Folding wing unmanned aerial vehicle vertical launching transition process control method

Abstract

The invention relates to the technical field of unmanned aerial vehicle flight control, and particularly discloses a folding wing unmanned aerial vehicle vertical launching transition process control method. The method aims at solving the problem of attitude stability control caused by aerodynamic configuration time varying and unfolding asymmetry in the slow wing unfolding stage of the folding wing unmanned aerial vehicle after vertical launching. According to the technical scheme, the method comprises the steps that a multi-rigid-body nonlinear dynamic model comprising a fuselage and four independent wings is constructed, and additional inertia force, moment and time-varying gravity moment in the wing unfolding process are considered; the nonlinear model is converted into a linear parameter variation (LPV) model with the fuselage pitch angle and the wing sweepback angle as double scheduling parameters; a gain scheduling controller is designed based on a protection mapping principle, stable control gain covering a full vertical envelope is generated offline, and attitude stable control is realized online through real-time parameter table look-up. According to the method, the vertical launching success rate and safety are remarkably improved, the wings are supported to be unfolded gently so as to reduce mechanism impact and maintenance cost, implementation of an airborne embedded platform is facilitated, hardware requirements are reduced, and the method is suitable for vertical launching attitude control scenes of the cylindrical launching type folding wing unmanned aerial vehicle.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) flight control technology, and in particular to a method for controlling the vertical launch transition process of a folding-wing UAV. Background Technology

[0002] After being ejected from the launch tube, the folded wings of a cannon-launched folding-wing UAV need to unfold rapidly. In existing technologies, when unfolding relies on passive mechanisms such as springs, manufacturing tolerances, wear, or aerodynamic load disturbances often make it difficult to achieve strict synchronization between the two wings during unfolding. This asymmetrical unfolding can instantly generate significant roll or yaw disturbance moments, easily leading to attitude divergence in the UAV, or even launch failure.

[0003] To circumvent the control challenges of the aforementioned transition phase, existing technologies tend to employ a high-speed deployment strategy, where the wing deployment process is completed within 0.25 seconds to minimize the UAV's dwell time in this unstable state. However, this high-speed deployment method places extremely high instantaneous impact loads on the deployment mechanism (such as springs and locking mechanisms), which can easily lead to structural fatigue and component damage, thereby increasing the maintenance costs throughout the entire lifecycle and the risk of mission failure.

[0004] Therefore, one improvement trend is to use independent actuators (such as servo motors) to achieve controlled and slow wing deployment, thereby reducing the impact on the mechanism. However, this significantly extends the wing deployment time, requiring the UAV to simultaneously complete the attitude transition from vertical launch to level flight while its aerodynamic shape is constantly changing. In this complex time-varying process, an advanced controller is urgently needed to ensure the stable convergence of the UAV's key attitude angles.

[0005] Traditional gain scheduling methods suffer from high design complexity and difficulty in ensuring stability. In scenarios where the wing slowly unfolds, causing drastic time-varying and uncertain aerodynamic parameters, they are prone to runaway risks due to improper interpolation point selection or dynamic inconsistency, and cannot meet the high-precision attitude control requirements of the vertical launch transition process of folding-wing UAVs. Summary of the Invention

[0006] The purpose of this invention is to overcome the deficiencies in the prior art and provide an attitude control method for the vertical launch transition process of a folding-wing unmanned aerial vehicle (UAV). This method aims to solve the attitude control problem caused by the drastic time-varying and uncertain aerodynamic parameters during the slow wing deployment process, particularly avoiding the high design complexity and instability issues associated with traditional gain scheduling methods. It improves the success rate and safety of vertical launch, reduces mechanical impact and maintenance costs, and facilitates airborne engineering implementation.

