A method and device for predicting landing trajectory of large-wingspan UAV
By constructing an aerodynamic model and expanding Kalman filter for the large wingspan drone under the sudden wind disturbance, the flight control adjustment difficulty and recycling route offset caused by sudden wind disturbance during landing are solved, and the problem of recovery route offset is achieved. Accurate prediction and safe recycling of drone trajectories.
Patent Information
- Application Number
- CN202311726793.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-15
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2043-12-15
AI Technical Summary
When the large wingspan drone encounters sudden wind disturbances during the near-ground stage, it is difficult to adjust the flight control, resulting in a deviation in the recycling route, increasing the difficulty of safe and reliable recycling of the drone.
Aerodynamic model of large wingspan drone under sudden air interference was constructed, and an extended Kalman filter was constructed based on this model, and the landing trajectory of the drone was predicted using real-time observation data.
By predicting the position coordinates and speed of the next trajectory, the safety and speed of the drone recycling of drones are improved, and the difficulty of recycling is reduced.
Smart Images

Figure CN117829016B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of unmanned aerial vehicle (UAV) control, and in particular relates to a method and device for predicting the landing trajectory of a large-wingspan UAV. Background Art
[0002] Compared to low-orbit satellites and high-altitude airships with similar mission characteristics, large-wingspan UAVs (UAVs) offer comprehensive advantages, including high mission altitude, long loiter time, strong maneuverability, and autonomy. These advantages make them well-suited for the information-based and autonomous nature of future warfare. They will play a vital role in space offense and defense, as well as information warfare, and represent a key development direction for UAVs. Large-wingspan UAVs feature a high lift-to-drag ratio, low flight speed, and wings with a large aspect ratio and low wing loading. Their design specifications are deeply coupled with energy parameters. This is particularly true for large-wingspan UAVs operating in near-space, where their weak payload capacity presents significant challenges in overall, aerodynamic, structural, flight control, power, propulsion, and trajectory design. UAV recovery technology, a crucial component of UAV combat units, is a key technology in the development of UAV weaponry. Ensuring precise recovery of large-wingspan UAVs within effective range is crucial for reducing the payload of large-wingspan UAVs and increasing their payload.
[0003] When encountering gusts during landing, large-wing fixed-wing drones inevitably experience wobbling. This makes flight control adjustments difficult, causing the recovery path to deviate and complicating safe and reliable recovery. To improve the real-time and accuracy of drone recovery, this paper proposes a method for predicting the landing trajectory of large-wing fixed-wing drones in the presence of gusts. Summary of the Invention
[0004] In order to solve the above problems existing in the prior art, the present invention provides a method and device for predicting the landing trajectory of a large-wingspan UAV.
[0005] In order to achieve the above objectives, the present invention adopts the following technical solutions.
[0006] In a first aspect, the present invention provides a method for predicting the landing trajectory of a large-wingspan UAV, comprising the following steps:
[0007] Construct an aerodynamic model for large-wingspan UAVs under gust disturbance;
[0008] constructing a Kalman filter based on the aerodynamic model;
[0009] The real-time observation data of the large-wingspan UAV landing is input into the Kalman filter to predict the landing trajectory of the large-wingspan UAV.
[0010] Furthermore, constructing an aerodynamic model for a large-wingspan UAV under gust interference includes:
[0011] Approximate parameters relevant to modeling;
[0012] List the dynamic and kinematic equations for a large wingspan UAV;
[0013] By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained:
[0014]
[0015] In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the roll static stability derivative, L p is the lateral damping derivative, L r is the transverse cross derivative, N β is the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the rudder control efficiency derivative; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed;
[0016] The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain:
[0017]
[0018] Where Δβ g is the reference disturbance of the sideslip angle without wind interference, V0 is the airspeed of the UAV in equilibrium state, v w is the crosswind speed;
[0019] Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance:
[0020]
[0021] Furthermore, the approximate processing of parameters related to modeling includes the following settings:
[0022] The large wingspan drone is a rigid body with a constant mass;
[0023] The geographic coordinate system is an inertial coordinate system;
[0024] Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV has not only a symmetrical geometric shape but also a symmetrical internal mass distribution, i.e., the inertial product I xy =I yz =0;
[0025] The surface of the Earth is horizontal;
[0026] The acceleration due to gravity does not change with flight altitude.
