Quadrotor unmanned aerial vehicle attitude decoupling-free control method based on improved active disturbance rejection technology

Abstract

The invention discloses a quadrotor unmanned aerial vehicle attitude decoupling-free control method based on an improved active disturbance rejection technology in the technical field of quadrotor unmanned aerial vehicles. Comprising the steps of defining a coordinate system and an attitude angle of the four-rotor unmanned aerial vehicle; performing four-rotor unmanned aerial vehicle dynamics modeling; designing an active disturbance rejection control strategy; designing a high-order expansion state observer; and introducing a Levant differentiator, and designing a parameter estimator. According to the method, the extended state observer is used for estimating system dynamic and environmental interference, unknown disturbance is compensated in real time in combination with parameter estimation, and the anti-interference capability and the dynamic response performance are enhanced. Meanwhile, a Levant differentiator is introduced to replace a traditional tracking differentiator (TD) link, and unknown disturbance is estimated through a parameter estimator. Through a decoupling-free control strategy, a complex decoupling process in a traditional method is avoided, and the control efficiency is improved. And finally, a simulation experiment proves that the improved active-disturbance-rejection control scheme still keeps the effectiveness of good tracking precision under high-frequency disturbance.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) control, and in particular to a posture control method and system for a quad-rotor UAV. Background Art

[0002] With the rapid development of drone technology, quadcopters (UAVs) have demonstrated tremendous application potential in multiple fields due to their flexible maneuverability and easy operation. For example, in the power industry, quadcopters can be used for high-voltage transmission line inspections and rapid fault detection. In agriculture, they can be used for farmland irrigation and pesticide spraying, improving operational efficiency. In logistics and distribution, quadcopters enable rapid last-mile delivery. Furthermore, they have found widespread application in environmental monitoring, emergency rescue, and aerial filming. However, quadcopters face numerous challenges in flight attitude control. Their flight systems are highly complex and nonlinear. The inherent uncertainty of their parameters (such as variations in motor performance, battery charge, and airframe mass distribution) and their strong coupling make them highly susceptible to external disturbances during flight. During modeling, coupling issues primarily manifest themselves in the following: aerodynamic coupling between the rotors, which affects lift and thrust output; the coupling between attitude and position, which causes attitude changes to affect the flight trajectory, while position adjustments, in turn, affect attitude stability. These coupling characteristics increase the complexity of the control system, making precise control and stable flight more difficult. Furthermore, drones are subject to a variety of external disturbances during flight, such as wind speed fluctuations, air turbulence, electromagnetic interference, and terrain undulations. These interfering factors interact with the coupled characteristics of drones, further exacerbating the complexity of flight attitude control. Currently, existing control methods are primarily designed based on small-disturbance linearization models or nonlinear models. While small-disturbance linearization methods offer theoretical simplicity, they struggle to adapt to the control requirements of drones operating over large ranges and in complex environments. While nonlinear control methods can better describe the dynamics of drones, they require high model accuracy and computational complexity, making them difficult to implement in real-time control systems. Furthermore, some research has explored using neural networks to compensate for the uncertainty introduced by unknown disturbances. However, neural network training requires extensive data support and may lack generalization capabilities when faced with unseen interference patterns. While these existing control methods can achieve a certain degree of control over drone attitude, they still have significant limitations when faced with complex coupling relationships and multi-source interference, making them difficult to meet the demanding requirements for stable flight in complex environments. In particular, tracking accuracy and attitude control stability are often difficult to maintain in situations where random airflow is excessively large or complex, leading to problems such as inaccurate tracking and poor attitude control stability. These issues not only affect the operational efficiency and safety of drones, but also limit their application in a wider range of fields. Therefore, developing a quadrotor drone attitude control method that can effectively cope with complex coupling relationships and multi-source interference is of great practical significance for improving the performance and application range of drones. Summary of the Invention

