A fixed-wing unmanned aerial vehicle wind disturbance observation and dynamic gain attitude control method
By combining a lightweight wind disturbance estimator and a dynamic gain adjustment term, the problems of fast response and overshoot suppression of UAVs are solved, realizing fast wind disturbance resistance control with low computational complexity, which is suitable for lightweight UAV platforms.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENYANG LIGONG UNIV
- Filing Date
- 2026-05-08
- Publication Date
- 2026-07-24
Smart Images
Figure CN122450147A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of UAV flight control, specifically relating to a method for wind disturbance observation and dynamic gain attitude control of fixed-wing UAVs, which can be applied to attitude stabilization control and wind disturbance resistance flight missions of fixed-wing UAVs. Background Technology
[0002] Fixed-wing unmanned aerial vehicles (UAVs) are susceptible to wind disturbance, airflow turbulence, and aerodynamic nonlinearity during low-altitude flight, leading to attitude angle oscillations, response lag, and control overshoot. Commonly used control methods include:
[0003] 1. Traditional PID control: simple to implement, but slow response and weak ability to suppress disturbances;
[0004] 2. Active disturbance rejection control based on extended state observer (ESO): It can estimate external disturbances, but it requires a large amount of computation and is difficult to adjust parameters;
[0005] 3. Sliding mode control: It is robust, but suffers from high-frequency chattering.
[0006] 4. Model Predictive Control (MPC): High accuracy but computationally burdensome, unsuitable for embedded flight controllers.
[0007] While ESO (Electrical Stability Oscillator) is widely used for attitude disturbance rejection in existing technologies, it typically relies on external wind speed measurements and requires real-time matrix operations, resulting in a large computational load and limiting its application on lightweight UAV platforms with limited computing power. For wind disturbance rejection control of fixed-wing UAVs, traditional methods generally suffer from problems such as gain fixation, insufficient recovery speed under large disturbances, and overshoot after rapid response. Meanwhile, existing sliding mode adaptive control often relies on… S The use of derivatives of order ˙ or higher and complex logical judgments further increases computational complexity, making it difficult to achieve stable implementation in actual flight control. Existing technologies cannot simultaneously solve the three technical challenges of "no external sensors required", "low computational cost O(1)" and "fast response without overshoot". Summary of the Invention
[0008] This invention aims to propose an attitude stabilization control method based on wind disturbance estimation and response acceleration. This method achieves real-time estimation of external wind disturbances by constructing a Lightweight Disturbance Observer (LDO) and introduces a dynamic gain adjustment term into the traditional control law. By combining a nonlinear weight function adjustment mechanism based on sliding mode surfaces, rapid convergence and overshoot suppression of attitude response are achieved. The entire algorithm can run in real time at 200Hz on an embedded flight control platform (such as PX4), with low computational complexity and no need for an external wind speed sensor. Under typical fixed-wing disturbance conditions, the maximum attitude angle overshoot can be reduced from 18° to 5°, and the disturbance recovery time is shortened by more than 40%. These performance improvements are achieved without adding sensors or system dimensions, solely through the control law itself, representing a substantial effect brought about by control structure innovation rather than parameter tuning. This invention introduces a dynamic gain adjustment structure with exponential decay and a single-parameter membership function, ensuring that the maximum gain only appears during large disturbances and automatically falls back to the steady-state value. Traditional methods lack this self-recovery mechanism. This invention requires no wind speed sensor, relying solely on attitude error. With angular velocity error Disturbance estimation and dynamic gain adjustment can be completed in one loop. The overall computational complexity is constant at O(1). O(1) represents constant-level computational complexity. The lightweight wind disturbance estimator LDO only requires 2 multiplications and 2 additions in a single loop, making it more suitable for embedded flight control environments.