[0027] Furthermore, the Kalman filter is an extended Kalman filter.
[0028] Furthermore, constructing a Kalman filter based on the aerodynamic model includes:
[0029] List the state equations and observation equations of a discrete nonlinear system:
[0030] X(k+1)=f[k,X(k)]+G(k)W1(k) (4)
[0031] Z(k)=h[k,X(k)]+W2(k) (5)
[0032] Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, W1(k) and W2(k) are noise matrices; X(k) and Z(k) are:
[0033] X(k)=[x(k),y(k),z(k),v x (k),v y (k),v z (k),a x (k),a y (k),a z (k)] T (6)
[0034] Z(k)=[d0(k),α0(k),θ0(k)] T (7)
[0035] In the formula, x(k), y(k), z(k), v x (k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d0(k), α0(k), and θ0(k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station, respectively;
[0036] f[k,X(k)] is determined by formula (3), and the expression of h[k,X(k)] is:
[0037]
[0038] Where (x0, y0, z0) is the coordinate of the ground observation station;
[0039] Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation:
[0040]
[0041] Z(k)=H(k)X(k)+Y(k)+W2(k) (10)
[0042] Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion;
[0043] The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
[0044] In a second aspect, the present invention provides a landing trajectory prediction device for a large-wingspan UAV, comprising:
[0045] Model building module, used to build aerodynamic models of large-wingspan UAVs under gust disturbance;
[0046] A filter building module, configured to build a Kalman filter based on the aerodynamic model;
[0047] The trajectory prediction module is used to input the real-time observation data of the large-wingspan UAV during landing into the Kalman filter to predict the landing trajectory of the large-wingspan UAV.
[0048] Furthermore, the model building module is specifically used to:
[0049] Approximate parameters relevant to modeling;
[0050] List the dynamic and kinematic equations for a large wingspan UAV;
[0051] By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained:
[0052]
[0053] In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the roll static stability derivative, L p is the lateral damping derivative, L r is the transverse cross derivative, N β is the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the rudder control efficiency derivative; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed;
[0054] The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain:
[0055]
[0056] Where Δβ g is the reference disturbance of the sideslip angle without wind interference, V0 is the airspeed of the UAV in equilibrium state, v w is the crosswind speed;
[0057] Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance:
[0058]
[0059] Furthermore, the approximate processing of parameters related to modeling includes the following settings:
[0060] The large wingspan drone is a rigid body with a constant mass;
[0061] The geographic coordinate system is an inertial coordinate system;
[0062] Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV has not only a symmetrical geometric shape but also a symmetrical internal mass distribution, i.e., the inertial product I xy =I yz =0;
[0063] The surface of the Earth is horizontal;
[0064] The acceleration due to gravity does not change with flight altitude.
[0065] Furthermore, the Kalman filter is an extended Kalman filter.
[0066] Furthermore, the filter building module is specifically used to:
[0067] List the state equations and observation equations of a discrete nonlinear system:
[0068] X(k+1)=f[k,X(k)]+G(k)W1(k) (4)
[0069] Z(k)=h[k,X(k)]+W2(k) (5)
[0070] Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, W1(k) and W2(k) are noise matrices; X(k) and Z(k) are:
[0071] X(k)=[x(k),y(k),z(k),v x (k),v y (k),v z (k),a x (k),a y (k),a z (k)] T (6)
[0072] Z(k)=[d0(k),α0(k),θ0(k)] T (7)
[0073] In the formula, x(k), y(k), z(k), v x(k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d0(k), α0(k), and θ0(k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station, respectively;
[0074] f[k,X(k)] is determined by formula (3), and the expression of h[k,X(k)] is:
[0075]
[0076] Where (x0, y0, z0) is the coordinate of the ground observation station;
[0077] Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation:
[0078]
[0079] Z(k)=H(k)X(k)+Y(k)+W2(k) (10)
[0080] Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion;
[0081] The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
[0082] Compared with the prior art, the present invention has the following beneficial effects.
