Transition mode comprehensive weight intelligent control method for compound wing unmanned aerial vehicle

Abstract

The application discloses a transition mode comprehensive weight intelligent control method for a compound wing unmanned aerial vehicle, and belongs to the technical field of unmanned aerial vehicle flight control and intelligent control. In view of the problems of lift switching, nonlinear coupling, control chattering and stall risk in the transition from a rotor mode to a fixed wing mode, a dual-variable comprehensive weight distribution mechanism with airspeed dominance and angle of attack correction is adopted to construct rotor / fixed wing control weights which are continuous, smooth and meet the normalization constraint; meanwhile, the unmanned aerial vehicle is modeled as an integral nonlinear system, and strong robustness, low chattering and high precision integral control input solutions are realized through integral sliding mode control, neural network disturbance compensation and Moore-Penrose pseudo-inverse control distribution; finally, the control commands are distributed to the rotor, rudder and propulsion system according to the comprehensive weight, and the control dominance is smoothly transferred. The height and attitude stability, anti-interference ability and flight safety margin in the transition stage are improved, the engineering realizability is strong, and the method is suitable for mode transition control of various compound wing unmanned aerial vehicles.

Description

Technical Field

[0001] This invention belongs to the field of UAV flight control and intelligent control technology, specifically relating to an intelligent control method for the transition mode integrated weight of a compound wing UAV. It is applicable to compound wing UAVs with rotor vertical take-off and landing capabilities and fixed-wing cruise capabilities, and can be used to achieve coordinated distribution of rotor lift and wing aerodynamic lift during the transition flight phase, as well as stable control of key longitudinal flight states such as altitude, pitch attitude and airspeed. Background Technology

[0002] Compound-wing UAVs, by integrating rotor and fixed-wing flight modes, offer significant advantages in tasks such as vertical takeoff and landing, hovering, and high-speed cruise. However, during the transition from rotor to fixed-wing mode, the aircraft's lift source, aerodynamic layout, and control effects undergo significant changes. Rotor lift gradually decreases while wing aerodynamic lift gradually increases. The system is simultaneously subjected to multiple forces, including rotor thrust, wing lift, and propulsion system thrust, resulting in highly nonlinear, strongly coupled, and time-varying dynamic characteristics. This can lead to problems such as altitude fluctuations, pitch instability, and control command chattering, thereby affecting the safety and smoothness of the transition flight.

[0003] Existing transition control methods for compound-wing UAVs mostly employ airspeed-based mode switching or weighted scheduling strategies. This involves setting a speed threshold or predefined linear or piecewise functions to achieve the switching or fusion of rotor control and fixed-wing control. However, these methods typically use airspeed as the sole scheduling variable, failing to adequately consider the impact of angle-of-attack variations on lift distribution, attitude safety margin, and control requirements. The weighted design is also limited in scope, making it difficult to adapt to complex transitional flight conditions. Furthermore, fixed weighted functions lack adaptive adjustment capabilities during flight. When aerodynamic parameters, load conditions, or external disturbances change, control performance cannot remain consistent, posing control safety risks at high angles of attack.

[0004] On the other hand, it is difficult to establish accurate dynamic models for compound-wing UAVs in transition modes, and traditional control methods that rely on accurate models have limited engineering adaptability. Although sliding mode control has strong robustness, it is prone to chattering problems if used alone; although neural networks have nonlinear approximation capabilities, if directly used as controller outputs, system stability is difficult to analyze and guarantee, and cannot meet the high reliability requirements of UAV flight control. Therefore, how to introduce intelligent methods to improve the adaptability and smoothness of transition mode control while ensuring system stability, and address issues such as incomplete weight scheduling, control chattering, and weak disturbance rejection capability, are problems that need to be solved in the field of compound-wing UAV control. Summary of the Invention

[0005] The purpose of this invention is to address the problems encountered by existing compound-wing unmanned aerial vehicles (UAVs) during the transition from rotor-dominated to fixed-wing flight modes. These problems arise because the lift source gradually shifts from rotor-dominated to wing-dominated, leading to significant changes in longitudinal dynamic characteristics, increased model uncertainty, and complex control channel coupling. This invention provides a comprehensive weighted intelligent control method for the transition mode of a compound-wing UAV. This method uses flight speed and angle of attack as the core scheduling variables in the transition mode. By constructing a comprehensive weighted allocation mechanism between the rotor control channel and the fixed-wing control channel, it achieves a continuous transfer of control dominance. Furthermore, to address the nonlinear strong coupling and disturbance uncertainty issues in the transition phase of the compound-wing UAV, it achieves rapid error convergence and robustness through sliding mode control, introduces the Moore-Penrose pseudo-inverse to ensure continuous and achievable multi-actuator control allocation, and uses neural networks to compensate for equivalent disturbances online, thereby improving the stability and safety margin of the transition process.