[0009] The technical solution adopted by this invention to achieve the above objectives is: a method for wind disturbance observation and dynamic gain attitude control of a fixed-wing unmanned aerial vehicle, comprising the following steps:
[0010] The angular velocity and acceleration information of the UAV are obtained by the inertial measurement unit, and the attitude angle is obtained by filtering, thereby obtaining the attitude angle error. and angular velocity error ;
[0011] Wind disturbance estimates are obtained through the wind disturbance estimation module. ;
[0012] Based on the PID control framework, and according to the attitude angle error and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. According to the outer ring attitude control rate Obtain the inner ring angular velocity control rate To control the servo motor's movements.
[0013] The wind disturbance estimation value is obtained through the wind disturbance estimation module:
[0014]
[0015] in, This is an estimate of the wind disturbance. , For observation gain parameters, Used to provide rapid angular velocity error correction Used to suppress the accumulation of attitude angle errors.
[0016] The PID control framework is based on the attitude angle error. and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. The details are as follows:
[0017]
[0018] in, , For proportional gain and differential gain, For wind disturbance compensation, This is the dynamic gain, serving as a response acceleration term that is dynamically adjusted based on attitude error.
[0019] The dynamic gain This can be achieved using the following formula:
[0020]
[0021] in, For membership functions, the sliding surface , and All are coefficients. For gain parameters, Scaling factor The decay time constant, This is the timestamp for the gain's start.
[0022] The outer ring attitude control rate Obtain the inner ring angular velocity control rate The details are as follows:
[0023]
[0024] Wherein, saturation function Used to limit servo commands within a safe range For angular velocity feedback gain, This is the instruction limit.
[0025] A wind disturbance observation and dynamic gain attitude control system for a fixed-wing unmanned aerial vehicle (UAV) includes:
[0026] The data acquisition module is used to acquire the angular velocity and acceleration information of the UAV through the inertial measurement unit, obtain the attitude angle through filtering, and then obtain the attitude angle error. and angular velocity error ;
[0027] The wind disturbance estimation module is used to obtain wind disturbance estimates. ;
[0028] The control module is used to control the attitude angle error based on the PID control framework. and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. According to the outer ring attitude control rate Obtain the inner ring angular velocity control rate To control the servo motor's movements.
[0029] A device for wind disturbance observation and dynamic gain attitude control of a fixed-wing unmanned aerial vehicle (UAV) includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV when the computer program is executed.
[0030] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned method for wind disturbance observation and dynamic gain attitude control of a fixed-wing unmanned aerial vehicle.
[0031] The present invention has the following beneficial effects and advantages:
[0032] 1. This invention achieves real-time estimation of wind disturbance using a lightweight disturbance observer (LDO) with a computational complexity of O(1), and employs a composite dynamic gain adjustment law driven by a sliding surface, modulated by a single-parameter membership function, and exhibiting exponential decay characteristics. It achieves both fast response and low overshoot control without the need for a wind speed sensor.
[0033] 2. The lightweight disturbance observer of this invention does not rely on wind speed sensor input, but indirectly estimates the external disturbance torque only through inertial navigation information. Its advantages are that the update quantity is of constant order, the computational complexity is low, and it is suitable for real-time control.
[0034] 3. The present invention The regulation law does not rely on traditional fuzzy inference systems, but instead uses the error rate of change to directly drive the fuzzy weight function. This allows for single-parameter dynamic fuzzy adjustment, achieving low computational complexity. Because... All parameters change with attitude error S The convergence of the term naturally decreases, and the term will not continue to grow in steady state, thus ensuring the stability of the closed loop of the system.
[0035] 4. This method does not include the augmented state structure of ESO, nor does it rely on higher-order terms of sliding mode control. It belongs to a lightweight perturbation observation and dynamic gain fusion method with different technical paths. Attached Figure Description
[0036] Figure 1 The overall system structure block diagram of the present invention;
[0037] Figure 2 Flowchart of the LDO wind disturbance estimator of the present invention;
[0038] Figure 3 The Δ of the present invention K(t) Dynamic gain adjustment law structure diagram;
[0039] Figure 4 A schematic diagram of the signal flow of the dual-loop attitude control law of the present invention. Detailed Implementation
[0040] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.