[0083] This invention constructs an aerodynamic model for large-wingspan drones subjected to gusts of wind, builds a Kalman filter based on this model, and inputs real-time observation data from large-wingspan drones during landing into the Kalman filter. This method can predict the landing trajectory of large-wingspan drones. By constructing an aerodynamic model for large-wingspan drones subjected to gusts of wind and using the Kalman filter, the present invention can quickly predict the position coordinates and velocity of the next trajectory based on historical trajectories, thereby improving the safety and speed of drone recovery by unmanned vehicles. BRIEF DESCRIPTION OF THE DRAWINGS
[0084] Figure 1 This is a flowchart of a method for predicting the landing trajectory of a large-wingspan UAV according to an embodiment of the present invention.
[0085] Figure 2Schematic diagram of the body coordinate system and the inertial coordinate system.
[0086] Figure 3 Schematic diagram of the airflow coordinate system and the body coordinate system.
[0087] Figure 4 Schematic diagram of the speed triangle in crosswind.
[0088] Figure 5 This is a block diagram of a landing trajectory prediction device for a large-wingspan UAV according to an embodiment of the present invention. DETAILED DESCRIPTION
[0089] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the present invention is further described below with reference to the accompanying drawings and specific embodiments. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0090] Figure 1 This is a flow chart of a method for predicting the landing trajectory of a large-wingspan UAV according to an embodiment of the present invention, comprising the following steps:
[0091] Step 101: construct an aerodynamic model of a large-wingspan UAV under gust interference;
[0092] Step 102: constructing a Kalman filter based on the aerodynamic model;
[0093] Step 103: input the real-time observation data of the large-wingspan UAV during landing into a Kalman filter to predict the landing trajectory of the large-wingspan UAV.
[0094] In this embodiment, step 101 is primarily used to construct an aerodynamic model for a large-wingspan UAV subjected to gust disturbances. Due to the large wingspan of a large-wingspan UAV, its response to gust disturbances during landing is significant, significantly impacting its flight trajectory. Therefore, it is necessary to establish an aircraft dynamics model under gusts. This aircraft dynamics model under gusts primarily considers a linearized model of the UAV's small perturbations in the presence of crosswinds, ignoring the impact of wind gradients on the UAV's torque and aerodynamic derivatives. It only examines the dynamics of the aircraft under crosswinds.
[0095] In this embodiment, step 102 is mainly used to construct a Kalman filter. In this embodiment, the landing trajectory of a large-wingspan UAV is predicted by performing Kalman filtering on the observation data during landing of the large-wingspan UAV, so it is necessary to first establish a Kalman filter. The target of Kalman filtering processing is a random signal. The observed quantity of the system is used as the input of the filtering algorithm, and the required signal estimation value is used as the output of the filtering algorithm. These two quantities are closely related through time update and observation update algorithm. The statistical characteristics of observation noise and system noise are utilized, and the required signal is estimated based on the state equation and observation equation of the system. The Kalman filter algorithm is a real-time recursive filtering algorithm implemented by a computer, which belongs to the optimal estimation method.
[0096] In this embodiment, step 103 primarily predicts the landing trajectory of a large-wingspan drone. This embodiment implements the prediction of the large-wingspan drone's landing trajectory by inputting real-time observation data, such as the drone's position coordinates and velocity, into a Kalman filter. This predicted landing trajectory can be used for recovery and docking between the drone and the unmanned vehicle. The predicted point is used as a local preview point, and the heading angle between this preview point and the vehicle's current attitude is calculated for pure tracking control.