[0007] To achieve the above objectives, the present invention provides the following solution: A method for controlling the vertical launch transition process of a folding-wing unmanned aerial vehicle (UAV) includes: The UAV is decoupled into five rigid body modules: the fuselage and four independent wings. A corresponding nonlinear dynamic model is established, including aerodynamic force, thrust, gravity, and additional inertial force, inertial torque, and time-varying gravitational torque generated during the slow and controlled deployment of the wings. The fuselage pitch angle and wing sweep angle are selected as dual scheduling parameters. A grid is set in a two-dimensional parameter space. The nonlinear model is balanced and Jacobian linearized at each grid vertex, and the nonlinear dynamic model is transformed into a linear parameter variation model. Based on the principle of protection mapping, a sequence of control parameters covering the transition region is generated. Stable control gains under each flight mode are established through offline boundary extension calculation. Lightweight table lookup is performed online based on real-time pitch angle and wing deployment angle to achieve speed and trajectory climb angle tracking, and complete attitude transition control from vertical launch to level flight.

[0008] Furthermore, the wing deployment angle is a known function of time t, as shown in the following formula: in, and These are the initial deployment angle and the final deployment angle of the wing, respectively. This refers to the time taken for the wings to deploy.

[0009] Furthermore, the additional inertial force, inertial torque, and time-varying gravitational torque generated during the slow and controlled deployment of the wing are: When the front and rear wings are deployed symmetrically from left to right, they will generate the following additional inertial forces in the airframe coordinate system: Meanwhile, the inertial torque caused by wing rotation and center of mass shift can be expressed in the body coordinate system as: In addition, the gravitational torque caused by gravity and the displacement of the wing position is: The geometric relationships between the terms are defined as follows: In the formula, This indicates the distance from the fuselage center of mass to the canard hinge axis along the fuselage coordinate system. Distance between axes From the fuselage center of mass to the rear wing hinge axis Axial position; The wing's center of mass relative to the fuselage's center of mass is at Distance along the axial direction; Indicates the mass of a single wing; l This is the distance from the wing's center of mass to the hinge axis; These represent the sweep angles of each wing surface, where 1 and 2 correspond to the forewings, and 3 and 4 to the afts, symmetrically distributed left and right. α is the fuselage pitch angle, and α is the UAV angle of attack; L It is aerodynamic lift.

[0010] Furthermore, the expression for aerodynamic force is: In the formula, The lift coefficient, The drag coefficient, This is the pitch moment coefficient. For reference wing area, For the average aerodynamic chord length, air density, For drone speed.

[0011] Furthermore, the expression for gravity is: in, m For the overall quality of the drone, This is the acceleration due to gravity.

[0012] Furthermore, the expression for the thrust is: in, Throttle position, This is the maximum thrust coefficient.

[0013] Furthermore, the trim and Jacobian linearization process includes: solving for the static trim point in the two-dimensional parameter space of the wing deployment angle and pitch angle, temporarily treating the time-varying term caused by the change in wing deployment angle as a constant value, and performing Jacobian linearization near the trim point to obtain a small perturbation model in the state space.

[0014] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: This method constructs a control law covering the entire vertical envelope in one go through a systematic stable boundary expansion mechanism. Even under time-varying aerodynamic characteristics caused by controlled wing deployment, it maintains stable fuselage attitude, fundamentally avoiding the risk of runaway due to improper interpolation point selection or dynamic inconsistencies in traditional methods, and reducing the probability of launch failure.

[0015] This method allows the wing to transition at a smoother and more controllable speed, effectively reducing the dynamic impact on the folding mechanism and locking device. In vertical launch missions, this means an extended fatigue life of the deployment mechanism, significantly reduced life-cycle maintenance costs, while maintaining rapid launch response capabilities and avoiding the structural fatigue and component damage problems associated with traditional high-speed deployment strategies.

