A method for quickly locating actuator faults of quadrotor drones

Abstract

The present invention discloses a method for quickly locating a fault of a quad-rotor UAV actuator, which relates to the technical field of multi-rotor aircraft fault diagnosis and solves the technical problem that the fault location of a quad-rotor UAV actuator is not accurate enough. The key points of the technical solution are as follows: first, a fault detection observer is designed based on a linear variable parameter model of a quad-rotor UAV; then, a residual logic table for locating the fault is established by combining the motion characteristics of the quad-rotor UAV and the fault matrix; then, a robust L ∞ 、H ‑ The performance indicators and pole configuration conditions are used to solve the gain of the fault detection observer, so that the residual signal generated by the observer is sensitive to faults but insensitive to interference. Finally, the evaluation function value is calculated based on the residual signal. If the evaluation function value exceeds the threshold, the system is judged to have a fault. At the same time, the residual signal is compared with the fault location residual logic table to quickly locate the fault location.

Description

Technical Field

[0001] The present application relates to the technical field of multi-rotor aircraft fault diagnosis, and in particular to a method for quickly locating a fault of a quad-rotor UAV actuator. Background Art

[0002] Quadcopter drones, with their vertical takeoff and landing, fixed-point hovering, and high maneuverability, have excelled in a variety of fields, including aerial photography, urban patrols, and power inspections. However, quadcopter drones are also underactuated and tightly coupled systems. A failure in their actuators can lead to performance degradation, resulting in mission failure, or even instability and crashes. Therefore, ensuring the safety and reliability of quadcopter drones is a topic worthy of in-depth research.

[0003] The high safety and reliability requirements make fault detection in quadrotor drones a critical issue. For quadrotor drone systems, if faults can be quickly and accurately detected and located, timely countermeasures can be taken, thereby avoiding accidents and reducing losses to a certain extent. Current fault detection methods primarily include model-based, knowledge-based, and data-based approaches. Knowledge-based approaches, such as fault tree analysis and symbolic directed graphs, leverage the system's knowledge base to learn fault types. Data-based approaches establish fault models through reinforcement learning and machine learning, but the results lack interpretability. Model-based approaches combine analysis and modeling of system characteristics to diagnose faults by analyzing the differences between observed and actual signals.

[0004] Model-based methods primarily locate and detect faults by designing observers. The mainstream fault location strategy involves designing multi-channel fault detection observers, which increases location complexity and hinders rapid fault location. Furthermore, factors such as measurement noise and external interference can affect the performance of observer-based fault location methods, leading to false alarms. Therefore, further research is needed on fault location techniques for quadrotor UAV systems. Summary of the Invention

[0005] The present application provides a method for quickly locating actuator faults of a quad-rotor UAV, the technical purpose of which is to achieve rapid and accurate actuator fault detection and positioning.

[0006] The above technical objectives of this application are achieved through the following technical solutions:

[0007] A method for quickly locating a fault in a quadrotor UAV actuator comprises:

[0008] S1: Analyze the motion characteristics of the quadrotor UAV system to construct a linear variable parameter model of the quadrotor UAV system;

[0009] S2: Design the observer for fault detection according to the linear variable parameter model, and obtain the error dynamic equation of the quadrotor UAV system according to the linear variable parameter model. According to the error dynamic equation, combined with the interference robustness index L ∞ , Fault sensitivity index H - Solve the observer gain L(α) using the pole placement conditions;

[0010] S3: generating a residual signal according to the measurement output and the observation output of the observer, designing a state evaluation function according to the residual signal, and calculating the value of the state evaluation function in real time according to the residual signal;

[0011] S4: The influence of the actuator fault on the residual signal is analyzed through the error dynamic equation, and the residual logic table for locating the actuator fault is designed in combination with the fault matrix; S5: The detection threshold is set according to the known noise and interference conditions, and the value of the state evaluation function is compared with the detection threshold. If the value of the state evaluation function does not exceed the detection threshold, the quadrotor UAV actuator state is normal; if the value of the state evaluation function exceeds the detection threshold, the quadrotor UAV actuator fails, and then the generated residual signal is compared with the residual logic table to locate the fault position.