[0097] As an optional embodiment, constructing an aerodynamic model of a large-wingspan UAV under gust interference includes:
[0098] Approximate parameters relevant to modeling;
[0099] List the dynamic and kinematic equations for a large wingspan UAV;
[0100] By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained:
[0101]
[0102] In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the roll static stability derivative, L p is the lateral damping derivative, L r is the transverse cross derivative, N βis the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the rudder control efficiency derivative; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed;
[0103] The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain:
[0104]
[0105] Where Δβ g is the reference disturbance of the sideslip angle without wind interference, V0 is the airspeed of the UAV in equilibrium state, v w is the crosswind speed;
[0106] Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance:
[0107]
[0108] This embodiment provides a technical solution for constructing an aerodynamic model of a large-wingspan UAV under gust interference. In order to simplify the dynamic model of the large-wingspan UAV, the parameters of the large-wingspan UAV are first approximated. For example, it is assumed that the large-wingspan UAV is a rigid body with a constant mass. Then, the dynamic and kinematic equations of the large-wingspan UAV are obtained based on the coordinate system transformation. This embodiment involves an inertial coordinate system, an aircraft body coordinate system, and an airflow coordinate system. Inertial coordinate system Ox g y g z g Fixed on the surface of the earth, with the origin located at an arbitrary fixed point on the ground, and the three coordinate axes pointing to the north (N), east (E), and ground (D). b y b z b It is fixed to the drone and moves with it. The origin is at the center of mass of the drone. b 、z b The axis is located in the symmetry plane of the drone, xb is parallel to the fuselage axis and points forward, z b Axis perpendicular to x b Axis, pointing down, y b The axis is perpendicular to the symmetry plane and points to the right. The schematic diagram of the body coordinate system and the inertial coordinate system is as follows Figure 2As shown. Airflow coordinate system Ox w y w z w The origin is located at the center of mass of the drone, x w The axis always coincides with the airspeed direction of the drone, z w The axis is located in the symmetry plane of the drone and is perpendicular to the x w Axis, pointing down, y w Perpendicular to Ox w z w Plane, pointing to the right. The schematic diagram of the airflow coordinate system and the body coordinate system is as follows Figure 3 The longitudinal, lateral, and directional motions of the drone are all defined in the body coordinate system. The three components of the aerodynamic torque (rolling moment L, pitching moment M, and yaw moment N) are also defined relative to the three axes of the body coordinate system. Their relative relationships are as follows:
[0109]
[0110]
[0111] Where, F x 、F y 、F z is the component of the net external force along the three axes of the body, p, r, q are the angular velocities of the body coordinate system relative to the geographic coordinate system, u, v, w are the linear velocities of the body coordinate system relative to the geographic coordinate system, L, M, N are the projections of the external torque on the body axis, I x , I y , I z are the inertia moments of the drone about the three axes of the body, I xy , I yz , I zx are the three inertial products of the drone to the body. The attitude angle of the drone is determined by the relative relationship between the body coordinate system and the inertial coordinate system. The pitch angle θ is the body axis x b and ground plane Ox g y g The yaw angle Ψ is the angle of the body axis x b At ground level Ox g y g The projection of x g The angle between the axes; the roll angle φ is z g Axis containing x b The angle between the axis and the vertical plane is converted as follows:
[0112]
[0113]
[0114] The transformation relationship between the airflow coordinate system and the body coordinate system is:
[0115]
[0116] Where, the angle of attack α is the difference between the projection of the UAV velocity vector on the UAV symmetry plane and x b The sideslip angle β is the angle between the UAV velocity vector and the UAV symmetry plane.