[0016] This method can generate the applicable scheduling parameter range corresponding to the control gain offline before vertical launch. During the transition, it is only necessary to select the corresponding gain according to the scheduling parameter range in which the UAV is located. This avoids complex online optimization or high-frequency real-time gain calculation, making it suitable for resource-constrained airborne embedded platforms, reducing the computing performance requirements of airborne hardware, and possessing good engineering feasibility. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1. Overall schematic diagram; Figure 2. Comparison of amplitude-frequency response from throttle to airspeed; Figure 3. Comparison of amplitude-frequency response from throttle to track climb angle; Figure 4. Comparison of amplitude-frequency response from elevator to airspeed; Figure 5. Comparison of amplitude-frequency response from elevator to track climb angle; Figure 6. Schematic diagram of four common protection mappings; Figure 7. Flowchart of the two-parameter scheduling algorithm based on protection mapping; Figure 8. Comparison of open-loop and closed-loop pole locations; Figure 9. Comparison of controlled variables (track angle, velocity) curves; Figure 10. Comparison of control quantities. Detailed Implementation

[0019] 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, and 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.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] This invention provides a method for controlling the vertical launch transition process of a folding-wing unmanned aerial vehicle (UAV), comprising: First, this invention constructs a high-precision multi-rigid-body nonlinear dynamic model. This model decouples the UAV system into five rigid-body modules: the fuselage and four independent wings, and accurately derives the system's dynamic equations using the Newton-Euler method. The core of this model lies in its consideration not only of conventional aerodynamic forces, thrust, and gravity, but also, and more importantly, of the additional inertial forces, additional inertial moments, and time-varying gravitational moments generated during the slow, controlled deployment of the wings.

[0022] Secondly, to facilitate controller design, this invention transforms the nonlinear model into a linear parameter variation (LPV) model. This method selects the fuselage pitch angle, which has the most significant impact on system dynamics. With wing sweep angle As a dual scheduling parameter. (Through...) By setting up a grid in a two-dimensional parameter space and balancing and Jacobian linearizing the nonlinear model at each grid vertex, this invention obtains a family of linear state spaces describing the local dynamics of the system. These vertex systems collectively constitute a multicellular LPV model, which can be used for real-time bilinear interpolation to... and The form accurately approximates the time-varying characteristics of the UAV under arbitrary configurations during the transition process.

[0023] Finally, this invention designs a gain scheduling controller for the vertical launch control process. This gain scheduling control addresses the time-varying disturbances and aerodynamic coefficient fluctuations in the pitch channel during the transition from vertical to level flight for folding-wing UAVs. Based on the principle of protective mapping, it systematically generates a sequence of control parameters covering the transition region. Through offline boundary extension calculations, stable control gains for each flight mode are pre-established. Online tracking of speed and trajectory climb angle can be achieved simply by performing a lightweight lookup based on the real-time pitch angle and wing deployment angle. This method significantly reduces the onboard computational load while ensuring launch safety, providing an engineering-featured solution for the stable control of longitudinal motion during the vertical launch phase.

[0024] 1. Coordinate system definition and unfolding motion analysis After catapult launch, the wings begin to deploy. Under the action of the actuators, the front wing rotates 90° towards the nose, and the rear wing rotates 90° towards the tail. Therefore, during deployment, the front wing is swept forward, and the rear wing is swept backward. The vertical tail fin's function is to improve lateral stability. Let... Indicates the center of gravity of the fuselage. The rotation center of each airfoil, This represents the center of mass of the corresponding wing, and arrive The distance is denoted as ; This represents the sweep angle of the airfoil. Among them, Corresponding in order to the left front wing, right front wing, left rear wing, and right rear wing (e.g.) Figure 1 (As shown). When fully deployed, all sweep angles are zero. From a top-down view, the clockwise sweep angle is defined as positive; therefore, the sweep angle of the left front wing and right rear wing changes from -90° to 0°, while that of the right front wing and left rear wing changes from +90° to 0°. This is because the fuselage is relative to its coordinate system... Planar symmetry, negligible and Two inertial products.