[0012] Furthermore, step S1 includes: considering the actual working conditions and combining the small angle assumption, selecting the premise variable ρ, and designing the membership function α i (ρ), combined with the analysis of motion characteristics, the linear variable parameter model of the quadrotor UAV system is expressed as:

[0013]

[0014] y(t)=Cx(t)+Eω(t);

[0015] Among them, ρ represents the premise variable; α i (ρ) represents the membership function; A i represents the state matrix, i = 1, 2, 3, 4; S represents the input matrix; F represents the fault matrix; B d represents the interference matrix; C represents the output matrix; E represents the noise matrix; x(t) represents the state; u(t) represents the control input; f(t) represents the actuator fault of the quadrotor drone; d(t) represents the interference; y(t) represents the measured output of the system; ω(t) represents the measurement noise.

[0016] Furthermore, in step S2, the observer is expressed as:

[0017]

[0018] in, Represents the state value of the observation; represents the gain of the fault detection observer to be solved; r represents the residual value of the system;

[0019] The error dynamic equation is expressed as:

[0020]

[0021] Among them, A C (α)=A(α)-L(α)C,D e =[B d 0], D r =[0 E].

[0022] Furthermore, in step S3, the measured output value y(t) and the observed output value y(t) of the quadrotor UAV system are used to calculate the output value y(t). For the residual signal r = [r1, r2, r3, r4] T Calculate and express it as: Among them, r1, r2, r3, and r4 correspond to the four state channel signals x1, x2, x3, and x4 respectively, and the state signal x = [x1, x2, x3, x4] T , x1, x2, x3, and x4 represent the z-axis velocity, pitch angular velocity, roll angular velocity, and yaw angular velocity of the quadrotor UAV system, respectively;

[0023] Then the state evaluation function is expressed as:

[0024] Furthermore, in step S4, the fault matrix F is expressed as:

[0025]

[0026] Among them, c 11 、c 21 、c 31 、c 41 Represents the impact of the failure of the first actuator on r1, r2, r3, and r4 respectively; c 12 、c 22 、c 32 、c 42 Represents the impact of the second actuator failure on r1, r2, r3, and r4 respectively; c 13 、c 23 、c 33 、c 43 Represents the impact of the third actuator failure on r1, r2, r3, and r4 respectively; c 14 、c 24 、c 34 、c 44They represent the impact of the fourth actuator failure on r1, r2, r3, and r4 respectively;

[0027] According to the fault matrix, the corresponding symbol matrix sign(F) is expressed as:

[0028]

[0029] According to the sign matrix sign(F), a residual logic table between single motor fault and positive and negative residual values ​​is obtained. Further, in step S5, the detection threshold is expressed as:

[0030] R th =f(||d|| ∞ ,||w|| ∞ , t);

[0031] Among them, ||·|| ∞ represents the ∞ norm of the vector; R th Represents a function related to interference d, noise w and time t.

[0032] The beneficial effect of the present application is that the method for quickly locating the actuator fault of the quadrotor UAV described in the present application utilizes the fault sensitivity index H when solving the observer gain. - , interference robustness index L ∞ The proposed method combines the pole configuration conditions to reduce the impact of external interference and noise on fault detection while enhancing the sensitivity of the residual signal to fault information. This allows the residual signal to change dramatically at the moment a fault occurs, reducing the false alarm rate and improving the accuracy of actuator fault detection in quadrotor systems. The fault detection structure is simple, eliminating the need for a multi-channel observer. Instead, a unified fault detection observer is required to generate the residual signal. The fault matrix is ​​then established by combining the motion principles of the quadrotor system with analysis of the error dynamics system. By analyzing the correspondence between the residual signal symbols and the fault matrix symbols when a fault occurs, fault location can be quickly achieved at the moment detection is completed. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of the structure of an "X" type quadrotor drone;

[0034] Figure 2 This is a flow chart of a method for quickly locating a fault in a quadrotor drone actuator according to an embodiment of the present application;

[0035] Figure 3 Schematic diagram of the detection threshold and state evaluation function curve for a quadrotor drone with a random single motor failure in an embodiment of the present application;

[0036] Figure 4This is a schematic diagram of the residual signal curve under random single motor failure of the quadrotor drone in an embodiment of the present application. DETAILED DESCRIPTION

[0037] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0038] In the embodiment of the present application, the structure of the quad-rotor drone is as follows Figure 1 shown.