[0003] The purpose of the present invention is to address the deficiencies of the existing technology and to provide a decoupling-free control method for the attitude of a four-rotor UAV based on an improved active disturbance rejection technology, aiming to achieve high-precision control of the UAV's flight attitude. The present invention uses active disturbance rejection control (ADRC) as the core main controller, which is responsible for the main attitude adjustment tasks. The adaptive estimator serves as an auxiliary module to estimate unknown disturbances in real time and input the results into ADRC to assist in precise control. The attitude controller is based on the improved active disturbance rejection control technology and minimizes the deviation between the actual attitude and the target attitude through an optimization algorithm to achieve efficient and precise adjustment. In order to improve the robustness of the system, an adaptive estimator is introduced to pre-process the sensor data, reduce the sensitivity to external interference, and enhance the anti-interference ability. At the same time, the introduction of a decoupler reduces the complexity and computational complexity of the system, simplifies the system structure, and ensures the stability of the control effect.

[0004] In order to achieve the above-mentioned purpose, the technical solutions provided by the present invention are as follows: A decoupling-free attitude control method for a quadrotor UAV based on an improved active disturbance rejection technology comprises the following steps: Step 1: Establish a dynamic model of the quadrotor drone, which includes the body coordinate system ( b X b Y b Z b ) and the inertial coordinate system (O e X e Y e Z e ), define the pitch angle, roll angle and yaw angle of the quadrotor drone; determine the dynamic model of the quadrotor drone as follows: Among them, φ, θ, ψ represent the pitch angle, roll angle and yaw angle of the quadrotor drone respectively, J x ,J y ,J z The moment of inertia around the body axis and the external disturbance torque are respectively around the body axis X b 、Y b and Z b The moment of inertia, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k t is the air resistance coefficient, f i (i=1,2,3) is the total disturbance; Define f i Differentiable, and denoted as h i : Step 2: Introduce the Levant differentiator to replace the traditional tracking differentiator (TD) link to optimize the command change rate response speed; Step 3: Design a high-order extended state observer to accurately estimate the tracking error change rate and total disturbance to improve system robustness. The linear extended state observer (LESO) is as follows: Among them, z1, z2 and z3 are the state variables of LESO, which estimate the output y and output derivative of the system respectively. and the total disturbance f(·); β1, β2, and β3 are adjustable observer gains, and β1 = 3ω o 、 ω o is the adjustable observer bandwidth, u is the control input, b0 is the control gain, o represents the observer parameter, and h is the differential of the total disturbance; Step 4: Design the ADRC controller by analyzing the dynamic model in step 1 and combining the output data from steps 2 and 3 to calculate the system input. Step 5: Use the least squares method with exponential forgetting to design an estimator to estimate the unknown disturbance in real time; Step 6: Design a decoupling-free controller according to the dynamic model to achieve three-channel decoupling-free control; Step 7: Combine parameter estimation to compensate for unknown disturbances in real time, enhance anti-interference capability and dynamic response performance, and build an auto-disturbance rejection controller that integrates error change rate estimation to achieve precise control of the quadrotor UAV's attitude.

[0005] Furthermore, the parameters of the Levant differentiator in step 2 are set to ensure its signal tracking and differential evaluation capabilities in a random noise environment. The Levant differentiator algorithm is as follows: Where r is the signal to be differentiated, c1 is the tracking signal of r, c2 is the first-order differential signal tracking r, c3 is the second-order differential signal tracking r, c4 is the third-order differential signal tracking r, λ1, λ2, and λ3 are the parameters to be adjusted in the differentiator, v1, v2, and v3 are intermediate variables, and sign(·) is the sign function.

[0006] Furthermore, in the step of optimizing the transfer function of the linear extended state observer using the proportional derivative term active disturbance rejection control law established in step 3 and outputting the transfer function as a control signal, the optimized transfer function is expressed as: Among them, ω c is the adjustable observer bandwidth, c represents the observer parameter, and s is a complex variable representing the frequency in the Laplace transform domain. Since all the poles of the transfer function are located in the left half of the complex plane (the real part is negative), the system is guaranteed to be stable.