[0006] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0007] A method for intelligent control of transition modes of a compound-wing unmanned aerial vehicle (UAV) with integrated weights includes the following steps:

[0008] S1: Collect flight status information of the compound wing UAV and determine whether it has entered the transition phase from rotor mode to fixed wing mode based on flight speed;

[0009] S2: After determining that the transition phase has begun, a basic weight value is constructed based on the flight speed, and the weight value is corrected in combination with the angle of attack to form a combined control weight value for rotor / fixed wing.

[0010] S3: In the transition mode, the compound wing UAV is modeled as an overall nonlinear coupled system, a unified error vector and vector sliding surface are constructed, and online compensation of the unmodeled dynamics is performed by combining neural networks to solve the overall robust control input;

[0011] S4: Based on the rotor weight and fixed wing weight, the control commands are weighted and distributed, and the weighted control commands are output to the rotor system, fixed wing control surfaces and propulsion system respectively, so as to realize the continuous and smooth transfer of control dominance.

[0012] Step S1 specifically includes the following steps:

[0013] S11: Real-time acquisition of flight speed V and angle of attack via onboard sensors Altitude h, Pitch Vertical velocity And at least a portion of the pitch angular velocity q, after low-pass filtering to eliminate sensor noise, the collected state variables are used to construct the flight state vector. S12: Pre-set the speed range [V1,V2] for the transition from rotor flight mode to fixed-wing flight mode of the compound wing UAV, where V1 represents the initial speed threshold for entering the transition phase and V2 represents the speed threshold for ending the transition phase; V1 and V2 are set according to the UAV type, wing parameters, and rotor thrust characteristics. When the flight speed satisfies V1≤V≤V2, it is determined that the compound wing UAV has entered the transition flight phase.

[0014] S13: When the flight speed is within the transition speed range [V1,V2], the integrated weight scheduling and overall robust control strategy for the transition mode is triggered; when the flight speed is lower than V1 or higher than V2, the rotor mode control strategy or the fixed wing mode control strategy is adopted respectively to achieve seamless connection of the three control strategies and avoid sudden changes in control commands caused by mode switching.

[0015] Step S2 specifically includes the following steps:

[0016] S21: Regarding flight speed V and angle of attack Normalization was performed separately to obtain normalized velocity index and normalized angle of attack index, so as to eliminate the influence of the difference in the dimensions of different physical quantities on the weight construction process and improve the universality and engineering applicability of the weight function design;

[0017] S22: A velocity weight function is constructed based on the normalized flight speed to describe the process of the rotor control dominance gradually shifting to the fixed-wing control dominance as the flight speed increases; the velocity weight function changes continuously in the interval [V1,V2] and maintains first-order continuity at the boundary of the interval to avoid control command chattering caused by abrupt changes in weight.

[0018] when At this time, the rotor weight is 1, and the rotor is completely dominant and controlled;

[0019] when At that time, the rotor weight is 0, and the fixed wing completely dominates control;

[0020] When the weights change smoothly in the intermediate interval V1≤V≤V2;

[0021] S23: Construct an angle-of-attack correction weight function based on the angle of attack magnitude to reflect the aerodynamic safety margin of the fixed wing under the current flight attitude; when the angle of attack is small ( No correction is made to the speed weights; when the angle of attack gradually approaches the stall risk threshold ( This increases the rotor control weight to slow the transfer rate of control to the fixed wing, thereby preventing the fixed wing from bearing excessive lift load under high angle-of-attack conditions. When the angle of attack reaches or exceeds the risk threshold ( When ), the rotor weight is increased to the maximum extent to completely avoid the risk of fixed-wing stall.

[0022] S24: The velocity weights and angle-of-attack correction weights are fused, and the weights are constrained to a certain value using a saturation function. Within the interval, the control weights of the rotor and the fixed wing are finally obtained, and the two always satisfy the normalization constraint relationship;

[0023] By employing the above methods, the control authority of the rotor and fixed wing changes continuously throughout the transition phase, avoiding drastic fluctuations in altitude, attitude, and control commands caused by sudden changes in weights.

[0024] Step S3 specifically includes the following steps:

[0025] S31: During the transition from rotor mode to fixed-wing mode, rotor thrust, propulsion system thrust, and fixed-wing aerodynamic lift and torque act together on the compound wing UAV, causing the system to exhibit significant nonlinear, strong coupling, and time-varying parameter characteristics.