[0039] Figure 2 This is a flow chart of a method for quickly locating a fault in a quadrotor drone actuator according to an embodiment of the present application. The method includes:

[0040] S1: Analyze the motion characteristics of the quadrotor UAV system to construct a linear variable parameter model of the quadrotor UAV system.

[0041] Specifically, the motion characteristics of the quadrotor UAV system are analyzed, which is specifically realized through the dynamic model. The dynamic model of the quadrotor UAV system is expressed as:

[0042]

[0043] Where m represents the total mass of the quadrotor drone, g represents the acceleration of gravity, and J x , J y , J z Respectively represent the moment of inertia around the x, y, and z axes of the body coordinate system, x, y, and z represent the position coordinates of the quadrotor drone, φ, θ, and ψ represent the roll angle, pitch angle, and yaw angle of the drone, respectively, T represents the total lift generated by the four propellers, and Γ φ , Γ θ , Γ ψ They represent the rolling moment, pitching moment and yaw moment acting on the aircraft respectively.

[0044] The state variables and control inputs are set as:

[0045] u=[u z u φ u θ u ψ ] T =[T-mg Γ φ Γ θ Γ ψ ] T The state space model of the quadrotor UAV system is expressed as:

[0046] x(t)=A(x)x+Bu+Ff+B d d;

[0047] y=Cx+Eω

[0048] The state space model is linearized and converted into a linear variable parameter (LPV) model. Considering that the pitch angular velocity and roll angular velocity are bounded under general working conditions, the premise variables are selected as Design membership function α i (ρ), and combined with the analysis of motion characteristics, the linear variable parameter model of the quadrotor UAV system is constructed as follows:

[0049]

[0050] y(t)=Cx(t)+Eω(t);

[0051] Among them, ρ represents the premise variable; α i (ρ) represents the membership function; A i represents the state matrix, i = 1, 2, 3, 4; B represents the input matrix; F represents the fault matrix; B d represents the interference matrix; C represents the output matrix; E represents the noise matrix; x(t) represents the state; u(t) represents the control input; f(t) represents the actuator fault of the quadrotor drone; d(t) represents the interference, d = [d1, d2, d3, d4] T ; y(t) represents the measured output of the system; ω(t) represents the measurement noise.

[0052] In the embodiment of the present application, the membership function α i The specific form of (ρ) is as follows:

[0053]

[0054] S2: Design the observer for fault detection according to the linear variable parameter model, and obtain the error dynamic equation of the quadrotor UAV system according to the linear variable parameter model. According to the error dynamic equation, combined with the interference robustness index L ∞ , Fault sensitivity index H - The observer gain L(α) is solved using the pole placement conditions.

[0055] Specifically, the observer is expressed as:

[0056]

[0057] in, Represents the state value of the observation; represents the gain of the fault detection observer to be solved; r represents the residual value of the system.

[0058] Defining new aggregate interference Defining state error The system state error dynamic equation can be obtained as follows:

[0059]

[0060] Among them, A C (α)=A(α)-L(α)C,D e =[B d 0], D r =[0 E].

[0061] In the absence of faults, f=0, Selecting a suitable observer gain L(α) makes the system stable, that is, When a single motor fault exists, Ff will affect the changing trend of e and also the change of the residual signal r. Therefore, the fault matrix F and the change of the residual signal r can be combined to locate the system fault.