[0117] Finally, the aerodynamic model of the large-wingspan UAV under gust interference is constructed. The dynamic equations and kinematic equations of the large-wingspan UAV are shown in the coordinate transformation formula above. The lateral and longitudinal coupled motions are decoupled and simplified under typical flight conditions. For the lateral motion analysis, the small perturbation linearization method is used to obtain the lateral linearization matrix equations, as shown in formula (1). In formula (1), Y β 、Y p 、Y r , L β , L p , L r 、N β 、N p 、N r and The derivatives are the lateral dynamic derivatives and the control derivatives. The derivative here is a figurative name similar to the derivative in mathematics. In mathematics, the differential of a variable can be expressed as the product of the change of the variable and the derivative of the variable. The left side of the equal sign in formula (1) is the differential of the disturbance, so the right side of the equal sign is regarded as the product of the disturbance and the derivative parameter.
[0118] When a UAV encounters crosswind during flight, the velocity triangle on the horizontal plane is as follows: Figure 4 As shown. Crosswinds primarily affect the sideslip angle of the drone. The windless state is considered the drone's equilibrium or undisturbed state. The drone's response to wind can be considered as a small perturbation motion in which the drone deviates from the equilibrium state after the wind acts. Therefore, small perturbation linearization can be performed to obtain Equation (2). Substituting Equation (2) into Equation (1) yields the aerodynamic model under gust disturbance, as shown in Equation (3).
[0119] As an optional embodiment, the approximate processing of parameters related to modeling includes the following settings:
[0120] The large wingspan drone is a rigid body with a constant mass;
[0121] The geographic coordinate system is an inertial coordinate system;
[0122] Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV has not only a symmetrical geometric shape but also a symmetrical internal mass distribution, i.e., the inertial product I xy =I yz =0;
[0123] The surface of the Earth is horizontal;
[0124] The acceleration due to gravity does not change with flight altitude.
[0125] This embodiment provides a method for approximating parameters related to modeling. In order to simplify the model, this embodiment performs five approximation processes on the parameters related to modeling. The specific contents are as described above and will not be described in detail here.
[0126] As an optional embodiment, the Kalman filter is an extended Kalman filter.
[0127] To improve filtering accuracy, this embodiment employs an extended Kalman filter. The standard Kalman filter, based on linear algebra and hidden Markov models, is only suitable for linear systems. However, in real-world engineering practice, systems often exhibit varying degrees of nonlinearity. Only a very small number of systems can be approximated as linear systems. The vast majority cannot be simply treated through approximate linearization, or are difficult to quantitatively describe using linear difference equations or thread differential equations. The flight state of a large-wingspan UAV is a nonlinear system. Therefore, this embodiment employs the extended Kalman filter, which can handle nonlinear systems. The core concept of the extended Kalman filter is that, for general nonlinear systems, the nonlinear function is first expanded into a Taylor series around the filter value, omitting second-order and higher terms to obtain an approximate linearized model. The Kalman filter is then applied to estimate the target. Simulation results show that the extended Kalman filter achieves higher prediction accuracy than the standard Kalman filter in predicting the low-speed landing trajectory of a large-wingspan UAV under gust interference. It also offers excellent real-time performance, demonstrating its practicality.