[0026] Therefore, in the transition mode, the compound wing UAV is regarded as a nonlinear dynamic system. In its dynamic model, the rotor-wing interference, aerodynamic parameter changes, model simplification errors and external disturbances are uniformly represented as equivalent uncertain disturbance terms.

[0027] S32: Select key longitudinal variables such as altitude, pitch angle, and flight speed to form the control output vector for the transition phase, and construct a unified error vector with its expected trajectory. By sorting out the error and its derivative, the overall error dynamics model is obtained.

[0028] S33: To ensure rapid convergence of errors during the transition phase and system robustness, an integral vector sliding surface is constructed based on the error vector; the sliding variables are differentiated, and combined with the error dynamics model, a continuous arrival law is selected to replace the traditional discontinuous sign function, so that the system state reaches and remains near the sliding surface in a finite time, while effectively reducing control chattering and improving engineering feasibility;

[0029] S34: To address the model simplification errors, rotor-wing interference, and external disturbances present during the transition phase, a neural network estimator is introduced to estimate the equivalent disturbances in the error channel online. The neural network takes error-related quantities, flight speed, angle of attack, and angular velocity as inputs, adjusts the network weights online through an adaptive update law, and performs amplitude limiting on the network output to avoid abrupt changes in control commands caused by estimation anomalies.

[0030] S35: Considering that the control input dimension and error dimension are generally inconsistent during the transition phase, and the control allocation matrix is ​​not invertible, the Moore-Penrose pseudo-inverse is used to solve the overall control input in the sense of minimum norm, thereby ensuring the continuity, rationality and engineering feasibility of the control allocation, and obtaining the overall robust control law.

[0031] Step S4 specifically includes the following steps:

[0032] S41: Based on the rotor control weights and fixed-wing control weights constructed in step S2, the control commands output by the overall robust control module are weighted and allocated to obtain rotor system control commands and fixed-wing system control commands respectively.

[0033] S42: Control commands assigned to the rotor system are used to adjust rotor thrust or speed, control commands assigned to the fixed-wing control surfaces are used to adjust lift and pitch moment, and propulsion system control commands are output simultaneously to establish the airspeed required for fixed-wing cruise, thereby achieving a continuous and smooth transition in which rotor lift gradually withdraws and fixed-wing aerodynamic lift gradually takes over during the transition phase.

[0034] Compared with the prior art, the present invention has the following beneficial technical effects:

[0035] (1) To address the problem of discontinuous weight switching and altitude / attitude fluctuations caused by sudden command changes, this invention constructs a smooth comprehensive weight function that satisfies normalization constraints. The cosine smoothing design ensures that the weights are continuous and first-order continuous within the transition interval. At the same time, an angle-of-attack correction weight function is introduced to construct a bivariate adaptive weight scheduling mechanism that is dominated by airspeed and corrected by angle of attack. Under high angle-of-attack conditions, the rotor control weight is automatically increased to avoid the risk of fixed-wing stall and achieve a seamless and continuous transition between rotor and fixed-wing control commands, significantly improving the smoothness of transition flight and attitude safety margin.

[0036] (2) This invention addresses the problems of strong nonlinear coupling during the transition period, slow error convergence under disturbance, and control chattering. It adopts a collaborative design of integral vector sliding surface and continuous saturation arrival law, which effectively reduces chattering while ensuring system robustness, and improves the engineering feasibility and error convergence speed of the actuator. Furthermore, it combines the equivalent disturbance estimated online by neural network, and simplifies uncertainties such as error and rotor-wing interference through adaptive update law compensation model, thereby further improving the system's anti-disturbance capability and achieving high-precision stable control of the transition process.

[0037] (3) This invention addresses the problems of inconsistent control input and error dimensions and non-invertible control allocation matrix under multi-actuator collaboration. It introduces Moore-Penrose pseudo-inverse to solve continuous control allocation in the sense of minimum norm, and combines the equivalent disturbance compensation results of neural network to provide accurate control input for multi-actuator collaboration. Even under complex working conditions, it can still maintain stable control accuracy and system stability margin, effectively solving the core technical bottleneck of multi-actuator collaborative control allocation in the transition stage of compound wing UAV. Attached Figure Description

[0038] Figure 1 This is a block diagram of the overall control system structure for the transition mode of the compound-wing UAV of the present invention.

[0039] Figure 2 This is a flowchart of the transition mode control method for the compound-wing unmanned aerial vehicle of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the specific embodiments of the intelligent control method for transition mode comprehensive weights of a compound-wing unmanned aerial vehicle (UAV) according to this invention will be described in detail below with reference to the accompanying drawings. It should be understood that the following embodiments are only used to illustrate the technical solutions of this invention and are not intended to limit the scope of protection of this invention.

