Quadrotor UAV attitude control method based on improved fuzzy anti-disturbance control

By improving the attitude control method of fuzzy self-immunity, the suppression problem of four-rotor drones under multi-source disturbance is solved, the robustness and anti-interference are improved, and more efficient attitude control effect is achieved.

CN118747004BActive Publication Date: 2024-11-15XUZHOU NORMAL UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411031317.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-15
Estimated Expiration
2044-07-30

AI Technical Summary

Technical Problem

Existing quadrotor UAVs are difficult to achieve effective suppression in complex environments under multi-source disturbance, and their robustness is insufficient in a stable state.

Method used

Using an attitude control method with improved fuzzy self-immunity, the design of a cascade PID controller and an improved expansion state observer are achieved by establishing a mathematical model of multi-source perturbation, combining nonlinear functions and fuzzy control, and suppressing and pose of multi-source perturbation Accurate control of angles.

Benefits of technology

It improves the anti-interference and robustness of the steady-state process of the quadrotor drone under multi-source disturbance, enhances the ability to suppress multi-source disturbance, and improves the observation performance and control effect.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118747004B_ABST
    Figure CN118747004B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for attitude control of a quad-rotor unmanned aerial vehicle based on improved fuzzy self-disturbance rejection, and the steps include: establishing a mathematical model of a quad-rotor unmanned aerial vehicle with multi-source disturbances; designing a position loop controller; designing a nonlinear function; designing an improved extended state observer; improving ESO convergence condition analysis; and designing an attitude loop controller. The present invention realizes parameter self-tuning by designing an improved nonlinear function in the extended state observer and adding fuzzy control; designs a cascade PID control in the position loop to realize position tracking control; and designs an improved fuzzy self-disturbance rejection controller in the attitude loop to realize attitude control. The present invention improves the observation performance of the observer, thereby improving the controller's ability to suppress multi-source disturbances, making the steady-state process of the quad-rotor unmanned aerial vehicle more robust, and improving the anti-interference performance of the quad-rotor unmanned aerial vehicle when working in a complex environment under multi-source disturbances.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a quad-rotor unmanned aerial vehicle attitude control method based on improved fuzzy self-anti-disturbance control, and belongs to the technical field of unmanned aerial vehicle intelligent control. Background Art

[0002] A drone is an unmanned aircraft controlled by a radio remote control device and a self-contained program control device. It is powered by four independently controlled DC brushless motors driving propellers, and the flight control of the drone can be achieved by adjusting the speed. In recent years, with the rapid development of micromechanical sensors, composite materials, visual positioning and navigation technologies, drones have gradually become more intelligent and lightweight. Due to their versatility and ability to perform tasks autonomously or remotely, drones have obvious advantages in operating in dangerous and complex environments.

[0003] In order to achieve precise control of the attitude of quad-rotor drones, the current mainstream control methods include proportional integral differential control, sliding mode control, model predictive control, backstepping control, etc. Quad-rotor drones are highly complex nonlinear systems, and their accurate models are difficult to obtain in practice. For example, PID control is widely used in engineering and does not rely on system models. However, in the face of multi-source interference such as unknown parameter perturbations, sensor noise, airflow and wind disturbances, the control effect of quad-rotor drones is not ideal.

[0004] Therefore, how to achieve the ability of a quadrotor drone to suppress multi-source interference and have high robustness in a stable state when working in a complex environment with multi-source disturbances is still a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the problems existing in the above-mentioned prior art, the present invention provides a quadrotor UAV attitude control method based on improved fuzzy self-interference rejection, which enables the quadrotor UAV to suppress multi-source interference when working in a complex environment under multi-source disturbance and has higher robustness in a stable state.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: a quadrotor drone attitude control method based on improved fuzzy anti-disturbance control, comprising the following steps:

[0007] S1. Establish a mathematical model of a quadrotor drone with multi-source disturbances. The steps are as follows:

[0008] S1.1. Establish the ground coordinate system E(O) based on the three-dimensional space of the earth E ,X E ,Y E ,Z E ) and the body coordinate system B(O based on the quadrotor drone B ,XB ,Y B ,Z B ); In the ground coordinate system, three Euler angles [φ, θ, ψ] are defined, corresponding to the roll angle, pitch angle and yaw angle respectively; among them, the value range of the roll angle φ and the pitch angle θ are both (-0.5π, 0.5π), and the value range of the yaw angle ψ is (-π, π);

[0009] S1.2. According to the Newton-Euler equation, the dynamic model of the quadrotor drone is established as follows:

[0010]

[0011] Where m is the total mass of the quadrotor drone; g is the acceleration of gravity; K1, K2, ..., K6 are the drag coefficients; d1, d2, ..., d6 are the multi-source total disturbances including wind disturbance, sensor noise and parameter perturbation; l is the distance from the rotor tip to the center of gravity of the quadrotor drone; I x ,I y ,I z Indicates the moment of inertia of the three axes; u1, u2, u3, u4 indicate control inputs:

[0012]

[0013] Where F1, F2, F3, and F4 are the thrusts generated by the four rotors, which are regarded as the actual control inputs of the system;

[0014] S1.3. Introduction of virtual control quantity u into the position subsystem x ,u y ,u z They are:

[0015]

[0016] Formula (1) can be rearranged as:

[0017]

[0018] In the formula, X1=[x,y,z,φ,θ,ψ] T ,

[0019] U=[u x ,u y ,u z ,u2,u3,u4] T , D=[d1,d2,d3,d4,d5,d6] Τ , and D has a known upper limit;

[0020] S1.4. According to equation (3), the expected pitch angle θ is inversely solved d and the desired roll angle φ d :

[0021]

[0022] S2. Design the position loop controller. The steps are as follows:

[0023] S2.1. Design a horizontal channel cascade PID controller. The subsystem is divided into an inner loop and an outer loop. The outer loop uses the position error as input and adopts a proportional link. The designed controller is:

[0024]

[0025] In the formula, k 1px , k 1py is the proportional link coefficient, e 1d 、e 2d is the expected position x d ,y d and the error between the actual position x and y;

[0026] The inner loop takes the output of the outer loop as input. This link uses proportional-integral-differential control. The designed controller is:

[0027]

[0028] In the formula, K 2p , K 2i , K 2d They are the constant coefficient matrices of the proportional link, integral link, and differential link respectively.

[0029] Substituting formula (6) into formula (1), we get:

[0030]

[0031] S2.2. Design a high-pass cascade PID controller. The subsystem is divided into an inner loop and an outer loop. The outer loop uses the position error as input and adopts a proportional link. The controller is designed as follows:

[0032]

[0033] Substituting (9) into (1) we obtain:

[0034]

[0035] S3. Design a nonlinear function. The steps are as follows:

[0036] According to the principle of nonlinear function selection, the nonlinear function is designed as follows:

[0037]

[0038] In the formula, δ>0, a>0 is the adjustment factor, which corresponds to the adjustment factor in the original nonlinear function one by one, and has the same effect. When the input error |e|≤1, if only the inverse tangent function is used, the gain of the small error interval will be small. To solve this problem, a sine function is added when |e|≤1 to increase the error gain; when the error |e|>1, in order to ensure that the error gain of this interval is less than the gain of the original fal function, the narcfal function is defined as a constant, and to ensure the continuity of the function, the constant value is set equal to the value of the narcfal function when |e|=1;

[0039] S3.2. Take the error gain of fal function and narcfal function when e>1 as:

[0040]

[0041] The unknown terms in g2(e)·e are considered as independent terms, and their maximum values ​​are:

[0042]

[0043] Then: g2(e)·e<Max[g2(e)·e]<1(14)

[0044] In summary, when e>1, g2(e)·e<1<g1(e)·e always holds;

[0045] S4. Design and improve the extended state observer. The specific steps are as follows:

[0046] S4.1. Add fuzzy control to realize ESO self-adaptation: define the input as error signal e1 and error differential signal e2, and the output as ESO gain parameter Δβ 01 , Δβ 02 , Δβ 03 ; Define the basic domain of input quantity as [-50,50], and the basic domain of output quantity as [-100,100], [-100,100], [-30,30] respectively; The language subset is defined as: negative big (NB), negative middle (NM), negative small (NS), positive big (PB), positive middle (PM), positive small (PS); The fuzzy reasoning is:

[0047]

[0048] And establish the fuzzy rule table:

[0049] Δβ 01 , Δβ 02 , Δβ 03 Fuzzy rule table

[0050]

[0051] S4.2. Based on S3 and S4.1, establish an improved ESO as follows:

[0052]

[0053] In the formula, e1 is the error between the observed value and the actual value, β 0i +Δβ 0i is the gain parameter of the improved ESO; let β 0i +Δβ 0i =P i , and further obtain the observation error:

[0054]

[0055] S4.3. According to S4.2, the matrix equation is established as:

[0056]

[0057] Where, F1 = narcfal1 / e1, F2 = narcfal2 / e1;

[0058] S5. Improve the ESO convergence condition analysis, the steps are as follows:

[0059] S5.1. Establish the convergence analysis matrix HA(e) as:

[0060]

[0061] Where H is a matrix with positive diagonals, defined as:

[0062]

[0063]

[0064] F is a bounded quantity, and its range is:

[0065]

[0066] S5.2. There exists H such that HA(e) is a positive definite symmetric matrix and the zero solution of (17) is Lyapunov asymptotically stable. Then HA(e) needs to satisfy the following conditions:

[0067]

[0068] Let h 11 =1,h 22 =h 33=λ, λ is a positive number infinitely close to 0. Combining equation (23) and equation (24), the convergence condition is:

[0069] P1P2>P3 (25);

[0070] S6. Design the attitude loop controller. The steps are as follows:

[0071] S6.1. Attitude system decoupling processing:

[0072] Since the attitude angles of the quadrotor drone are coupled with each other, the coupled part of the quadrotor drone is regarded as a disturbance, the system model is simplified, and the state space equation of the system is established:

[0073]

[0074] in,

[0075]

[0076] S6.2. Design of tracking differentiator:

[0077] Active disturbance rejection control establishes the fastest feedback system based on the fastest comprehensive function fhan(·) and constructs a tracking differentiator:

[0078]

[0079] The fhan function is:

[0080]

[0081] Where v is the set reference input, v1 is the tracking output signal, v2 is the differential signal, r is the fast factor parameter; h is the filter factor.

[0082] S6.3. Design and improve the extended state observer:

[0083] In formula (27), [ω φ ,ω θ ,ω ψ ] T Expand to new state variables:

[0084]

[0085] Considering the high symmetry of the quadrotor UAV system model, the controller designs of the three attitude angles are consistent, so the pitch channel is selected to establish an improved extended state observer consistent with formula (16);

[0086] S6.4. Design of nonlinear state error feedback control law:

[0087] The errors of ESO and TD are nonlinearly combined, and the specific expression is:

[0088] e1=v1-z1

[0089] e2=v2-z2

[0090] u0=β1fal(e1,a1,δ)+β2fal(e2,a2,δ) (32)

[0091] u= u0-z3

[0092] b0

[0093] In the formula, e1 and e2 are the error signal and error differential signal respectively; b0 is the compensation factor.

[0094] The present invention realizes parameter self-tuning by designing and improving nonlinear functions in the extended state observer and adding fuzzy control; designs cascade PID control in the position loop to realize position tracking control; and designs and improves the fuzzy anti-disturbance controller in the attitude loop to realize attitude control. The present invention improves the observation performance of the observer, thereby improving the controller's ability to suppress multi-source disturbances, making the steady-state process of the quad-rotor drone more robust, and improving the anti-interference performance of the quad-rotor drone when working in a complex environment under multi-source disturbances. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 : is a coordinate system relationship diagram of a quad-rotor drone according to an embodiment of the present invention;

[0096] Figure 2 It is a comparison diagram of the cascade PID control and expected position response curves according to an embodiment of the present invention;

[0097] Figure 3 is a comparison diagram of observation error curves of the improved ESO and the traditional ESO according to the embodiment of the present invention;

[0098] Figure 4 It is a comparison of the response curves of the attitude angle of the improved ADRC in the embodiment of the present invention and the traditional ADRC. DETAILED DESCRIPTION

[0099] The present invention will be further described below in conjunction with the accompanying drawings.