[0062] For the error dynamic equation established, use H - Performance, L ∞ The performance and pole configuration conditions make the fault detection observer fault-sensitive and disturbance-robust. - The performance index enhances the sensitivity of the residual value to the fault signal, making Application L ∞ The performance index weakens the influence of interference and noise signals on the residual value; the pole configuration condition is used to improve the transient response of the fault detection observer.

[0063] Comprehensively consider the fault sensitivity index H - , interference robustness index L ∞ and pole placement conditions, the gain problem of the fault detection observer is transformed into a feasibility problem of linear matrix inequality, and the gain of the observer is obtained with the help of the matrix toolbox.

[0064] S3: generating a residual signal according to the measurement output and the observation output of the observer, designing a state evaluation function according to the residual signal, and calculating the value of the state evaluation function in real time according to the residual signal.

[0065] Specifically, the measured output value y(t) and the observed output value of the quadrotor UAV system are For the residual signal r = [r1, r2, r3, r4] T Calculate and express it as: Among them, r1, r2, r3, and r4 correspond to the four state channel signals x1, x2, x3, and x4 respectively, and the state signal x = [x1, x2, x3, x4] T, x1, x2, x3, and x4 represent the z-axis velocity, pitch angular velocity, roll angular velocity, and yaw angular velocity of the quadrotor UAV system, respectively;

[0066] Then the state evaluation function is expressed as:

[0067] S4: Analyze the influence of the actuator fault on the residual signal through the error dynamic equation, and design a residual logic table for locating the actuator fault in combination with the fault matrix.

[0068] Specifically, Figure 1 Taking the "X" type quadrotor drone as an example, the speeds of the four propeller motors are ω1, ω2, ω3, and ω4 respectively. The relationship between the propeller lift and the motor speed is b>0 indicates the lift coefficient of rotor rotation; the relationship between the counter torque and the motor speed is c M >0 indicates the anti-torque coefficient of rotor rotation, L roll is the distance between the tumbling motors, L pitch is the distance between the pitch motors, so the fault matrix F is expressed as:

[0069]

[0070] The symbol matrix sign(F) corresponding to the transformation matrix is ​​expressed as:

[0071]

[0072] According to the sign matrix sign(F), a residual logic table corresponding to single motor faults and positive and negative residual values ​​is obtained as shown in Table 1.

[0073] Table 1 Residual logic table for single motor fault location of “X” type quadrotor UAV

[0074]

[0075] Here, δ represents the absolute value of the additive fault.

[0076] S5: Set the detection threshold according to the known noise and interference conditions, and compare the value of the state evaluation function with the detection threshold. If the value of the state evaluation function does not exceed the detection threshold, the quadrotor UAV actuator is in normal state; if the value of the state evaluation function exceeds the detection threshold, the quadrotor UAV actuator fails. Then compare the generated residual signal with the residual logic table to locate the fault position.

[0077] Specifically, the detection threshold is expressed as:

[0078] R th=f(||d|| ∞ ,||w|| ∞ , t);

[0079] Among them, ||·|| ∞ represents the ∞ norm of the vector; R th Represents a function related to interference d, noise w and time t.

[0080] The obtained state evaluation function value is compared with the set detection threshold. If the value of the state evaluation function J(t) exceeds the set detection threshold R th , that is, J(t)>R th , it is determined that the quadcopter execution system has a fault. Figure 3 As shown in Figure 2, after 30 seconds, the value of the state evaluation function rises rapidly and exceeds the set detection threshold after 30.3 seconds, indicating that the method of the present invention successfully detects the actuator fault 0.3 seconds after the fault occurs, and then Figure 4 Comparing the sign of the residual signal after 30.3 seconds with that in Table 1, it can be found that its sign corresponds to the fault condition of motor 4, and it can be determined that motor 4 has a fault.

[0081] The above experiments verified the accuracy and speed of the quadrotor drone fault detection and positioning method of the present invention.