[0128] As an optional embodiment, constructing a Kalman filter based on the aerodynamic model includes:
[0129] List the state equations and observation equations of a discrete nonlinear system:
[0130] X(k+1)=f[k,X(k)]+G(k)W1(k) (4)
[0131] Z(k)=h[k,X(k)]+W2(k) (5)
[0132] Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, W1(k) and W2(k) are noise matrices; X(k) and Z(k) are:
[0133] X(k)=[x(k),y(k),z(k),vx (k),v y (k),v z (k),a x (k),a y (k),a z (k)] T (6)
[0134] Z(k)=[d0(k),α0(k),θ0(k)] T (7)
[0135] In the formula, x(k), y(k), z(k), v x (k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d0(k), α0(k), and θ0(k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station, respectively;
[0136] f[k,X(k)] is determined by formula (3), and the expression of h[k,X(k)] is:
[0137]
[0138] Where (x0, y0, z0) is the coordinate of the ground observation station;
[0139] Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation:
[0140]
[0141] Z(k)=H(k)X(k)+Y(k)+W2(k) (10)
[0142] Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion;
[0143] The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
[0144] This embodiment provides a specific technical solution for implementing the extended Kalman filter. First, the discrete nonlinear system state equation and observation equation are given, such as equations (4) and (5), where f[k, X(k)] and h[k, X(k)] are nonlinear functions. Then, the expressions of the drone state vector X(k) and the observation vector Z(k) are listed, such as equations (5) and (6), where X(k) is a 9-dimensional vector consisting of the position coordinates, velocity, and acceleration of the three components x, y, and z, and Z(k) is a 3-dimensional vector consisting of the distance, azimuth, and pitch angle of the drone relative to the ground observation station. f[k, X(k)] is determined by the drone aerodynamic model shown in equation (3), and the expression of h[k, X(k)] is shown in equation (8). The nonlinear functions f[k,X(k)] and h[k,X(k)] are Taylor expanded with respect to X(k), that is, only the first-order term and the constant term are retained and the higher-order terms above the second order are ignored, so as to obtain an approximate simple state equation and observation equation, such as (9) and (10), where the matrices Φ(k+1|k) and H(k) and Y(k) are determined by the first-order Taylor expansion, where Φ(k+1|k) and The expression is:
[0145]
[0146]
[0147] Where Δt is the interval time of state transition, V is the airspeed of the UAV in the OXY plane, and r(k) is determined by formula (3).
[0148] Finally, the gain matrix and error matrix are calculated according to the standard Kalman filter algorithm to predict the position coordinates and velocity at the k+1th moment, thereby realizing the prediction of the UAV landing trajectory.
[0149] Figure 5 This is a schematic diagram of the composition of a large-wingspan UAV landing trajectory prediction device according to an embodiment of the present invention, the device comprising:
[0150] Model building module 11, used to build an aerodynamic model of a large-wingspan UAV under gust interference;
[0151] A filter construction module 12, configured to construct a Kalman filter based on the aerodynamic model;
[0152] The trajectory prediction module 13 is used to input the real-time observation data of the large-wingspan UAV during landing into the Kalman filter to predict the landing trajectory of the large-wingspan UAV.
[0153] The device of this embodiment can be used to perform Figure 1The technical solution of the method embodiment shown has similar implementation principles and technical effects, which will not be described in detail here. The same is true for the following embodiments, which will not be described in detail.
[0154] As an optional embodiment, the model building module 11 is specifically configured to:
[0155] Approximate parameters relevant to modeling;
[0156] List the dynamic and kinematic equations for a large wingspan UAV;
[0157] By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained:
[0158]
[0159] In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the roll static stability derivative, L p is the lateral damping derivative, L r is the transverse cross derivative, N β is the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the rudder control efficiency derivative; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed;
[0160] The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain:
[0161]
[0162] Where Δβ gis the reference disturbance of the sideslip angle without wind interference, V0 is the airspeed of the UAV in equilibrium state, v w is the crosswind speed;
[0163] Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance:
[0164]
[0165] As an optional embodiment, the approximate processing of parameters related to modeling includes the following settings:
[0166] The large wingspan drone is a rigid body with a constant mass;
[0167] The geographic coordinate system is an inertial coordinate system;
[0168] Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV has not only a symmetrical geometric shape but also a symmetrical internal mass distribution, i.e., the inertial product I xy =I yz =0;
[0169] The surface of the Earth is horizontal;
[0170] The acceleration due to gravity does not change with flight altitude.
[0171] As an optional embodiment, the Kalman filter is an extended Kalman filter.