[0007] Furthermore, the active disturbance rejection controller in step 4 adopts a linear state error feedback control law based on disturbance estimation. The specific expression of the control input is as follows: in, is the estimated output of the Levant differentiator for the desired attitude angle and its derivative; 11 、z 21 and z 31 are the first outputs of the state observers designed for φ, θ, and ψ, i.e., the estimated values of the system input y; z 12 、z 22 and z 32 is the second output of the state observer, i.e., the derivative estimate of the system input y; z 13 、z 23 and z 33 are the third output of the state observer, namely the estimated value of the total disturbance, and b0 is the control gain.

[0008] Furthermore, the control bandwidth ω in step 4 c , the calculation formula is as follows: Where N is the order of the system, T 98％ The time required for the system response curve to first enter and remain at 98% of its final value.

[0009] Furthermore, the method in step 5 also includes estimating the unknown parameters in the system based on online parameter estimation and combining the attitude angle change rate tracking error. The method for estimating the unknown parameter d is as follows: Where W(t) is the signal matrix that describes the relationship between the system and the unknown parameters. It can be obtained from the measurement of the system signal. The update rule of the gain P is as follows: Among them, λ0 and k0 are positive constants, representing the preset boundaries of the maximum forgetting rate and the gain matrix P, respectively.

[0010] Furthermore, the active disturbance rejection control module in step 6 uses three-channel decoupling-free control, which eliminates the need to decouple the coupled part, thus reducing the difficulty of controller design. The decoupler-free design is as follows: in,

[0011] Furthermore, the method in step 7 also includes solving the actual control variable, that is, the rotor speed, under the condition of force balance in the height direction, as follows: Among them, U L For along O b Z b Total lift on the axis, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k L is the lift coefficient, b is the anti-torque coefficient, l is the distance from the rotor center to the center of mass of the UAV, and ω1, ω2, ω3, and ω4 are the rotational speeds of the four rotors.

[0012] The quadrotor UAV attitude control method based on the composite improved ADRC method proposed in this paper significantly improves the performance of the quadrotor UAV attitude control system, mainly in the following aspects: 1. Improved system stability: The designed high-order extended state observer accurately estimates and compensates for disturbances in the system in real time, ensuring that the drone maintains attitude stability in the face of external interference and internal changes. Furthermore, the optimized control law design and stable system transfer function further enhance the system's stability and anti-interference capabilities in complex environments, effectively avoiding attitude fluctuations and system instability caused by disturbances. 2. Reduced decoupling complexity: Three-channel decoupling-free control based on active disturbance rejection technology eliminates the need for complex and difficult-to-accurate decoupling operations on the coupled parts, directly controlling the coupled system. This significantly reduces the controller design difficulty and system complexity, improves the reliability and feasibility of the control system, and simplifies the overall control process. 3. Optimized Control Precision: The introduction of the Levant differentiator optimizes the response speed of the command change rate, enabling the system to quickly and accurately respond to attitude command changes and reducing control latency. Furthermore, online parameter estimation, which compensates for unknown parameter changes in real time, and the high-order extended state observer, which provides precise system state estimation, help the controller accurately calculate control inputs, thereby achieving high-precision control of the drone's attitude, significantly improving control accuracy and the system's dynamic response performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a decoupling-free attitude control method for a quadrotor UAV based on an improved active disturbance rejection technology proposed by the present invention; Figure 2 This is a structural diagram of the decoupling-free attitude control method for a quadrotor UAV based on the improved active disturbance rejection technology proposed in the present invention; Figure 3 is the target tracking curve in Experimental Example 1; Figure 4 is the random airflow disturbance curve in Experimental Example 1; Figure 5 is the pitch angle trajectory tracking curve in Experimental Example 1; Figure 6 is the rolling angle trajectory tracking curve in Experimental Example 1; Figure 7 is the yaw angle trajectory tracking curve in Experimental Example 1; Figure 8 is the pitch angle trajectory tracking error curve in Experimental Example 1; Figure 9 is the rolling angle trajectory tracking error curve in Experimental Example 1; Figure 10 is the yaw angle trajectory tracking error curve in Experimental Example 1; Figure 11 is the estimated curve of the unknown parameters in the pitch angle in Experimental Example 1; Figure 12 is the estimated curve of the unknown parameters in the roll angle in Experimental Example 1; Figure 13 is the estimated curve of the unknown parameters in the yaw angle in Experimental Example 1; Figure 14 is the pitch angle tracking error curve under noise in Experimental Example 1; Figure 15 is the roll angle tracking error curve under noise in Experimental Example 1; Figure 16 is the pitch angle tracking error curve under noise in Experimental Example 1; Figure 17 This is the rotor speed response curve in Experimental Example 1. DETAILED DESCRIPTION