[0025] 2. Modeling of wing deployment motion To simplify the dynamic analysis of the wing deployment process, this paper treats the wing deployment angle as a known function of time t, as shown in the following formula: (1) in, and These are the initial deployment angle and the final deployment angle of the wing, respectively. This refers to the time taken for the wing deployment process. Therefore, combining this with the fore and aft wing deployment angle process from the previous section, the initial wing deployment angle can be set manually. Target deployment angle , unfolding time Therefore, the following relationship holds: (2) 3. Overview of Multi-rigid-body Dynamics Modeling for Unmanned Aerial Vehicles Since the wings cannot be simply considered as a single rigid body structure during deployment, this paper divides the entire aircraft into five rigid body modules, including the fuselage and four independent wings, to more accurately describe the dynamic characteristics of this complex system, and establishes a corresponding multibody dynamics model. Considering that the vertical tail has a relatively small impact on the aerodynamic and dynamic response of the system during deployment, this part is not included in the model for the time being.

[0026] In terms of analytical methods, this paper adopts the Newton-Euler method, which combines the classic Newton's laws and Euler's rotation equations. This method can effectively describe the translational and rotational coupling between components in vector form and has good physical interpretability.

[0027] The changes in momentum and angular momentum of the system during deployment can be represented in the body coordinate system as follows: (3) in, For the system momentum, Angular momentum Let the linear velocity of the center of mass be denoted as . Angular velocity, The rotational speeds of each wing are given, and the system static moment is: (4) in This indicates the position of the fuselage's center of mass in the fuselage coordinate system. For the first The position vector of the center of mass of each wing.

[0028] Assuming the wing mass is uniformly distributed along its span, its inertia matrix in the body coordinate system is... It can be calculated using line integrals, in the following form: (5) The overall inertia matrix of the aircraft is obtained by superimposing the inertia matrices of the fuselage and each wing: (6) According to the law of conservation of momentum and the relationship between the conservation of angular momentum, we have: (7) Finally, combining the above equations yields the multibody dynamics system equations describing the motion characteristics of the UAV during deployment. Based on these equations, it is then necessary to model the forces and moments, aerodynamic forces and moments, gravity and thrust resulting from the symmetrical deployment of the wing, and substitute them back to obtain the nonlinear dynamic equations for the UAV's longitudinal plane.

[0029] 3.1. Additional forces and moments caused by symmetrical wing deployment In deriving the dynamic model of the UAV wing deployment phase, a complete six-degree-of-freedom multibody dynamics equation has been constructed. Since this paper primarily focuses on the motion characteristics in the longitudinal plane, variables related to the lateral and longitudinal directions are ignored to simplify the analysis. Based on this, expressions for the additional inertial forces and moments caused by wing deployment can be further derived.

[0030] When the front and rear wings are deployed symmetrically from left to right, they will generate the following additional inertial forces in the airframe coordinate system: (8) Meanwhile, the inertial torque caused by wing rotation and center of mass shift can be expressed in the body coordinate system as: (9) In addition, the gravitational torque caused by gravity and the displacement of the wing position is: (10) The geometric relationships between the terms are defined as follows: (11) In the formula, This indicates the distance from the fuselage center of mass to the canard hinge axis along the fuselage coordinate system. Distance between axes From the fuselage center of mass to the rear wing hinge axis Axial position; The wing's center of mass relative to the fuselage's center of mass is at Distance along the axial direction; Indicates the mass of a single wing; denoted as the sweep angle of each wing surface, where 1 and 2 correspond to the forewing and 3 and 4 to the aft wing, and are symmetrically distributed on the left and right sides.

[0031] 3.2. Aerodynamic and Torque Modeling The aerodynamic lift, drag, and pitching moment are shown below: (12) The aerodynamic coefficients are in the following form: (13) In the formula, The lift coefficient, The drag coefficient, This is the pitch moment coefficient. For reference wing area, For the average aerodynamic chord length, For reference exhibition length, This refers to air density.