[0100] Embodiment: In order to verify the effectiveness of the posture control method of the present invention, Figure 1 In the quadrotor UAV system model shown, the improved ESO proposed in the present invention is first simulated and compared with the traditional ESO to verify whether the observation error performance is improved; then the improved active disturbance rejection controller (ADRC) proposed in the present invention in the attitude loop is simulated and verified with the traditional active disturbance rejection controller.

[0101] The present invention builds a simulation model in the Matlab / Simulink environment to simulate the trajectory tracking of three attitude angles and three positions of the quad-rotor UAV system. The relevant parameter settings are shown in Tables 1 to 4:

[0102] Table 1 Quadrotor UAV model parameters

[0103]

[0104] Table 2 Cascade PID simulation parameters

[0105]

[0106] Table 3 Simulation comparison parameters of improved ESO and traditional ESO

[0107]

[0108] Table 4 Simulation comparison parameters between improved ADRC and traditional ADRC

[0109]

[0110] The expected positions of the quadrotor drone are [6m, 3m, 10m], and the corresponding time is [5s, 3s, 0s]; the expected value of the yaw angle is 20°, and the corresponding time is 2s.

[0111] like Figure 2 The figure shows a comparison of the response curves based on cascade PID control and the expected position. The red color is the expected position and the blue color is the actual position. It can be seen from the figure that the position curve starts to respond at a given time and eventually reaches the expected position.

[0112] like Figure 3 The figure shows the comparison of the observed error curves of the improved ESO of the present invention and the traditional ESO, where the blue one is the error curve after improvement and the green one is the error curve before improvement. Overall, the observed errors of the two can converge to 0. Compared with the traditional ESO, the error amplitude of the improved ESO of the present invention is much lower, and the amplitude fluctuation is smaller than that of the traditional ESO;

[0113] like Figure 4 The figure shows the response curve comparison of the attitude angle of the improved ADRC of the present invention and the traditional ADRC. The red color is the expected attitude angle, the blue color is the response curve of the improved ADRC, and the green color is the response curve of the ADRC before the improvement. Overall, the ADRC before and after the improvement can suppress the influence of disturbances and achieve good attitude control effects. In comparison, the overshoot of the improved ADRC of the present invention is slightly lower than that of the traditional ADRC, and it has a faster response speed. And compared with the traditional ADRC, the improved ADRC of the present invention has a better suppression effect on interference.

Claims

1. A quadrotor drone attitude control method based on improved fuzzy anti-disturbance control, characterized in that: The steps include: S1. Establish a mathematical model of a quadrotor drone with multi-source disturbances. The steps are as follows: S1.

1. Establish the ground coordinate system E(O) based on the three-dimensional space of the earth E ,X E ,Y E ,Z E ) and the body coordinate system B(O based on the quadrotor drone B ,X B ,Y B ,Z B ); In the ground coordinate system, three Euler angles [φ, θ, ψ] are defined, corresponding to the roll angle, pitch angle and yaw angle respectively; among them, the value range of the roll angle φ and the pitch angle θ are both (-0.5π, 0.5π), and the value range of the yaw angle ψ is (-π, π); S1.

2. According to the Newton-Euler equation, the dynamic model of the quadrotor drone is established as follows: Where m is the total mass of the quadrotor drone; g is the acceleration of gravity; K1, K2, ..., K6 are the drag coefficients; d1, d2, ..., d6 are the multi-source total disturbances including wind disturbance, sensor noise and parameter perturbation; l is the distance from the rotor tip to the center of gravity of the quadrotor drone; I x ,I y ,I z Indicates the moment of inertia of the three axes; u1, u2, u3, u4 indicate control inputs: Where F1, F2, F3, and F4 are the thrusts generated by the four rotors, which are regarded as the actual control inputs of the system; S1.

3. Introduction of virtual control quantity u into the position subsystem x ,u y ,u z They are: Formula (1) can be rearranged as: In the formula, X1=[x,y,z,φ,θ,ψ] T , U=[u x ,u y ,u z ,u2,u3,u4] T , D=[d1,d2,d3,d4,d5,d6] Τ , and D has a known upper limit; S1.