[0041] like Figure 1 As shown, the present invention provides an intelligent control method for the transition mode integrated weights of a compound-wing unmanned aerial vehicle, comprising the following steps:

[0042] S1. Collect flight status information of the compound wing UAV and determine whether the UAV has entered the transition phase from rotor flight mode to fixed wing flight mode based on airspeed.

[0043] S2. When the transition phase is determined based on the airspeed, a basic weight value that changes smoothly between 0 and 1 is first generated based on the airspeed, and the weight value is corrected in combination with the angle of attack to obtain the comprehensive weight value used to fuse the control commands of the rotor and fixed wing, namely the rotor control weight value and the fixed wing control weight value.

[0044] S3. In the transition mode, the compound-wing UAV is modeled as an overall nonlinear coupled system. An overall error vector is constructed based on the overall system output and the desired trajectory, and a vector sliding surface is constructed on the basis of the error vector. Under the constraints of the sliding surface, the overall robust control input is solved by combining the online estimation of the unmodeled dynamics by the neural network.

[0045] S4. Based on the rotor control weights and fixed-wing control weights obtained in step S2, the overall robust control input described in step S3 is weighted and allocated, and the weighted control commands are output to the rotor system, fixed-wing control surfaces and propulsion system respectively, so as to realize the continuous and smooth transfer of rotor control dominance to fixed-wing control dominance.

[0046] Step S1 specifically includes:

[0047] S11. Real-time acquisition of airspeed V and angle of attack via airborne sensors (including but not limited to airspeed sensor, angle of attack sensor, IMU inertial measurement unit, altimeter). Altitude h, pitch angle θ, vertical velocity At least a portion or a combination of the pitch angular velocity q, which constitute the state vector. As control inputs; all acquired state variables are filtered to eliminate the impact of sensor noise on control accuracy.

[0048] S12. Determine the transition mode according to the preset transition interval, the preset transition interval including the speed range. When satisfied The transition phase is determined at a certain time; among them, The initial velocity for the transition phase, The speed at which the transition phase ends;

[0049] S13. When the flight speed is within the transition range between the rotorcraft flight mode and the fixed-wing flight mode, the UAV is determined to have entered the transition flight phase, thereby triggering the comprehensive weight scheduling and overall robust control strategy for the transition mode; when When, a conventional rotor mode control strategy is adopted; when At the same time, a conventional control strategy for fixed-wing modes is adopted to achieve seamless integration of the three control strategies and avoid sudden changes in commands during mode switching.

[0050] Step S2 specifically includes:

[0051] S21, Regarding flight speed V and angle of attack Normalization was performed separately to obtain the normalized speed index. with normalized angle of attack index This is to eliminate the influence of differences in the dimensions of different physical quantities on the weight construction process and improve the versatility of weight function design;

[0052] The normalization formula is as follows:

[0053]

[0054]

[0055] S22, Based on the normalized speed index A velocity weighting function is constructed to describe the process of control shifting from the rotor system to the fixed-wing system as airspeed V increases. Its form is as follows:

[0056]

[0057] in, The initial velocity for the transition phase, The velocity at the end of the transition phase is π, where π is the mathematical constant pi. It is a cosine function; when hour, =0 indicates that the rotor system is fully under control; when hour, =1 indicates that the fixed-wing system has complete control; when in the intermediate range The velocity weight function is a smooth S-shaped curve, with the weights increasing smoothly with airspeed. exist The weights are continuous at the top and first-order continuous at the boundary to avoid control command chattering caused by abrupt changes in weights.

[0058] S23. Construct an angle-of-attack correction weight function based on the angle of attack magnitude to reflect the safety margin of fixed-wing aerodynamic operation under the current flight attitude. Its form is as follows:

[0059]

[0060] in, For the angle of attack of the compound-wing UAV, This is the absolute value of the angle of attack; For the angle of attack safety threshold, The angle of attack risk threshold, and It is generally determined by the stall angle of attack; when the absolute value of the angle of attack... hour, It does not correct the velocity weights; when hour, The angle of attack increases linearly with increasing absolute value, gradually increasing the rotor weight; when the absolute value of the angle of attack... hour, The core function of increasing the rotor weight to the maximum extent is to temporarily increase the "rotor weight" when the angle of attack is too large, slow down the transfer to the fixed wing, prevent the fixed wing from entering the stall risk range due to the angle of attack exceeding the limit, and improve the attitude safety margin of the transition flight.

[0061] S24. Combining the aforementioned velocity weights With angle of attack correction weight Construct rotor control weights Its form is:

[0062]

[0063] in, To adjust the coefficient for the degree of influence of angle of attack, This is a saturation function used to adjust the rotor control weights. The value of is restricted to [0,1], and the control weights of the fixed wing are further constructed. :

[0064]

[0065] Make rotor control weight With fixed-wing control weights It changes continuously throughout the entire transition phase and satisfies The normalization constraint is used to avoid sudden changes in control dominance.