[0032] 3.3. Gravity and Thrust Modeling (1) Gravity modeling The gravitational force acting on a drone during flight originates from its own mass. With gravitational acceleration The product of these forces, the force, is directed downwards vertically in the ground coordinate system. Its vector form in the geographic coordinate system is shown below: (14) It is evident that gravity only acts on the ground coordinate system. The axis direction, and its size is proportional to the overall mass of the machine.

[0033] (2) Engine thrust modeling The magnitude of thrust is mainly determined by the maximum thrust coefficient. With control input throttle quantity Together, they determine that its direction of action is along the body coordinate system. Axial direction. The thrust vector can be expressed as: (15) To facilitate modeling and ensure the system has sufficient power reserves, this paper will Set to a constant value of 500, and assume thrust and throttle position. The relationship is linear.

[0034] 3.4. Longitudinal Plane Nonlinear Equations When an aircraft transitions between attitudes, it often performs large pitch maneuvers. When the pitch angle... Beyond ±45°, traditional models based on the small-angle assumption will fail to accurately describe the aircraft's motion, and may even exhibit mathematical singularities. To overcome this problem, this paper employs the Euler angle parameterization method, selecting an appropriate attitude description method based on the pitch angle: when the pitch angle is within ±45°, conventional horizontal Euler angles are used (…). Modeling is used; when the value exceeds this range, vertical Euler angles are introduced. To avoid the problem of unusual posture.

[0035] To accurately reflect the motion characteristics of the aircraft during transition flight and to simplify and decouple the mathematical model, this paper extracts the longitudinal motion model based on the no-slip assumption and constructs dynamic expressions suitable for both horizontal and vertical Euler angle systems. The following transformation relationship is required when switching between the two coordinate systems: (16) At this point, the longitudinal nonlinear dynamic model of the system is as follows: (17) Given the inherently time-varying dynamic characteristics of the deployment process, traditional fixed-parameter control strategies are insufficient to meet the control requirements of this stage. Therefore, this paper introduces a Linear Parameter Variation (LPV) modeling framework to characterize the nonlinear time-varying dynamic behavior of the wing deployment process of a variator aircraft. Using the LPV method, the nonlinear system can be approximated as a family of state-dependent linear systems within a specific state interval, thus laying the foundation for the design of a gain-scheduled controller.

[0036] Within this framework, the nonlinear dynamics model of the variator can be constructed into a set of multi-cell LPV models based on dual-parameter scheduling of wing deployment angle and pitch angle by trimming and linearizing the nonlinear system under different configurations.

[0037] 1. Balancing and Linearization Based on the longitudinal nonlinear motion equations, to obtain the vertex system of the LPV model, trim analysis must first be performed under multiple typical configurations, followed by Jacobian linearization of the system. Unlike traditional fixed-configuration aircraft, the trim point of a variant aircraft will vary with the wing deployment angle. and pitch angle The deviation occurs due to changes in the wing's spread angle. Therefore, it is necessary to adjust the wing's spread angle accordingly. With pitch angle Within the two-dimensional parameter space, the static trim points for each typical configuration are solved. The settings for each trim point (taking level flight as an example) are shown in the table below: Table 1 Single-point balancing settings

[0038] During the trim process, the time-varying term caused by the change in wing spread angle can be temporarily treated as a constant value, ignoring its dynamic disturbance effect. Thus, Jacobian linearization is performed near the trim point, resulting in a small-perturbation state-space model of the following form: (18) in, For state vectors, To control the input, For a bounded time-varying perturbation, the parameter vector The weighting coefficients corresponding to the current flight state are defined as follows: (19) 2. Weight Calculation and Vertex Interpolation Methods To achieve model interpolation under arbitrary combinations of pitch and wing deployment angles, it is necessary to calculate online the weight coefficients of each vertex within the current working point's cell. Taking a rectangular cell with 4 vertices as an example, its linear interpolation coefficients are calculated using the following formula: (20) in, , These represent the current pitch angle and wing deployment angle, respectively.