[0014] The present application is described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but are not intended to limit the present application in any form. It should be noted that, for those skilled in the art, several variations and improvements can be made without departing from the concept of the present application. These fall within the scope of protection of the present application.

[0015] Example 1 The present application discloses a decoupling-free attitude control method for a quadrotor UAV based on an improved active disturbance rejection technology, comprising the following steps: Step 1: Establish a dynamic model of the quadrotor drone, which includes the body coordinate system ( b X b Y b Z b ) and the inertial coordinate system (O e X e Y e Z e ), define the pitch angle, roll angle and yaw angle of the quadrotor drone; determine the dynamic model of the quadrotor drone as follows: Among them, φ, θ, ψ represent the pitch angle, roll angle and yaw angle of the quadrotor drone respectively, J x ,J y ,J z The moment of inertia around the body axis and the external disturbance torque are respectively around the body axis X b 、Y b and Z b The moment of inertia, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k t is the air resistance coefficient, f i (i=1,2,3) is the total disturbance; Define f i Differentiable, and denoted as h i : Step 2: Introduce the Levant differentiator to replace the traditional tracking differentiator (TD) link to optimize the command change rate response speed; Step 3: Design a high-order extended state observer to accurately estimate the tracking error change rate and total disturbance to improve system robustness. The linear extended state observer (LESO) is as follows: Among them, z1, z2 and z3 are the state variables of LESO, which estimate the output y and output derivative of the system respectively. and the total disturbance f(·); β1, β2, and β3 are adjustable observer gains, and β1 = 3ω o 、 ω o is the adjustable observer bandwidth, u is the control input, b0 is the control gain, o represents the observer parameter, and h is the differential of the total disturbance; Step 4: Design the ADRC controller by analyzing the dynamic model in step 1 and combining the output data from steps 2 and 3 to calculate the system input. Step 5: Use the least squares method with exponential forgetting to design an estimator to estimate the unknown disturbance in real time; Step 6: Design a decoupling-free controller according to the dynamic model to achieve three-channel decoupling-free control; Step 7: Combine parameter estimation to compensate for unknown disturbances in real time, enhance anti-interference capability and dynamic response performance, and build an auto-disturbance rejection controller that integrates error change rate estimation to achieve precise control of the quadrotor UAV's attitude.

[0016] Furthermore, in step 2, the parameters of the Levant differentiator are set to ensure its signal tracking and differential evaluation capabilities in a random noise environment. The Levant differentiator algorithm is as follows: Where r is the signal to be differentiated, c1 is the tracking signal of r, c2 is the first-order differential signal tracking r, c3 is the second-order differential signal tracking r, c4 is the third-order differential signal tracking r, λ1, λ2, and λ3 are the parameters to be adjusted in the differentiator, and sign(·) is the sign function.

[0017] Furthermore, in step 3, in the step of optimizing the transfer function of the linear extended state observer by establishing the active disturbance rejection control law of the proportional derivative term and outputting the transfer function as the control signal, the optimized transfer function is expressed as: Among them, ω cis the adjustable observer bandwidth, c represents the observer parameter, and s is a complex variable representing the frequency in the Laplace transform domain. Since all the poles of the transfer function are located in the left half of the complex plane (the real part is negative), the system is guaranteed to be stable.

