanism was designed, including: Substituting formula (22) into formula (21) yields: ； Based on the above formula, perform energy analysis and select... As an evaluation metric, to determine whether a composite disturbance is beneficial, it includes: Determine whether If yes, then enable disturbance suppression or compensation, that is, let ,in It is the identity matrix; otherwise, i.e. When this happens, disturbance suppression is turned off or compensation is reduced, causing... ; Position loop conditional perturbation suppression smoothing judgment function matrix after smoothing factor processing The first in j Each component Represented as: (23) in, It is a smoothing factor and a positive constant.

6. The autonomous landing output feedback control method for a quadcopter UAV according to claim 5, characterized in that, In S5, the specific steps for obtaining the desired attitude angle and thrust command of the UAV by combining the aforementioned outer-loop control law include: The thrust command for the UAV is obtained based on the position outer loop control law. Based on the longitudinal, lateral, and vertical components of the outer-loop control law, and combined with the current desired heading angle of the UAV, the desired roll angle and desired pitch angle of the UAV are calculated using the following formulas: (24) in, For the expected roll angle of the drone, For the desired pitch angle of the drone, The desired heading angle for the UAV.

7. The autonomous landing output feedback control method for a quadcopter UAV according to claim 6, characterized in that, In S6, the specific steps for constructing the attitude inner-loop control law based on the desired attitude angle of the UAV include: Within the attitude inner loop, the desired attitude of the UAV is obtained according to formula (24). Thus, attitude error is constructed. , is represented as: (25) Taking the derivative of the attitude error, we obtain the dynamic equation of the attitude error: (26) To ensure the convergence of attitude error, a virtual control law for the attitude loop is designed. for: (27) Among them, positive definite diagonal matrix ; A first-order dynamic surface filter is introduced to filter the virtual control law of the attitude loop, including: Let the output of the dynamic surface filter be The filtering process is as follows: (28) in, The filtering time constant; Based on formula (28), the attitude loop filtering error is defined as follows: Taking the derivative of the attitude loop filtering error, we get ; Construct the angular velocity error based on formula (28) for: (29) in, The angular velocity of the drone; Differentiating the angular velocity error yields: (30) The Lyapunov preselected function is constructed to perform attitude loop performance analysis on the UAV, as shown in Equation (31): (31) Differentiating equations (30) and (31) yields: (32) in, It is a composite disturbance in the attitude channel of the UAV. ; Based on attitude error Angular velocity error Attitude loop filtering error Formulas (31) and (32) are used to design the attitude inner-loop control law, and the expression is: (33) Among them, positive definite diagonal matrix , To estimate the angular velocity error of the attitude loop obtained by reconstructing the angular velocity based on the neural network extended state observer, , It is the attitude loop conditional perturbation suppression smoothing judgment function after being processed by the smoothing factor. It is an estimate of the composite perturbation in the attitude channel of the UAV. , and It is the reconstructed information obtained by the neural network extended state observer; A mechanism for suppressing attitude loop conditional perturbations based on Lyapunov energy analysis is designed, including: Substituting formula (33) into formula (32) yields: ； Based on the above formula, perform energy analysis and select... As evaluation indicators to determine whether a disturbance is beneficial, the following are included: Determine whether If yes, then enable disturbance suppression or compensation, that is, let ,in It is the identity matrix; otherwise, i.e. When this happens, disturbance suppression is turned off or compensation is reduced, causing... ; Specifically, the attitude loop conditional perturbation suppression smoothing judgment function matrix after smoothing factor processing. The first in j Each component Represented as: (34) in, It is a smoothing factor and a positive constant.

8. A quadcopter unmanned aerial vehicle (UAV) autonomous landing output feedback control system, used to implement the quadcopter unmanned aerial vehicle (UAV) autonomous landing output feedback control method as described in claim 1, characterized in that, include: Airborne sensors, shipborne sensors, collaborative control units, quadcopter actuators, and wireless transmission networks; The airborne sensor is used to collect motion status information of the quadcopter drone, and the shipborne sensor is used to collect motion status information of the unmanned surface vessel platform. The motion status information of the quadcopter drone and the unmanned surface vessel platform are transmitted to the collaborative control unit through the wireless transmission network. The collaborative control unit processes the collected motion status information of the quadcopter UAV and the unmanned surface vessel platform, and completes the autonomous landing output feedback control of the quadcopter UAV on the mobile unmanned surface vessel platform in complex marine environments. The collaborative control unit includes: a construction module, a reconstruction module, a position loop module, a discrimination module, an attitude calculation module, an attitude control module, and a driving module; The construction module is used to construct the dynamic and kinematic models of UAVs and unmanned surface vessels, and to obtain measurable output information. It is also used to design a UAV trajectory stage switching mechanism based on the kinematic and dynamic models of the unmanned surface vessels, thereby generating a reference landing trajectory for the UAV. The reconstruction module is used to construct a neural network extended state observer based on the dynamic and kinematic models of the UAV, thereby reconstructing the unmeasurable velocity state, internal unknown nonlinearity, and external disturbances of the UAV online. The position loop module is used to design the position outer loop control law based on the reference landing trajectory and the reconstructed information obtained based on the neural network extended state observer; The discrimination module is used to design a conditional disturbance suppression mechanism to discriminate and compensate for composite disturbances under different operating conditions, so as to reduce the impact of environmental disturbances and model uncertainties on control performance. The attitude calculation module is used to combine the position outer loop control law to obtain the desired attitude angle and thrust command of the UAV. The attitude control module is used to construct an inner-loop attitude control law based on the desired attitude angle of the UAV. The drive module is used to convert the thrust command output by the outer loop control law of position and the control torque command output by the inner loop control law of attitude into target drive parameters corresponding to each rotor, so as to drive the quadcopter UAV to achieve autonomous landing. The quadcopter actuator is used to execute control commands sent by the controller.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is called by the processor, it is used to execute the autonomous landing output feedback control method for a quadcopter UAV as described in claim 1.

10. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, is used to implement the autonomous landing output feedback control method for a quadcopter UAV as described in claim 1.

Images