[0066] The specific control logic is as follows: when the airspeed approaches the initial speed of the transition phase... hour, When the airspeed is 0, regardless of the angle of attack, the rotor dominates; when the airspeed approaches the end speed of the transition phase... If the angle of attack is less than the safe threshold, the rotor will disengage and the fixed wing will take over. When the airspeed is greater than the end speed of the transition phase, if the angle of attack is too large, the rotor can still retain some weight to help maintain altitude and avoid drastic changes in pitch / altitude due to the wing approaching stall.

[0067] Step S3 specifically includes:

[0068] S31. During the transition from rotor flight mode to fixed-wing flight mode of a compound-wing UAV, the rotor thrust, propulsion system thrust, and fixed-wing aerodynamic lift / torque act together on the aircraft, causing the system to exhibit significant nonlinearity, strong coupling, and time-varying parameter characteristics. Therefore, in the transition mode, the UAV as a whole is regarded as a nonlinear dynamic system, and its state equation can be expressed as:

[0069]

[0070] in, This is the system state vector, which includes variables such as position, velocity, attitude angle, and angular velocity. The integrated control input vector can be represented as resultant force / torque or equivalent control command; This is the nominal dynamic model of the system, used to describe the inherent dynamic characteristics of the UAV under undisturbed conditions; An input assignment matrix is ​​used to represent the mapping relationship between the integrated control input and the system state;

[0071] Considering uncertainties such as rotor-wing interference, aerodynamic parameter variations, model simplification errors, and external disturbances, the system is represented as:

[0072]

[0073] in, It represents the system dynamics function of a compound-wing UAV under nominal aerodynamic parameters and dynamic model conditions, reflecting the system state change law formed by aerodynamic force, thrust and gravity, etc., when neglecting modeling errors and external disturbances; This is the system's control input matrix, used to describe the influence of control inputs such as rotor thrust, control surface deflection angle, and propulsion system control quantities on the system dynamics. This represents the overall uncertainty of the system, encompassing the combined effects of model simplification errors, rotor-wing coupling disturbances, aerodynamic parameter variations, and external disturbances. This overall uncertainty is projected onto the error channel via output mapping to form an equivalent disturbance. It is used to characterize factors such as model simplification errors, rotor-wing interference, and external disturbances during the transition phase.

[0074] S32. Define the control output vector for the transition phase of the compound-wing UAV as follows:

[0075]

[0076] in, To output the mapping function, This is the system state vector;

[0077] Assume system output The expected reference trajectory is y r (t), then the overall error vector representing the deviation between the actual output and the expected trajectory, and the derivative of the error vector are:

[0078]

[0079] When the output mapping function When an explicit time term is included, it further includes To improve the derivative calculation, the system in step S31 is then... Substituting into the derivative formula for the error vector above, we get:

[0080]

[0081] The above equation can be rearranged into an overall error dynamics equation that facilitates controller design, as follows:

[0082]

[0083] in, , which is the nominal term of the global error dynamics equation; , is the input assignment matrix for the error channel; , which is the equivalent perturbation term of the error channel;

[0084] S33. To ensure rapid convergence of errors during the transition phase of the compound-wing UAV and the robustness of the system, an integral vector sliding surface is constructed for the error dynamics model:

[0085]

[0086] Where s(t) is the sliding mode variable, and its dimension is the same as that of the error vector e(t); It is a positive definite matrix used to adjust the error convergence rate. The integral term is the error term. By introducing the integral term, the strong robustness of sliding mode control to model uncertainty and external disturbances is retained, and the zero steady-state error convergence of the error can be achieved. At the same time, the chattering of control commands can be reduced, which is suitable for the control requirements of the transition mode of the compound wing UAV.

[0087] S34. Taking the derivative of the sliding mode variable s(t) defined in step S33 yields the derivative of the sliding mode variable:

[0088]

[0089] Substituting the overall error dynamics equation obtained in step S32 into the above equation and simplifying, we get:

[0090]

[0091] To ensure that the system state can reach and remain near the sliding surface within a finite time, while effectively suppressing the chattering problem of traditional sliding mode control, a continuous arrival law (boundary layer method) is selected, which takes the following form:

[0092]

[0093] Where K is the positive definite gain matrix. It is a saturation function. This is the boundary layer thickness, used to reduce chattering;

[0094] Combining the above equation with the aforementioned overall error dynamics, and rearranging, we get:

[0095]