[0041] This invention discloses a method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV. It achieves real-time estimation of wind disturbance using a lightweight disturbance observer (LDO) with a computational complexity of O(1). Simultaneously, it employs a composite dynamic gain adjustment law driven by a sliding surface, modulated by a single-parameter membership function, and exhibiting exponential decay characteristics. This achieves both fast response and low overshoot control without the need for a wind speed sensor. This method can be embedded in the PX4 control framework and is applicable to attitude stabilization control tasks for various fixed-wing UAVs. This method does not include the augmented state structure of an ESO and does not rely on higher-order terms of sliding mode control, belonging to a lightweight disturbance observation and dynamic gain fusion method with different technical approaches.
[0042] This invention mainly includes the following steps:
[0043] Inertial navigation data such as attitude angle and angular velocity are collected and then filtered and fused to obtain attitude estimation.
[0044] Constructing attitude dynamics equations: ;
[0045] A lightweight disturbance observer (LDO) is designed based on angular velocity error and attitude angle error to estimate external disturbances:
[0046]
[0047] in: .
[0048] The design includes dynamic gain adjustment. Control Law:
[0049]
[0050] The dynamic gain adjustment term The calculation depends only on the attitude angle error. and angular velocity error It does not require external wind speed signals or higher-order derivative inputs.
[0051] It is a type of sliding surface S Driven by a single-parameter membership function Modulation, and includes an exponential decay term. The composite adaptive gain law is calculated using the following structure:
[0052]
[0053] in: And all are non-zero positive numbers. , γ is a scaling factor, the value of which determines the steepness of the smooth curve tanh(). τdec is the adaptive gain decay time constant, and its value determines the rate of descent of the exponential curve; t a This is the gain trigger timestamp, its function is tt a The time elapsed since the last strong wind disturbance is calculated; t equals t0. a Time (i.e., tt) a =0) The exponential decay term equals 1, which has no compression effect in the system, and the system generates maximum gain to resist wind disturbance. Considering the actuator limitations and negative feedback characteristics of the actual physical system, the composite adaptive gain is calculated. Then, non-negative clipping is applied (i.e.) This is to prevent negative gain reversal when the sliding surface reverses due to slight overshoot of the system, thereby ensuring the attitude stability of the aircraft.
[0054] ;
[0055] It is a single-parameter membership function; For the adaptive gain, the exponential decay function is used. For the most recent satisfaction The moment; The adaptive gain decay time constant is a positive number; three terms , , None of them can be missing.
[0056] The control law output is used to correct the attitude via the servo motor actuator, and the system operates in a closed loop.
[0057] The output is obtained through a single-parameter membership function. Apply smoothing constraints.
[0058] To prevent overshoot, a velocity saturation function and a double-loop feedback mechanism are introduced into the control law: , To ensure that the control quantity does not exceed the allowable range of the servo motor, where The angular velocity feedback gain ranges from 0.2 to 1.0.
[0059] Example:
[0060] like Figure 1 As shown, the specific implementation process of the present invention is as follows:
[0061] 1. Flight status information collection:
[0062] Fixed-wing UAVs acquire angular velocity and acceleration information through an inertial measurement unit (IMU). While traditional systems often include airspeed sensors, the core algorithm of this invention does not rely on airspeed measurement. To eliminate drift, a fusion method based on complementary filtering or error state Kalman filtering (ESKF) is employed to obtain smooth and high-precision attitude angle estimates.
[0063]
[0064] in The acceleration of the aircraft as measured by the accelerometer. The aircraft's angular velocity as measured by the gyroscope. The speed is calculated using GPS measurement data. The attitude angle estimate is... As the filtered attitude angle measurement value Based on the desired target value, the attitude angle error and angular velocity error used for subsequent control calculations are calculated respectively.
[0065] Attitude angle error is defined as:
[0066] Angular velocity error is defined as:
[0067] The subscript d represents the expected value, which is the target attitude angle and target angular velocity that the flight control command requires the UAV to achieve; the subscript m represents the measured value, which is the current actual attitude and angular velocity of the UAV obtained by the sensor and after filtering and fusion.