4. According to equation (3), the expected pitch angle θ is inversely solved d and the desired roll angle φ d : S2. Design the position loop controller. The steps are as follows: S2.

1. Design a horizontal channel cascade PID controller. The subsystem is divided into an inner loop and an outer loop. The outer loop uses the position error as input and adopts a proportional link. The designed controller is: In the formula, k 1px , k 1py is the proportional link coefficient, e 1d 、e 2d is the expected position x d ,y d and the error between the actual position x and y; The inner loop takes the output of the outer loop as input. This link uses proportional-integral-differential control. The designed controller is: In the formula, K 2p , K 2i , K 2d They are the constant coefficient matrices of the proportional link, integral link, and differential link respectively. Substituting formula (7) into formula (1), we get: S2.

2. Design a high-pass cascade PID controller. The subsystem is divided into an inner loop and an outer loop. The outer loop uses the position error as input and adopts a proportional link. The controller is designed as follows: Substituting (9) into (1) we obtain: S3. Design a nonlinear function. The steps are as follows: According to the principle of nonlinear function selection, the nonlinear function is designed as follows: In the formula, δ>0, a>0 is the adjustment factor, which corresponds one-to-one with the adjustment factor in the original nonlinear function; S3.

2. Take the error gain of fal function and narcfal function when e>1 as: The unknown terms in g2(e)·e are considered as independent terms, and their maximum values ​​are: Then: g2(e)·e<Max[g2(e)·e]<1(14) In summary, when e>1, g2(e)·e<1<g1(e)·e always holds; S4. Design and improve the extended state observer. The specific steps are as follows: S4.

1. Add fuzzy control to realize ESO self-adaptation: define the input as error signal e1 and error differential signal e2, and the output as ESO gain parameter Δβ 01 , Δβ 02 , Δβ 03 ; Define the basic domain of input quantity as [-50,50], and the basic domain of output quantity as [-100,100], [-100,100], [-30,30] respectively; The language subset is defined as: negative large, negative medium, negative small, positive large, positive medium, positive small; Fuzzy reasoning is: And establish a fuzzy rule table; S4.

2. Based on S3 and S4.1, establish an improved ESO as follows: In the formula, e1 is the error between the observed value and the actual value, β 0i +Δβ 0i is the gain parameter of the improved ESO; let β 0i +Δβ 0i =P i , and further obtain the observation error: S4.

3. According to S4.2, the matrix equation is established as: Where, F1 = narcfal1 / e1, F2 = narcfal2 / e1; S5. Improve the ESO convergence condition analysis, the steps are as follows: S5.

1. Establish the convergence analysis matrix HA(e) as: Where H is a matrix with positive diagonals, defined as: F is a bounded quantity, and its range is: S5.

2. There exists H such that HA(e) is a positive definite symmetric matrix and the zero solution of (17) is Lyapunov asymptotically stable. Then HA(e) needs to satisfy the following conditions: Let h 11 =1,h 22 =h 33 =λ, λ is a positive number infinitely close to 0. Combining equation (23) and equation (24), the convergence condition is: P1P2>P3 (25); S6. Design the attitude loop controller. The steps are as follows: S6.

1. Attitude system decoupling processing: Establish the state-space equations of the system: in, S6.

2. Design of tracking differentiator: Active disturbance rejection control establishes the fastest feedback system based on the fastest comprehensive function fhan(·) and constructs a tracking differentiator: The fhan function is: In the formula, v is the set reference input, v1 is the tracking output signal, v2 is the differential signal, r is the fast factor parameter, and h is the filter factor; S6.

3. Design and improve the extended state observer: In formula (27), [ω φ ,ω θ ,ω ψ ] T Expand to new state variables: Selecting the pitch channel to establish the improved extended state observer is consistent with equation (16); S6.

4. Design of nonlinear state error feedback control law: The errors of ESO and TD are nonlinearly combined, and the specific expression is: In the formula, e1 and e2 are the error signal and error differential signal respectively; b0 is the compensation factor.