[0039] 3. Results and Analysis The Linear Parameter Variation (LPV) model constructed in this invention is based on multiple typical steady-state configurations. Local linearization is performed at these trim points, and a two-parameter interpolation technique is used to model and approximate the nonlinear time-varying characteristics of the system. To verify the adaptability and dynamic response fidelity of the LPV model under untrimmed conditions, a set of scheduling parameter points (wing span angle) not used in the modeling process are selected. Pitch angle Simulation verification was conducted. By comparing the frequency response results of the LPV model and the complete nonlinear system using Bode plots, the dynamic consistency of the model across different frequency ranges can be intuitively reflected, providing a reliable basis for subsequent controller design.

[0040] exist Figures 4 to 5 Medium, throttle to airspeed The amplitude-frequency responses of the two systems almost coincide, indicating that the LPV model can accurately approximate the nonlinear system in the main power channel; arrive Although there are slight deviations in the high-frequency bands within the channels, the overall trend is consistent. exist Figures 6 to 7 In the middle, elevator to airspeed In the response, the LPV model slightly underestimates the mid-frequency gain, possibly due to the pitch coupling effect not being fully expressed; while in the elevator... to track climb angle In the channel, the LPV model exhibits a slight phase lag at low frequencies, but this is within acceptable limits.

[0041] Controller Design Detailed Process As discussed above, given the longitudinal nonlinear model of the UAV and the known state point matrices within the scheduling parameter range, hybrid sensitivity can be applied. Hybrid sensitivity controller design is a method to address the trade-off between performance and robustness in multivariable feedback systems. It serves the following purposes: (1) Performance: By minimizing the weighted sensitivity function The norm directly constrains the tracking error of the system. To ensure the system responds to the reference signal It has good low-frequency tracking capability and resistance to external disturbances. Its low-frequency suppression capability.

[0042] (2) Control constraints: By minimizing the weighted average The norm of a function restricts and controls input. The amplitude is adjusted to avoid actuator saturation and suppress high-frequency noise.

[0043] (3) Robustness: This method can achieve a balance between performance and stability margin, and improves the tolerance of the closed-loop system to model uncertainty.

[0044] Therefore, the state quantities of the state-space equation Measure output Controlled output Control input quantity If the external input consists of a reference input and a disturbance, then we have .in Compared to the disturbance inputs calculated at each steady-state operating point, taking into account the wing deployment angle, angular velocity, state variables, and range of change, it can be considered as... .

[0045] Based on the preceding analysis of the longitudinal dynamics of the UAV, a design was made at the nominal state point where the wing is in a specific fixed shape. Hybrid Sensitivity Controller. This controller aims to minimize tracking errors related to the UAV's speed and flight path angle, and ensures the elimination of steady-state errors through an integral feedback mechanism. Simultaneously, the design incorporates attitude disturbances caused by wing deformation as equivalent external disturbance inputs, integrating them into the generalized controlled object to enhance the system's anti-interference capability.

[0046] However, considering the significant changes in aerodynamic coefficients and system dynamics caused by wing deformation in actual missions, the controller gain obtained from a single linearized state point cannot meet the robust performance and stability requirements of the aircraft throughout its entire flight envelope. Especially during transitions with drastic parameter changes, the performance of a fixed-gain controller may deteriorate significantly or even lead to instability. Therefore, it is necessary to apply the aforementioned scheduling algorithm based on protection mapping to solve the problem.

[0047] Specific applications of scheduling algorithms To achieve robust performance across the entire flight envelope, an iterative, protection-map-based gain scheduling algorithm is employed. Unlike traditional methods that design and interpolate controllers for each discrete operating point, this approach ensures generalized stability across the entire parameter space. The entire design process is divided into two main phases: first, at a fixed wing spread angle... Next, design a pitch angle-dependent function. A variable controller; then, it is extended to simultaneously cover pitch angle. and wing spread angle The complete parameter domain.