[0172] As an optional embodiment, the filter construction module 12 is specifically configured to:
[0173] List the state equations and observation equations of a discrete nonlinear system:
[0174] X(k+1)=f[k,X(k)]+G(k)W1(k) (4)
[0175] Z(k)=h[k,X(k)]+W2(k) (5)
[0176] Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, W1(k) and W2(k) are noise matrices; X(k) and Z(k) are:
[0177] X(k)=[x(k),y(k),z(k),v x (k),v y (k),v z (k),a x (k),ay (k),a z (k)] T (6)
[0178] Z(k)=[d0(k),α0(k),θ0(k)] T (7)
[0179] In the formula, x(k), y(k), z(k), v x (k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d0(k), α0(k), and θ0(k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station, respectively;
[0180] f[k,X(k)] is determined by formula (3), and the expression of h[k,X(k)] is:
[0181]
[0182] Where (x0, y0, z0) is the coordinate of the ground observation station;
[0183] Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation:
[0184]
[0185] Z(k)=H(k)X(k)+Y(k)+W2(k) (10)
[0186] Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion;
[0187] The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
[0188] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.
Claims
1. A method for predicting the landing trajectory of a large-wingspan UAV. It is characterized in that The following steps are involved: Construct an aerodynamic model for large-wingspan UAVs under gust disturbance; Constructing a Kalman filter based on the aerodynamic model; The real-time observation data of the large-wingspan UAV landing is input into the Kalman filter to predict the landing trajectory of the large-wingspan UAV; The aerodynamic model of the large-wingspan UAV under gust disturbance is constructed as follows: Approximate parameters relevant to modeling; List the dynamic and kinematic equations for a large wingspan UAV; By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained: In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the rolling static stability derivative, L p is the lateral damping derivative, L r is the lateral cross derivative, N β is the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the derivative of the rudder control efficiency; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed; The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain: In the formula, Δβ g is the reference disturbance of the sideslip angle without wind disturbance, V 0 is the airspeed of the drone in equilibrium state, v w is the crosswind speed; Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance: The Kalman filter is an extended Kalman filter; The constructing of a Kalman filter based on the aerodynamic model comprises: List the state equations and observation equations of a discrete nonlinear system: X(k+1)=f[k,X(k)]+G(k)W 1 (k) (4) Z(k)=h[k,X(k)]+W 2 (k) (5) Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, and W is 1 (k), W 2 (k) is the noise matrix; where X(k) and Z(k) are: X(k)=[x(k),y(k),z(k),v x (k),v y (k),v z (k),a x (k),a y (k),a z (k)] T (6) Z(k)=[d 0 (k),α 0 (k),θ 0 (k)] T (7) In the formula, x(k), y(k), z(k), v x (k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d 0 (k), α 0 (k),θ 0 (k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station; f[k,X(k)] is determined by equation (3), and the expression of h[k,X(k)] is: In the formula, (x 0 ,y 0 ,z 0 ) are the coordinates of the ground observation station; Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation: Z(k)=H(k)X(k)+Y(k)+W 2 (k) (10) Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion; The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
2. The method for predicting the landing trajectory of a large-wingspan UAV according to claim 1, It is characterized in that The approximate processing of parameters related to modeling includes the following settings: The large wingspan drone is a rigid body and its mass is constant; The geographic coordinate system is an inertial coordinate system; Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV not only has a symmetrical geometric shape, but also has a symmetrical internal mass distribution, that is, the inertia product I xy =I yz =0; The surface of the earth is horizontal; The acceleration due to gravity does not change with flight altitude.