[0096] S35. To address the comprehensive uncertainties in the transition phase of compound-wing UAVs, including model simplification errors, rotor-wing interference, and external disturbances, a neural network estimator is constructed to effectively handle the equivalent disturbances in the error channel. Perform online estimation, defining the equivalent perturbation as the mapping output of the neural network:

[0097]

[0098] Where N is the neural network mapping, This is the weight matrix of the neural network; For a neural network input vector, it must contain at least the error-related measurable quantities:

[0099]

[0100] To improve the accuracy of perturbation estimation, the neural network input vector It can further include characteristic quantities that reflect the transitional aerodynamic state, such as airspeed V, angle of attack α, and angular velocity q;

[0101] The neural network estimator can be implemented using a multilayer perceptron or a radial basis function network, and its output can be expressed as:

[0102]

[0103] in, Let W(t) be the activation function / basis function vector, W(t) be the neural network weight matrix, and T denote the transpose. To ensure the boundedness of network parameters and improve the stability of online learning, an adaptive update law based on sliding mode variables is used to adjust W(t) in real time. The update law has the following form:

[0104]

[0105] Where Γ is the positive definite learning rate matrix, For activation function / basis function vectors, Let s be the input vector of the neural network. T W(t) is the transpose of the sliding mode variable s(t), with the negative sign to ensure that the direction of weight update is consistent with the convergence requirements of sliding mode control; simultaneously, a projection / limiting operator is applied to W(t) to ensure that W(t) always remains within a preset bounded set, and the disturbance estimate is adjusted accordingly. Implement saturation limits:

[0106]

[0107] Where, Δ max The saturation limit is a preset equivalent disturbance upper limit. This saturation limit can effectively avoid abnormal neural network estimation output and prevent abnormal estimation output from causing sudden changes in control commands.

[0108] The neural network estimate Replace unknown disturbance terms Thus, the overall robust control law for engineering implementation is obtained:

[0109]

[0110] Where B(x,t) is the error channel input allocation matrix, u(t) is the integrated control input vector, F(x,t) is the nominal term of the error dynamics, which is composed of the nominal model and the derivative of the desired trajectory; Λ is a positive definite matrix, and e(t) is the error vector. Here, K is the equivalent perturbation estimate, K is the positive definite gain matrix, sat(·) is the saturation function, and s(t) is the sliding mode variable. The boundary layer thickness is used to decompose the actual disturbance into... ,in For bounded approximation error; robust term It is used to suppress approximation errors and external disturbances, ensuring that the system maintains stability margin and disturbance rejection performance even when there are errors in neural network estimation.

[0111] S36. Considering that the control input dimension and error dimension are usually inconsistent during the transition phase of a compound-wing UAV, resulting in the error channel input allocation matrix B(x,t) being typically non-invertible, a Moore-Penrose pseudo-inverse is adopted to solve for the control input in the sense of minimum norm, thereby ensuring the rationality, continuity, and engineering feasibility of the control allocation. Solving for the control input yields the overall sliding mode control law:

[0112]

[0113] Among them, u SMC (t) is the overall sliding mode control input vector. The Moore-Penrose pseudoinverse of the input matrix for the error channel is given, F(x,t) is the nominal term of the error dynamics, Λ is the positive definite matrix, and e(t) is the system error vector. Here, K is the equivalent perturbation estimate, K is the sliding mode arrival law gain matrix, sat(·) is the saturation function, and s(t) is the sliding mode variable. Boundary layer thickness;

[0114] Because the rotor thrust, propulsion system thrust, and fixed-wing aerodynamic lift / torque work together during the transition phase, the system is highly nonlinear and coupled with many uncertainties. Therefore, the control strategy design logic of this invention is as follows: First, an integral sliding mode surface is constructed at the error level, and rapid convergence and control chattering are reduced through a continuous arrival law. Then, the sliding mode condition is transformed into a control allocation equation. To address the issues of inconsistent input dimensions and non-invertible allocation matrix, the Moore-Penrose pseudo-inverse is introduced to obtain the overall control input with the minimum norm, ensuring that the control allocation is continuous and achievable. At the same time, an equivalent disturbance is estimated online using a neural network and amplitude-limited constraints are applied. The estimation error is covered by a sliding mode robust term, thereby achieving overall robust intelligent control under the transition mode and ensuring the stability and disturbance rejection performance of the system under complex operating conditions.