[0048] Step 1: Single-parameter gain scheduling In the first phase, the wing deployment angle is temporarily fixed. And treat the system as dependent only on pitch angle The single-parameter LPV model.

[0049] 1. Initialization: Starting from the boundary of the pitch angle working domain To begin, design an initial static controller. This controller must ensure that the closed-loop system meets performance specifications at the initial operating point; for example, the poles must be strictly located within the target region. Inside.

[0050] 2. Iteration and Expansion: (1) Robustness analysis: using protection mapping First, analyze the controller. Maximum pitch angle range that can maintain system stability This boundary The case where the protection mapping is zero (according to Theorem II.1) means that the system stability has reached its limit.

[0051] (2) New controller synthesis: If discovered The stable range is insufficient to cover the entire pitch angle domain, so the operating point is moved to the stable boundary. At this point, a new controller will be found in the gain parameter space through optimization. This new controller is able to pull the system poles back to the target region. This allows for the acquisition of a new stability margin within the internal structure.

[0052] (3) Repeat: Repeating the above analysis and synthesis steps can generate a controller sequence. Each controller in this sequence covers its corresponding pitch angle sub-range, and these sub-ranges together constitute the complete pitch angle operating domain. .

[0053] This approach avoids designing separately for each pitch angle, instead iteratively expanding the applicability of a single controller, thus ensuring efficiency and stability throughout the process.

[0054] Step 2: Expand to two parameters Building on the first phase, the design will now be expanded to include the wing deployment angle. The complete two-parameter domain.

[0055] 1. Layered Iteration: A layered iterative method is adopted. First, the wing spread angle is... It is considered as an outer iteration variable.

[0056] 2. Inner layer scheduling: In each fixed... Under the given value, repeat the single-parameter gain scheduling algorithm of the first stage, and design a algorithm for pitch angle. Dispatch controller This process will provide specific The optimal controller gain function under the given value.

[0057] 3. Outer Layer Expansion: Next, we will analyze this scheduled controller. The maximum wing deployment angle range that can stabilize the system Similarly, this boundary is determined by finding cases where the protection mapping is zero.

[0058] 4. Repetition: If the interval is insufficient to cover the entire wing deployment angle domain, repeat at the stability boundary. Repeat the inner scheduling process and design a new controller set. Repeat this process until the entire pitch angle-wing deployment angle parameter plane is completely covered.

[0059] Ultimately, through this nested iterative method, a controller gain surface that guarantees stability and performance across the entire two-parameter domain is obtained. This surface can be directly implemented using a lookup table method combined with multilinear interpolation, thus providing a complete and feasible controller solution for the LPV model.

[0060] Control effect verification 1.1. Comparison of open-loop and closed-loop pole locations This paper selects three basic protection mapping regions to superimpose to form the final protection region, with the following indicators: (1) (2) Damping ratio (3) Frequency .Depend on Figure 8 It can be seen that before control is applied, not all open-loop poles fall within the protection region, but after control is applied, all closed-loop poles are located in the protection map.

[0061] 1.2. Comparison of Control Effects To compare the improved gain scheduling algorithm, the simulation was started with initial conditions of a pitch angle of 90°, a wingspan angle of 50°, and an initial airspeed of 22 m / s. The termination condition was that the aircraft completed wing deployment and transitioned from vertical to level flight. The simulation time was 50 seconds. The algorithms applied were traditional gain scheduling (Algorithm 1), chained switching control (Algorithm 2), and an improved gain scheduling method based on guard mapping theory (Algorithm 3). Simulation images are all represented as deviations from the initial state.