Claims

1. A method for quickly locating a quadrotor UAV actuator fault, characterized in that: include: S1: Analyze the motion characteristics of the quadrotor UAV system to construct a linear variable parameter model of the quadrotor UAV system; S2: Design the observer for fault detection according to the linear variable parameter model, and obtain the error dynamic equation of the quadrotor UAV system according to the linear variable parameter model. According to the error dynamic equation, combined with the interference robustness index L ∞ , Fault sensitivity index H - Solve the observer gain L(α) using the pole placement conditions; S3: generating a residual signal according to the measurement output and the observation output of the observer, designing a state evaluation function according to the residual signal, and calculating the value of the state evaluation function in real time according to the residual signal; S4: Analyze the influence of the actuator fault on the residual signal through the error dynamic equation, and design a residual logic table for locating the actuator fault in combination with the fault matrix; S5: Set the detection threshold according to the known noise and interference conditions, and compare the value of the state evaluation function with the detection threshold. If the value of the state evaluation function does not exceed the detection threshold, the quadrotor UAV actuator is in normal state; if the value of the state evaluation function exceeds the detection threshold, the quadrotor UAV actuator fails. Then compare the generated residual signal with the residual logic table to locate the fault position.

2. The method for quickly locating a quadrotor UAV actuator fault according to claim 1, wherein: Step S1 includes: considering the actual working conditions and combining the small angle assumption, selecting the premise variable ρ, and designing the membership function α i (ρ), combined with the analysis of motion characteristics, the linear variable parameter model of the quadrotor UAV system is expressed as: y(t)=Cx(t)+Eω(t); Among them, ρ represents the premise variable; α i (ρ) represents the membership function; A i represents the state matrix, i=1,2,3,4; B represents the input matrix; F represents the fault matrix; B d Represents the interference matrix; C represents the output matrix; E represents the noise matrix; x(t) represents the true state value; u(t) represents the control input; f(t) represents the actuator fault of the quadrotor drone; d(t) represents the interference; y(t) represents the measured output of the system; ω(t) represents the measurement noise.

3. The method for quickly locating a quadrotor UAV actuator fault according to claim 2, wherein: In step S2, the observer is expressed as: in, Represents the state value of the observation; represents the gain of the fault detection observer to be solved, L i represents the gain of the i-th fault detection observer to be solved; r(t) represents the residual value of the system; represents the estimated value of y(t); The error dynamic equation is expressed as: Among them, A C (a)=A(a)-L(a)C,D e =[B d 0],D r = [0E].

4. The method for quickly locating a quadrotor UAV actuator fault according to claim 3, wherein: In step S3, the measured output value y(t) and the observed output value of the quadrotor UAV system are For the residual signal r = [r1, r2, r3, r4] T Calculate and express it as: Among them, r1, r2, r3, and r4 correspond to the four state channel signals x1, x2, x3, and x4 respectively, and the state signal x = [x1, x2, x3, x4] T , x1, x2, x3, and x4 represent the z-axis velocity, pitch angular velocity, roll angular velocity, and yaw angular velocity of the quadrotor UAV system, respectively; Then the state evaluation function is expressed as:

5. The method for quickly locating a quadrotor UAV actuator fault according to claim 4, wherein: In step S4, the fault matrix F is expressed as: Among them, c 11 、c 21 、c 31 、c 41 Represents the impact of the failure of the first actuator on r1, r2, r3, and r4 respectively; c 12 、c 22 、c 32 、c 42 Represent the impact of the second actuator failure on r1, r2, r3, and r4 respectively; c 13 、c 23 、c 33 、c 43 Represents the impact of the third actuator failure on r1, r2, r3, and r4 respectively; c 14 、c 24 、c 34 、c 44 They represent the impact of the fourth actuator failure on r1, r2, r3, and r4 respectively; According to the fault matrix, the corresponding symbol matrix sign(F) is expressed as: According to the sign matrix sign(F), a residual logic table between single motor fault and positive and negative residual values ​​is obtained.

6. The method for quickly locating a quadrotor UAV actuator fault according to claim 5, wherein: In step S5, the detection threshold is expressed as: R th =f(||d|| ∞ ,||w|| ∞ ,t); Among them, ||·|| ∞ represents the ∞ norm of the vector; R th Represents a function related to interference d, noise w and time t.

Images