[0068] 2. Establish an attitude dynamics model:
[0069] Establish the attitude dynamics equations of the UAV:
[0070]
[0071] in It is a 3×3 inertia matrix. For the body's angular acceleration, The control torque generated by the servo motor. This is due to structural nonlinearity or model error. This refers to the external wind disturbance moment. The control objective is to ensure that the external wind disturbance moment exists. and In this case, achieve attitude angle and desired attitude angle Rapid tracking.
[0072] 3. Wind disturbance estimation module design:
[0073] like Figure 2 As shown, for ease of expression, the formulas below are expressed in continuous-time form, but in actual flight control, they are implemented in discrete time. The discretization method is the forward Euler method. The control laws in this specification are all given in continuous form, and their discrete implementation is completed by the forward Euler method with Ts = 0.005–0.01s. To achieve real-time estimation of external wind disturbances, this invention introduces a lightweight disturbance observer (LDO) to replace the traditional ESO, and its state equation is:
[0074] Continuous form:
[0075] Discrete form:
[0076] in This is an estimate of the wind disturbance. , For observation gain parameters, The above parameter range is based on system stability margin analysis. The gain is selected using a Lyapunov function to ensure that the disturbance estimation error satisfies the exponential convergence condition, thus ensuring the LDO's stability throughout the entire flight envelope. In this invention, the stability of the LDO is achieved by constructing... The Lyapunov function proves that it satisfies .in Provides fast angular velocity error correction. Used to suppress the accumulation of attitude angle errors. For attitude angle error, The error is the angular velocity, determined by the target angular velocity. With measured value constitute( This observer does not rely on wind speed sensor input; it estimates external disturbance torque indirectly only through inertial navigation information. Its advantages include constant-order updates, low computational complexity, and suitability for real-time control. Compared to existing ESO observers, the LDO of this invention does not extend the system dimension, does not rely on the system Jacobian architecture, and does not perform the state decoupling and augmentation required in ESO. It only uses attitude angle error and angular velocity error as input, thus belonging to a lightweight disturbance observation structure based on a different principle.
[0077] 4. Rapid response mechanism and control law construction:
[0078] Based on the traditional PID control framework, this invention introduces a wind disturbance compensation term and a dynamic gain adjustment term into the control input. This forms a composite control law:
[0079]
[0080] in, , Based on the basic ratio and differential gain, For wind disturbance compensation, This is the response acceleration term that is dynamically adjusted based on the attitude error. The design is the key innovation of this invention. Its dynamic adjustment mechanism enables the control system to quickly improve its response speed under large disturbances, while automatically reducing the gain in a stable state to prevent overshoot and chattering.
[0081] 5. Response acceleration regulation law:
[0082] like Figure 3 As shown, to improve the speed of the attitude control system under sudden wind disturbances and avoid overshoot due to excessive gain, this invention proposes a composite adaptive gain law driven by a sliding surface, modulated with a single-parameter membership function, and exhibiting exponential decay. Its gain... The definition is as follows:
[0083]
[0084] in:
[0085] (1) Sliding surface
[0086]
[0087] (2) Three-term gain structure
[0088] : Enhances the speed of movement in the large error region; the approaching law term in sliding mode control has the function of giving the system state a large gain when it is far away from the sliding surface, so as to make it approach quickly.
[0089] It provides nonlinear compression with moderate error; and smoothing is used in the control field to reduce the "chattering" phenomenon in sliding mode.
[0090] : Compensates for low-frequency errors; Integral control term (similar to I in PID) is used to eliminate steady-state errors of the system and counteract constant or low-frequency disturbances.
[0091] The gain parameter range is set to , , .