[0018] Furthermore, in step 4, the composite active disturbance rejection controller adopts a linear state error feedback control law based on disturbance estimation, and the specific expression of the control input is as follows: in, is the estimated output of the Levant differentiator for the desired attitude angle and its derivative; 11 、z 21 and z 31 are the first outputs of the state observers designed for φ, θ, and ψ, i.e., the estimated values of the system input y; z 12 、z 22 and z 32 is the second output of the state observer, i.e., the derivative estimate of the system input y; z 13 、z 23 and z 33 are the third output of the state observer, i.e., the estimated value of the total disturbance, and b0 is the control gain. The bandwidth of the controller is calculated as follows: Where N is the order of the system, T 98％ The time required for the system response curve to first enter and remain at 98% of its final value.

[0019] Furthermore, in step 5, the method further includes estimating the unknown parameters in the system based on online parameter estimation in combination with the attitude angle change rate tracking error. The method for estimating the unknown parameter d is as follows: Where W(t) is the signal matrix that describes the relationship between the system and the unknown parameters. It can be obtained from the measurement of the system signal. The update rule of the gain P is as follows: Among them, λ0 and k0 are positive constants, representing the preset limits of the maximum forgetting rate and the gain matrix P, respectively. Two different methods are proposed for the estimation of unknown parameters. The first method involves an overall estimation of all unknown quantities, which requires the controlled object to provide information on the angle change rate in real time. In view of this, we further propose a second method, which first determines the periodic characteristics of the unknown parameters in an offline manner, and then determines other unknown coefficients based on this periodic information. The former is called the low-precision active disturbance rejection control method (LPADRC), and the latter is called the high-precision active disturbance rejection control method (HPADRC). Signal matrix of the high-precision estimator W=[1,sin(0.25πt),sin(3t)], the signal matrix W of the low-precision estimator is 1.

[0020] Then, the active disturbance rejection control module described in step 6 uses three-channel decoupling-free control, which eliminates the need to decouple the coupling part, thus reducing the difficulty of controller design. The decoupler-free design is as follows: in,

[0021] Finally, under the condition of balanced forces in the height direction, the actual control variable, the rotor speed, is solved as follows: Among them, U L For along O b Z b Total lift on the axis, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k L is the lift coefficient, b is the anti-torque coefficient, l is the distance from the rotor center to the center of mass of the UAV, and ω1, ω2, ω3, and ω4 are the rotational speeds of the four rotors.

[0022] The following simulation is conducted on a quadrotor UAV attitude decoupling-free control method based on the improved ADRC technology disclosed in this application to prove the effectiveness and feasibility of the control strategy. The specific situation is as follows: the UAV mass is 0.8 kg, the x-axis moment of inertia is 5.445×10 -3kg·m 2 The y-axis moment of inertia is 5.445×10 -3 kg·m 2 , the z-axis moment of inertia is 1.089×10 - 2 kg·m 2 , the distance from the rotor to the center of mass is 0.165m, and the control gain is 2×10 -6 , the lift coefficient is 2.98×10 -5 , the air resistance coefficient is 9×10 -2 .

[0023] The simulation results are as follows: The target tracking curve is as follows Figure 3 As shown in FIG. 3 , the target tracking curve, the proportional-integral-derivative controller (PID) and the traditional ADRC are control experimental curves, and the LPADRC and HPADRC are the attitude control algorithms designed by the present invention. Figure 3-Figure 16 It can be seen that after the disturbance occurs, the drone experiences small disturbances around 2s, 6s, 12s, and 25s. However, a comparison shows that the HPADRC and LPADRC methods can achieve high-precision tracking of attitude commands in strong interference environments. Traditional ADRC methods can ensure high-precision tracking of attitude commands under no or low interference conditions, but attitude angle tracking accuracy is significantly reduced under high-frequency disturbances. PID control methods can also experience steady-state errors in the presence of unknown disturbances.