[0062] Depend on Figure 9 and Figure 10 It can be seen that all three algorithms can quickly enter the convergence state. Compared with the traditional gain scheduling algorithm and chained switching control, the protection mapping-based algorithm has a much faster convergence speed and higher tracking accuracy in track angle tracking. It can enter a stable state within 0.5s and maintain a deviation of less than 1° throughout the tracking process, while other algorithms require more than 1s to stabilize in the final state and the tracking deviation exceeds 2°. In terms of speed stabilization, other algorithms perform better in speed stabilization because their response speed is not sensitive, but they are insufficient in dynamic response. Although the protection mapping-based algorithm has a larger deviation in speed stabilization, its dynamic response is much higher than the other two algorithms. After the transition to level flight, it can converge to around 22m / s within 0.8s, while other algorithms require more than 3s to fully enter the convergence state.

[0063] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0064] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for controlling the vertical launch transition process of a folding-wing unmanned aerial vehicle, characterized in that, Includes the following steps: The UAV is decoupled into five rigid body modules: the fuselage and four independent wings. A corresponding nonlinear dynamic model is established, including aerodynamic force, thrust, gravity, and additional inertial force, inertial torque, and time-varying gravitational torque generated during the slow and controlled deployment of the wings. The fuselage pitch angle and wing sweep angle are selected as dual scheduling parameters. A grid is set in a two-dimensional parameter space. The nonlinear model is balanced and Jacobian linearized at each grid vertex, and the nonlinear dynamic model is transformed into a linear parameter variation model. Based on the principle of protection mapping, a sequence of control parameters covering the transition region is generated. Stable control gains under each flight mode are established through offline boundary extension calculation. Lightweight table lookup is performed online based on real-time pitch angle and wing deployment angle to achieve speed and trajectory climb angle tracking, and complete attitude transition control from vertical launch to level flight.

2. The vertical launch transition process control method for folding-wing UAVs according to claim 1, characterized in that, The wing's deployment angle is a known function of time t, as shown in the following formula: in, and These are the initial deployment angle and the final deployment angle of the wing, respectively. This refers to the time taken for the wings to deploy.

3. The vertical launch transition process control method for folding-wing UAVs according to claim 1, characterized in that, The additional inertial force, inertial moment, and time-varying gravitational moment generated during the slow and controlled deployment of the wing are: When the front and rear wings are deployed symmetrically from left to right, they will generate the following additional inertial forces in the airframe coordinate system: Meanwhile, the inertial torque caused by wing rotation and center of mass shift can be expressed in the body coordinate system as: In addition, the gravitational torque caused by gravity and the displacement of the wing position is: The geometric relationships between the terms are defined as follows: In the formula, This indicates the distance from the fuselage center of mass to the canard hinge axis along the fuselage coordinate system. Distance between axes From the fuselage center of mass to the rear wing hinge axis Axial position; The wing's center of mass relative to the fuselage's center of mass is at Distance along the axial direction; Indicates the mass of a single wing; l This is the distance from the wing's center of mass to the hinge axis; These represent the sweep angles of each wing surface, where 1 and 2 correspond to the forewings, and 3 and 4 to the afts, symmetrically distributed left and right. α is the fuselage pitch angle, and α is the UAV angle of attack; L It is aerodynamic lift.

4. The vertical launch transition process control method for folding-wing UAVs according to claim 1, characterized in that, The expression for aerodynamic force is: In the formula, The lift coefficient, The drag coefficient, This is the pitch moment coefficient. For reference wing area, For the average aerodynamic chord length, air density, For drone speed.

5. The control method for the vertical launch transition process of a folding-wing UAV according to claim 1, characterized in that, The expression for gravity is: in, m For the overall quality of the drone, This is the acceleration due to gravity.

6. The control method for the vertical launch transition process of a folding-wing UAV according to claim 1, characterized in that, The expression for the thrust is: in, Throttle position, This is the maximum thrust coefficient.

7. The vertical launch transition process control method for folding-wing UAVs according to claim 1, characterized in that, The trim and Jacobian linearization process includes: solving for the static trim point in the two-dimensional parameter space of the wing deployment angle and pitch angle, temporarily treating the time-varying term caused by the change in wing deployment angle as a constant value, and performing Jacobian linearization near the trim point to obtain a small perturbation model in the state space.

Images