[0092] (3) Membership function
[0093]
[0094] Smax represents the scale factor (or effective boundary width) of the sliding surface, and its value is a preset positive real number. This single-parameter membership function differs from traditional fuzzy inference in that it does not rely on a rule base, but only on... S The amplitude is modulated, causing the gain to automatically decrease in the low-error region. There is a lack of publicly available single-parameter membership functions for dynamic gain adjustment in attitude control in existing technologies; therefore, this method exhibits significant structural novelty and superior technical performance.
[0095] (4) Exponential decay term
[0096]
[0097] in For the most recent satisfaction At a certain moment, it is used to automatically eliminate the high-gain residue after the error enters steady state, allowing the system to gradually recover to a low-gain state after a rapid response, thus improving stability. is the scaling factor of the membership function. To determine the threshold for gain triggering, a value of 0.05–0.2 can be used.
[0098] This structure, compared to existing sliding mode adaptive methods (which typically only rely on...), is a significant improvement. or This invention adds a single-parameter membership modulation function and an exponential attenuation mechanism, realizing a dual mechanism of "fast response - automatic gain reduction" to effectively avoid overshoot. The regulation law does not rely on traditional fuzzy inference systems, but instead uses the error rate of change to directly drive the fuzzy weight function. This allows for single-parameter dynamic fuzzy adjustment, achieving low computational complexity. Because... All parameters change with attitude error S The convergence naturally decreases, and this term will not continue to increase in steady state, thus ensuring the closed-loop stability of the system. Existing sliding mode adaptive methods are usually based on The present invention provides single-term gain or multi-condition reasoning based on a fuzzy rule base. The regulation law uses a single-parameter exponential membership function. Driven directly by the rate of change of error, it does not rely on fuzzy rule bases and association systems, and does not require a large amount of matrix reasoning, thus belonging to a new regulation law configuration.
[0099] In traditional control methods, if control parameters are set very high to achieve rapid response, the drone can pull back abruptly, but this will result in overshoot and oscillation due to excessive inertia. Conversely, if stability is prioritized, the drone will react slowly and be unable to withstand strong winds. Dynamic gain adjustment law... When the drone encounters wind disturbances and experiences large errors, it can unleash high gain to quickly track the target. When the actual attitude is about to catch up with the desired attitude, it automatically reduces the gain using membership functions and exponential decay mechanisms, achieving a fast response without overshoot.
[0100] 6. Dual-loop control mechanism for rapid response and overshoot prevention
[0101] like Figure 4 As shown, to achieve fast response and avoid overshoot caused by large gain, a dual-loop structure is required, with an outer loop constraining the attitude angle error and an inner loop rapidly adjusting the angular velocity.
[0102] Outer loop output reference control quantity:
[0103] Inner ring execute command:
[0104] Inner Ring Road Enhanced angular velocity feedback enables rapid response, among which The angular velocity feedback gain, which can be set to 0.2-1.0, is used to enhance the inner loop damping; the outer loop limits the rate of change of attitude angle error to prevent overshoot. Saturation function. Limit servo commands within a safe range to avoid violent oscillations. Traditional sliding mode adaptive control in dual-loop implementations generally requires... S Or angular acceleration signals, but this method does not require them at all. S Only by , Driving gain changes. (And dependency) S Unlike sliding mode adaptive methods with higher-order derivatives, this method does not require acceleration measurement or higher-order analysis, further reducing implementation complexity and making it suitable for embedded execution.
[0105] In fixed-wing aircraft, the outer and inner loops form a cascade control structure. The outer loop is the attitude angle control loop, and the control target is the attitude angle. In other words, at what tilt angle is the aircraft? The focus is on the aircraft's final position and steady-state error. The input is the attitude angle error. Wind disturbance estimation Angular velocity error The output signal is a torque command. The inner loop is an angular velocity control loop, and the control target is angular velocity. This refers to the speed at which the aircraft is changing its attitude (roll, pitch). The focus is on the instantaneous dynamics of the aircraft, and the input signal is the reference command from the outer loop. and angular velocity error The output signal is the final servo execution command. .