[0115] Step S4 specifically includes:

[0116] S41. The control commands are weighted and allocated according to the rotor weight and the fixed wing weight, so that the rotor system control commands are weighted and allocated accordingly. Control commands for fixed-wing systems They respectively satisfy:

[0117]

[0118] S42, Rotor system control commands Output to the rotor system is used to adjust rotor thrust or speed, and to transmit control commands to the fixed-wing system. Output to fixed-wing control surfaces for adjusting lift and pitch moment; and simultaneously output propulsion system control commands. To the propulsion system controller, where The throttle command is used to adjust the propeller thrust to establish and maintain the airspeed required for fixed-wing cruise, thereby achieving a smooth transition in which the rotor lift gradually withdraws and the wing aerodynamic lift gradually takes over during the transition phase.

[0119] like Figure 2 As shown, the transition mode control method for a compound-wing unmanned aerial vehicle of the present invention includes the following steps:

[0120] S1: Collect feedback status of the compound-wing UAV The feedback status includes at least airspeed V and angle of attack. A portion or a combination of altitude h, pitch angle θ, and used as input for subsequent transition decisions and control calculations;

[0121] S2: Determine whether the current stage is in transition based on the preset transition speed range [V1, V2]. When the conditions are met, the system enters a transition phase; when the conditions are not met, it is determined to be a non-transition phase. The initial velocity for the transition phase, The speed at which the transition phase ends;

[0122] S3: When it is determined that the transition phase has begun, for V and Normalization process is performed to obtain and And calculate the rotor control weights. With fixed-wing control weights When determining a non-transitional state, weights are pre-set: if V < V1, let... If V > V2, let Thus, the current weight pair is obtained ( , ), and satisfy ;

[0123] S4: Construct the overall error vector e(t) based on the state x(t) and the desired trajectory, and calculate the unified control variable. ;and with weight pairs ( , Input the weight constraint / scheduling module to obtain control commands. The control commands The output is sent to the actuator system and applied to the dynamic model of the compound wing UAV, forming a closed-loop updated feedback state x(t).

[0124] Where x(t) is the system state vector, and V is the airspeed. θ is the angle of attack, h is the altitude, and θ is the pitch angle; [V1, V2] is the transition speed range; and These are the normalized velocity index and the normalized angle of attack index, respectively. For rotor control weights, e(t) represents the control weights for the fixed-wing aircraft; e(t) represents the overall error vector. A unified control quantity generated for the overall control law; These are control commands after weight constraints / scheduling.

[0125] Through the above implementation methods, the present invention achieves continuous transfer of control dominance between rotor and fixed wing in transition mode by comprehensive weight scheduling based on speed and angle of attack; and through overall sliding mode control combined with neural network online compensation and pseudo-inverse control allocation, it can still ensure the stability, safety and control smoothness of the transition flight phase of the compound wing UAV even when dynamics and disturbances are not modeled.

[0126] In summary, the core innovation of this invention lies in constructing a dual-variable comprehensive weight allocation mechanism dominated by airspeed and corrected by angle of attack, solving the problems of discontinuous weight switching and fixed-wing stall risk. It employs a collaborative design of integral vector sliding surface and continuous saturation arrival law combined with neural network online disturbance compensation, balancing system robustness and low chattering requirements. The introduction of Moore-Penrose pseudo-inverse enables continuous and reasonable control allocation of multiple actuators, overcoming the technical bottleneck of control dimension mismatch. These three elements work together to achieve a continuous and smooth transfer of control dominance during the transition phase, effectively improving the stability, smoothness, and safety margin of transition flight. This method has strong engineering feasibility, is adaptable to various rotor-fixed-wing hybrid UAVs, and can be widely applied in various UAV operation scenarios requiring vertical take-off and landing and high-speed cruising, such as civilian surveying, power line inspection, emergency rescue, and logistics transportation. It significantly improves the reliability and mission adaptability of UAV transition flight, demonstrating good engineering application value and promising prospects for promotion.

[0127] The above embodiments are merely preferred embodiments of the present invention. All equivalent changes and improvements made in accordance with the scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent control of transition modes of a compound-wing unmanned aerial vehicle (UAV) with integrated weights, characterized in that, Includes the following steps: S1. Collect flight status information of the compound wing UAV and determine whether it has entered the transition stage from rotor mode to fixed wing mode based on airspeed; S2. When the transition phase is determined, a speed weight is constructed based on airspeed, and an angle-of-attack correction weight is constructed based on angle of attack. The speed weight and angle-of-attack correction weight are then fused to obtain the rotor control weight. With fixed-wing control weights The rotor control weights and fixed-wing control weights satisfy the normalization constraint. ; S3. Establish a nonlinear dynamic model of the transition mode of the compound wing UAV, construct the error vector and integral vector sliding surface, use neural network to estimate the equivalent disturbance online, and combine the sliding mode arrival law to solve the overall robust control input; S4. Based on rotor control weights With fixed-wing control weights The overall robust control input is weighted and distributed, and the weighted control commands are output to the rotor system, fixed-wing control surfaces and propulsion system respectively, so as to realize the continuous and smooth transfer of rotor control dominance to fixed-wing control dominance.

2. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 1, characterized in that, Step S1 specifically includes: S11. Real-time acquisition of flight speed V and angle of attack via airborne sensors. Altitude h, pitch angle θ, vertical velocity At least a portion or combination of the pitch angular velocity q, after filtering, constitutes the system state vector x(t)=[V, ,θ, ,q] T As a control input; S12, Preset transition speed range ,in, The initial velocity for the transition phase, The speed at the end of the transition phase; when It is determined at this time that the transition phase has begun; S13. After the compound-wing UAV enters the transition phase, it triggers a comprehensive weighted scheduling and overall robust control strategy for the transition mode. When, a rotor mode control strategy is adopted; when At that time, a fixed-wing modal control strategy was adopted.

3. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 1, characterized in that, Step S2 specifically includes: S21. Normalize the airspeed and angle of attack to obtain normalized airspeed and normalized angle of attack indices. S22. Construct a smooth and continuous velocity weight function based on the normalized velocity index; S23. Construct a piecewise linear angle-of-attack correction weight function based on the angle of attack to improve the rotor control weight under high angle-of-attack conditions to avoid fixed-wing stall. S24. The velocity weight function and the angle of attack correction weight function are fused through a saturation function to construct the rotor control weights. and according to Obtain fixed-wing control weights .

4. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 3, characterized in that, The velocity weighting function is constructed using a cosine smoothing function: in, The initial velocity for the transition phase, The velocity at the end of the transition phase is π, where π is the mathematical constant pi. It is a cosine function; when hour, =0 indicates that the rotor system is fully under control; when hour, =1 indicates that the fixed-wing system has complete control; when in the intermediate range At that time, the weights increase smoothly with airspeed, and the velocity weight function... exist The weights are continuous at the top and first-order continuous at the boundary to avoid control command chattering caused by abrupt changes in weights.

5. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 3, characterized in that, The angle of attack correction weight function is a piecewise function: When the absolute value of the angle of attack hour, No correction is made to the velocity weights; when hour, With the absolute value of the angle of attack The increase is linear, gradually raising the rotor weight; When the absolute value of the angle of attack hour, To maximize the rotor weight; in, For the angle of attack of the compound-wing UAV, For the angle of attack safety threshold, This is the angle of attack risk threshold.

6. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 1, characterized in that, Step S3 includes: S31. Establish the dynamic state equation containing comprehensive uncertainty terms. ,Will Mapped to equivalent perturbation of the error channel ; S32. Define the control output vector Expected trajectory Construct the error vector Derivation of the error dynamics model ; S33. Constructing an integral vector sliding surface ,in, It is a positive definite diagonal gain matrix; S34. Adopting the continuous saturation arrival law Where K is the positive definite gain matrix, and sat( ) is a saturation function. Given the boundary layer thickness, we obtain the fundamental equations for the control input: ； S35. Construct a neural network estimator to output an equivalent perturbation estimate. Adaptive update law is adopted Γ is the positive definite learning rate matrix, and substituting it into the matrix yields the overall robust control law; S36. Employing Moore–Penrose pseudo-inverse Solve for the overall sliding mode control input: Among them, u SMC (t) is the overall sliding mode control input vector, B † (x,t) is the Moore-Penrose pseudoinverse of the error channel input assignment matrix, F(x,t) is the error dynamics nominal term, Λ is the positive definite matrix, and e(t) is the system error vector. Here, K is the equivalent perturbation estimate, and SAT is the sliding mode arrival law gain matrix. ) is the saturation function, and s(t) is the sliding mode variable. This represents the boundary layer thickness.

7. The intelligent control method for the transition mode comprehensive weight of a compound-wing unmanned aerial vehicle according to claim 1, characterized in that, The weighted allocation in step S4 is as follows: S41, Weighted allocation yields rotor control commands. Fixed-wing control commands ; S42, Rotor system control commands Output to the rotor system is used to adjust rotor thrust or speed, and to transmit control commands to the fixed-wing system. Output to fixed-wing control surfaces for adjusting lift and pitch moment; and simultaneously output propulsion system control commands. To the propulsion system controller, where The throttle command is used to adjust the propeller thrust to establish and maintain the airspeed required for fixed-wing cruise, thereby achieving a smooth transition in which the rotor lift gradually withdraws and the wing aerodynamic lift gradually takes over during the transition phase.

8. A flight controller for a compound-wing unmanned aerial vehicle (UAV), characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the intelligent control method for the transition mode integrated weight of the compound wing UAV as described in any one of claims 1-7.

Images