3. A landing trajectory prediction device for large-wingspan UAVs, It is characterized in that include: Model building module, used to build aerodynamic model of large-wingspan UAV under gust disturbance; A filter building module, used for building a Kalman filter based on the aerodynamic model; The trajectory prediction module is used to input the real-time observation data of the large-wingspan UAV during landing into the Kalman filter to predict the landing trajectory of the large-wingspan UAV; The aerodynamic model of the large-wingspan UAV under gust disturbance is constructed as follows: Approximate parameters relevant to modeling; List the dynamic and kinematic equations for a large wingspan UAV; By decoupling and simplifying the lateral and longitudinal coupled motion under typical flight conditions and using the small disturbance linearization method, the lateral linearization matrix equation is obtained: In the formula, the state variables Δβ, Δp, Δr, They are the lateral sideslip angle β, roll rate p, yaw rate r and roll angle The amount of disturbance from the equilibrium state; Δδ a , Δδ r are the deflection angle disturbances of the aileron and rudder respectively; Y β is the lateral force derivative, Y p is the lateral force derivative of the rolling angular velocity, Y r is the lateral force derivative of the yaw angular velocity, L β is the rolling static stability derivative, L p is the lateral damping derivative, L r is the lateral cross derivative, N β is the heading static stability derivative, N p is the heading cross derivative, N r is the heading damping derivative, is the rudder side force derivative, is the aileron control effectiveness derivative, is the rudder derivative of the rolling moment coefficient, is the derivative of the yaw moment coefficient with respect to the aileron, is the derivative of the rudder control efficiency; α is the UAV angle of attack, θ is the UAV pitch angle, and V is the UAV airspeed; The windless state is regarded as the equilibrium state or undisturbed state of the UAV. When there is wind, the UAV's response to the wind is regarded as a disturbance motion that deviates from the equilibrium state. The disturbance is linearized to obtain: In the formula, Δβ g is the reference disturbance of the sideslip angle without wind disturbance, V 0 is the airspeed of the drone in equilibrium state, v w is the crosswind speed; Substituting equation (2) into equation (1), we can obtain the aerodynamic model under gust disturbance: The Kalman filter is an extended Kalman filter; The constructing of a Kalman filter based on the aerodynamic model comprises: List the state equations and observation equations of a discrete nonlinear system: X(k+1)=f[k,X(k)]+G(k)W 1 (k) (4) Z(k)=h[k,X(k)]+W 2 (k) (5) Where k is the current time, f[k,X(k)] and h[k,X(k)] are nonlinear functions, X(k) is the state vector, Z(k) is the observation vector, G(k) is the noise input matrix, and W is 1 (k), W 2 (k) is the noise matrix; where X(k) and Z(k) are: X(k)=[x(k),y(k),z(k),v x (k),v y (k),v z (k),a x (k),a y (k),a z (k)] T (6) Z(k)=[d 0 (k),α 0 (k),θ 0 (k)] T (7) In the formula, x(k), y(k), z(k), v x (k), v y (k), v z (k), a x (k), a y (k), a z (k) is the coordinate, velocity and acceleration of the UAV in the inertial coordinate system, d 0 (k), α 0 (k),θ 0 (k) are the distance, azimuth and pitch angle of the UAV relative to the ground observation station; f[k,X(k)] is determined by equation (3), and the expression of h[k,X(k)] is: In the formula, (x 0 ,y 0 ,z 0 ) are the coordinates of the ground observation station; Perform the first-order Taylor expansion of the nonlinear functions f[k,X(k)] and h[k,X(k)] on X(k), and ignore the second-order and higher terms to obtain the simplified state equation and observation equation: Z(k)=H(k)X(k)+Y(k)+W 2 (k) (10) Where, Φ(k+1|k), H(k) and Y(k) are determined by first-order Taylor expansion; The standard Kalman filter algorithm can be used to predict the position coordinates and speed of the UAV at the k+1th moment, and realize the prediction of the UAV landing trajectory.
4. The landing trajectory prediction device for a large-wingspan UAV according to claim 3, It is characterized in that The approximate processing of parameters related to modeling includes the following settings: The large wingspan drone is a rigid body and its mass is constant; The geographic coordinate system is an inertial coordinate system; Plane ox of the body coordinate system b z b The symmetry plane of the large wingspan UAV is that the UAV not only has a symmetrical geometric shape, but also has a symmetrical internal mass distribution, that is, the inertia product I xy =I yz =0; The surface of the earth is horizontal; The acceleration due to gravity does not change with flight altitude.