[0106] 7. Closed-loop control execution
[0107] The control law output is converted into aileron and elevator deflection via servo actuators, achieving attitude correction. The system operates in closed-loop real-time with a cycle of 5–10 ms.
[0108] The present invention has the following advantages:
[0109] 1. Low computational complexity: The LDO model contains only two additions and multiplications, and can run in real time on embedded flight controllers such as PX4 and Pixhawk.
[0110] 2. No wind speed sensor required: Wind disturbance estimation is achieved based on the angular velocity error feedback of the IMU, reducing reliance on sensors and improving the system's practicality and environmental adaptability.
[0111] 3. Fast, overshoot-free response: Based on sliding surface and single-parameter membership function suppression. Dynamic adjustment enables unified control of response acceleration and overshoot suppression, balancing response speed and stability.
[0112] 4. The algorithm is highly versatile: it can be embedded in Simulink or PX4 control frameworks and is compatible with existing attitude controller modules.
[0113] 5. Strong robustness: High tolerance to modeling errors and sudden wind disturbances;
[0114] 6. Compared to the ESO method, this method does not extend the system state dimension, thus reducing the computational load;
[0115] 7. Compared with traditional fuzzy control, the fuzzy adjustment law driven by direct error has a simpler structure;
[0116] 8. Compared with fixed-gain PID, it achieves adaptive disturbance compensation and dynamic gain adjustment.
Claims
1. A method for wind disturbance observation and dynamic gain attitude control of a fixed-wing unmanned aerial vehicle, characterized in that, Includes the following steps: The angular velocity and acceleration information of the UAV are obtained by the inertial measurement unit, and the attitude angle is obtained by filtering, thereby obtaining the attitude angle error. and angular velocity error ; Wind disturbance estimates are obtained through the wind disturbance estimation module. ; Based on the PID control framework, and according to the attitude angle error and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. ; Based on the outer ring attitude control rate Obtain the inner ring angular velocity control rate To control the servo motor's movements.
2. The method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV according to claim 1, characterized in that, The wind disturbance estimation value is obtained through the wind disturbance estimation module: ； in, This is an estimate of the wind disturbance. , For observation gain parameters, Used to provide rapid angular velocity error correction Used to suppress the accumulation of attitude angle errors.
3. The method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV according to claim 1, characterized in that, The PID control framework is based on the attitude angle error. and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. The details are as follows: ； in, , For proportional gain and differential gain, For wind disturbance compensation, This is the dynamic gain, serving as a response acceleration term that is dynamically adjusted based on attitude error.
4. The method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV according to claim 3, characterized in that, The dynamic gain This can be achieved using the following formula: ； in, For membership functions, the sliding surface , and All are coefficients. For gain parameters, Scaling factor The decay time constant, This is the timestamp for the gain's start.
5. The method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV according to claim 1, characterized in that, The outer ring attitude control rate Obtain the inner ring angular velocity control rate The details are as follows: ； Wherein, saturation function Used to limit servo commands within a safe range For angular velocity feedback gain, This is the instruction limit.
6. A wind disturbance observation and dynamic gain attitude control system for a fixed-wing unmanned aerial vehicle, characterized in that, include: The data acquisition module is used to acquire the angular velocity and acceleration information of the UAV through the inertial measurement unit, obtain the attitude angle through filtering, and then obtain the attitude angle error. and angular velocity error ; The wind disturbance estimation module is used to obtain wind disturbance estimates. ; The control module is used to control the attitude angle error based on the PID control framework. and angular velocity error Wind disturbance estimates and the constructed dynamic gain This forms an outer-loop attitude composite control law. According to the outer ring attitude control rate Obtain the inner ring angular velocity control rate To control the servo motor's movements.
7. A device for wind disturbance observation and dynamic gain attitude control of a fixed-wing unmanned aerial vehicle, characterized in that, It includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement, when executing the computer program, a method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV as described in any one of claims 1-5.
8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements a method for wind disturbance observation and dynamic gain attitude control of a fixed-wing UAV as described in any one of claims 1-5.