[0024] Figure 8 、 9 Figures 1 and 10 are the trajectory tracking error curves for pitch, roll, and yaw angles. As can be seen in the figure, the improved controller design in the attitude loop significantly outperforms ADRC and PID, and its robustness is also superior to traditional ADRC and PID control algorithms. The pitch angle-time curve clearly shows that the LPADRC controller has fewer amplitudes and jitters, a smaller jitter amplitude, and a shorter convergence time. HPADRC, on the other hand, has a stronger ability to suppress high-frequency disturbances. Overall, different control methods have different response speeds when adjusting the target angle. It can be observed that LPADRC always responds the fastest when disturbances occur. In terms of stability, HPADRC consistently stays close to the target curve most of the time, demonstrating better stability. PID, on the other hand, experiences larger fluctuations when disturbances occur, and exhibits relatively poor stability.

[0025] Figure 11 、 12Figures 1 and 13 show the estimated curves for the unknown parameters of pitch, roll, and yaw. As can be seen, the curves exhibit error fluctuations at 2, 6, and 12 seconds. This is due to the addition of a disturbance to the drone before and after these time points, simulating random airflow disturbances. However, through offline parameter estimation, the instrument is able to quickly adapt to these changes within approximately 1.4 seconds, achieving an accurate estimate of the unknown disturbance.

[0026] Figure 14 、 15 Figures 1 and 16 show the attitude tracking error curves under noise, demonstrating the attitude tracking performance in a noisy environment. Gaussian white noise with a mean of zero and a covariance of 0.001 is introduced into the attitude angle measurement. It can be clearly observed from the figure that despite the influence of measurement noise, the attitude tracking error decreases rapidly after the system starts and eventually converges to a small range close to zero. This result demonstrates that the developed control strategy not only effectively handles system uncertainties and external disturbances but also maintains high control accuracy despite noise interference.

[0027] Figure 17 The response curve of the UAV rotor speed when the lift offsets the gravity is given.

[0028] The technical means disclosed in the solutions of the present invention are not limited to those disclosed in the above-mentioned embodiments, but also include technical solutions composed of any combination of the above-mentioned technical features. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, which are also considered to be within the scope of protection of the present invention.

Claims

1. A decoupling-free attitude control method for a quadcopter unmanned aerial vehicle (UAV) based on improved active disturbance rejection technology is designed to achieve high-precision control of the UAV's flight attitude. The present invention uses Active Disturbance Rejection Control (ADRC) as the core main controller, responsible for the main attitude adjustment tasks. An adaptive estimator serves as an auxiliary module, which estimates unknown disturbances in real time and inputs the results into ADRC to facilitate precise control. The attitude controller is based on improved active disturbance rejection control technology and minimizes the deviation between the actual and target attitudes through an optimization algorithm to achieve efficient and precise adjustment. To improve the robustness of the system, an adaptive estimator is introduced to preprocess sensor data, reduce sensitivity to external interference, and enhance anti-interference capability. At the same time, the introduction of a decoupler reduces system complexity and computational complexity, simplifies the system structure, and ensures the stability of the control effect.

2. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technology according to claim 1 is characterized in that: The method may include the following steps: Step 1: Establish a dynamic model of the quadrotor drone, which includes the body coordinate system ( b X b Y b Z b ) and the inertial coordinate system (O e X e Y e Z e ), define the pitch angle, roll angle and yaw angle of the quadrotor drone; determine the dynamic model of the quadrotor drone as follows: Among them, φ, θ, ψ represent the pitch angle, roll angle and yaw angle of the quadrotor drone respectively, J x ,J y ,J z The moment of inertia around the body axis and the external disturbance torque are respectively around the body axis X b 、Y b and Z b The moment of inertia, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k t is the air resistance coefficient, f i (i=1,2,3) is the total disturbance; Define f i Differentiable, and denoted as h i : Step 2: Introduce the Levant differentiator to replace the traditional tracking differentiator (TD) link to optimize the command change rate response speed; Step 3: Design a high-order extended state observer to accurately estimate the tracking error change rate and total disturbance to improve system robustness. The linear extended state observer (LESO) is as follows: Among them, z1, z2 and z3 are the state variables of LESO, which estimate the output y and output derivative of the system respectively. and the total disturbance f(·); β1, β2, and β3 are adjustable observer gains, and β1 = 3ω o 、 ω o is the adjustable observer bandwidth, u is the control input, b0 is the control gain, o represents the observer parameter, and h is the differential of the total disturbance; Step 4: Design the ADRC controller by analyzing the dynamic model in step 1 and combining the output data from steps 2 and 3 to calculate the system input. Step 5: Use the least squares method with exponential forgetting to design an estimator to estimate the unknown disturbance in real time; Step 6: Design a decoupling-free controller according to the dynamic model to achieve three-channel decoupling-free control; Step 7: Combine parameter estimation to compensate for unknown disturbances in real time, enhance anti-interference capability and dynamic response performance, and build an auto-disturbance rejection controller that integrates error change rate estimation to achieve precise control of the quadrotor UAV's attitude.

3. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technology according to claim 2 is characterized in that: The parameters of the Levant differentiator in step 2 are set to ensure its signal tracking and differential evaluation capabilities in a random noise environment. The Levant differentiator algorithm is as follows: Where r is the signal to be differentiated, c1 is the tracking signal of r, c2 is the first-order differential signal tracking r, c3 is the second-order differential signal tracking r, c4 is the third-order differential signal tracking r, λ1, λ2, and λ3 are the parameters to be adjusted in the differentiator, v1, v2, and v3 are intermediate variables, and sign(·) is the sign function.

4. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technology according to claim 2 is characterized in that: In step 3, in the step of establishing the active disturbance rejection control law of the proportional derivative term to optimize the transfer function of the linear extended state observer and outputting the transfer function as the control signal, the optimized transfer function is expressed as: Among them, ω c is the adjustable observer bandwidth, c represents the observer parameter, and s is a complex variable representing the frequency in the Laplace transform domain. Since all the poles of the transfer function are located in the left half of the complex plane (the real part is negative), the system is guaranteed to be stable.

5. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technology according to claim 2 is characterized in that: The active disturbance rejection controller in step 4 adopts a linear state error feedback control law based on disturbance estimation. The specific expression of the control input is as follows: in, is the estimated output of the Levant differentiator for the desired attitude angle and its derivative; 11 、z 21 and z 31 are the first outputs of the state observers designed for φ, θ, and ψ, i.e., the estimated values of the system input y; z 12 、z 22 and z 32 is the second output of the state observer, i.e., the derivative estimate of the system input y; z 13 、z 23 and z 33 are the third output of the state observer, namely the estimated value of the total disturbance, and b0 is the control gain.

6. The control bandwidth ω according to claim 5 c , which is calculated as follows: in, N is the order of the system, T 98％ The time required for the system response curve to first enter and remain at 98% of its final value.

7. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technique according to claim 2 is characterized in that: The method described in step 5 further includes estimating the unknown parameters in the system based on online parameter estimation and in combination with the attitude angle change rate tracking error. The method for estimating the unknown parameter d is as follows: Where W(t) is the signal matrix, It is an estimate of the unknown parameter d, describing the relationship between the system and the unknown parameter. It can be obtained from the measurement of the system signal. The update rule of the gain P is as follows: Among them, λ0 and k0 are positive constants, representing the preset boundaries of the maximum forgetting rate and the gain matrix P, respectively.

8. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technique according to claim 2 is characterized in that: The active disturbance rejection control module described in step 6 uses three-channel decoupling-free control, eliminating the need to decouple the coupled part, which reduces the difficulty of controller design. The decoupler-free design is as follows: in, 9. The decoupling-free attitude control method for a quadrotor drone based on an improved active disturbance rejection technique according to claim 2 is characterized in that: The method described in step 7 also includes solving the actual control variable, that is, the rotor speed, under the condition of balanced forces in the height direction, as follows: Among them, U L For along O b Z b Total lift on the axis, τ φ , τ θ and τ ψ To circle O b X b ,O b Y b ,O b Z b The torque of the shaft, k L is the lift coefficient, b is the anti-torque coefficient, l is the distance from the rotor center to the center of mass of the UAV, and ω1, ω2, ω3, and ω4 are the rotational speeds of the four rotors.

Images