Multi-unmanned aerial vehicle adaptive distributed control method based on relative measurement information

Abstract

The invention discloses a multi-unmanned aerial vehicle adaptive distributed control method based on relative measurement information, and belongs to the field of multi-unmanned aerial vehicle formation. The implementation method comprises the following steps: establishing kinematics and dynamics equations of the quad-rotor unmanned aerial vehicle, constructing a relation between a position pi and thrust Ti, and constructing a relation between a unit quaternion qi and torque tau i; according to the position, obtaining an instruction quaternion # imgabs0 #, realizing parameter mi identification by designing two dynamic systems, and based on an auxiliary system and a variable zi, combining a kinematics equation, designing adaptive distributed control force, obtaining a closed-loop position system, and realizing position formation tracking; two dynamic systems are designed to realize parameter Ji identification, an instruction quaternion # imgabs1 # is obtained according to a specific position and a position calculation instruction quaternion method, a closed-loop attitude system is obtained based on a variable si, a kinetic equation is combined, an adaptive control moment is designed, and attitude tracking is realized on the basis of position tracking. According to the invention, the precision and efficiency of multi-unmanned aerial vehicle adaptive distributed control can be improved.

Description

Technical Field

[0001] The invention relates to a multi-UAV adaptive distributed control method based on relative measurement information, belonging to the field of multi-UAV formation. Background Art

[0002] In recent years, the rapid development of artificial intelligence has led to the widespread application of multi-agents in military and civilian fields. Quadrotors have attracted widespread attention due to their sophisticated structure, flexible use and cost-effectiveness. They have been used in disaster relief, environmental testing, military reconnaissance and other fields.

[0003] In order to improve the performance of multiple UAVs, leader-follower mode or leaderless mode is usually adopted, both of which can realize coordinated flight of multiple UAVs. However, since most control strategies require network communication to realize information exchange, there is a heavy communication burden. From the practical application point of view, network communication may be subject to potential network attacks or interference, resulting in mission failure; and since the inertial parameters of UAVs may change when flying in the air. Therefore, it is necessary to propose a distributed control scheme that uses airborne sensors to obtain local and relative measurements, and includes inertial parameter identification function.

[0004] In order to solve this problem, the existing method can use an adaptive controller to solve the problem of parameter changes. Although this method can realize UAV formation in the presence of parameter uncertainty, it is mainly to ensure the stability of the closed loop, and cannot complete the accurate identification of unknown parameters; fuzzy control can also be used. Through fuzzy rules and membership functions, fuzzy control can adapt to changes in system parameters, but the design of fuzzy rules and membership functions in this method is more subjective and requires a lot of experiments to determine; intelligent control including neural network control and other methods can also be used to solve the problem of parameter changes, but this method requires a large amount of training data, and the training process may be time-consuming, and its control performance is affected by the quality of training samples and data; this problem can also be solved by model predictive control, which optimizes control actions by predicting future system behavior, but this method requires very high accuracy in mathematical modeling and large amount of calculation, and may not be applicable to systems with high real-time requirements. Summary of the invention

[0005] In order to solve the problem of inertial parameter changes on the basis of good control performance, small calculation amount and real-time control, the purpose of the present invention is to provide a multi-UAV adaptive distributed control method based on relative measurement information. On the one hand, this method no longer relies on network communication for information exchange, but adopts a distributed control strategy and only uses relative position information for local information exchange; on the other hand, the present invention uses data-based system identification to enable the inertial parameters to converge to their true values, thereby improving the accuracy and efficiency of multi-UAV adaptive distributed control.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] The present invention discloses a multi-UAV adaptive distributed control method based on relative measurement information, comprising the following steps:

[0008] Step 1: Establish the kinematics and dynamics equations of the quadrotor drone and construct the position p i and thrust T i The connection between them and construct the unit quaternion q i and torque τ i The connection between them; get the command quaternion according to the position Thereby expressing the command attitude of the quadrotor.

[0009] Step 2: Due to the mass m i Unknown, parameter m is realized by designing two dynamic systems i Identification, further propose auxiliary systems and variables z i ,Combined with the kinematic equation established in step 1, an adaptive distributed control force is designed to obtain a closed-loop position system, thereby realizing position formation tracking.

[0010] Step 3: Due to the moment of inertia J i Unknown, realized by designing two dynamic systems i Identify, and obtain the command quaternion based on the specific position obtained in step 2 and the method of calculating the command quaternion from the position mentioned in step 1 Further propose variable s i , combined with the dynamic equation established in step one, the adaptive control torque is designed to obtain a closed-loop attitude system, and finally attitude tracking is achieved based on the position tracking in step two.

[0011] The process of establishing the kinematic equation and the dynamic equation described in step 1 is as follows:

[0012] According to the Euler-Newton formula, the kinematic and dynamic equations of the quadrotor drone are expressed as follows:

[0013]

[0014] In the formula, p i is the position of the drone, v i and ω i are velocity and angular velocity, m i is the mass, g is the acceleration due to gravity, is the unit vector in the z-axis direction, T i It is along The thrust applied in the direction is the unit quaternion expressing the attitude of the quadrotor drone, q i0 is the real part of the quaternion, q ir is the imaginary part of the quaternion. The imaginary part specifically contains three items and can be expressed as q ir =[q ir1 ;q ir2 ;q ir3 ],function I3 is the third-order unit matrix; J i is the moment of inertia, τ i is the torque, R i ∈SO(3) is the rotation matrix, which is expressed as follows:

[0015]

[0016] Further based on the kinematic equation proposed in formula (2), the command quaternion is calculated according to the position

[0017] According to the cascade system control design, divide both sides of formula (2) by m i , thus defining the right-hand side formula as an intermediate variable where u ix 、u iy and u iz are intermediate variables u i There are three specific items in the command rotation matrix is the instruction unit quaternion, then:

[0018]

[0019] In the formula, is a unit vector. To ensure the consistency of formula (6), Formula (6) is further derived as:

[0020]

[0021] Then the command thrust T i for:

[0022]

[0023] To solve the redundancy of quaternions, let Thus we get:

[0024]

[0025] Then the instruction quaternion is extracted as:

[0026]

[0027] Among them, the design of the distributed formation control scheme described in step 2, the closed-loop position system establishment process is as follows:

[0028] m i is an unknown constant. To achieve parameter identification, two dynamic systems are designed as follows:

[0029]

[0030] Among them, Q i and P i The dynamic system state in equation (12) is represented by zero initial state. The filter is defined as and filter θ i and Represents the filter state, and the initial state is zero; the constant η p >0;f i To control; define variables definition ι1, ι2 and All are positive numbers, ι1<ι2;

[0031] Assistance Systems:

[0032]

[0033] in, Represents the state of the auxiliary system (13), and the initial state is zero; represents the set of drones adjacent to drone i; constant Constant κ r >0; relative position error p ij represents the relative position of UAV i and UAV j, δ ij Define a new variable for the expected position difference between drones where κ1 is a positive constant, i Multiply both sides by m i And derive it to get:

[0034]

[0035] Combined with formula (2), we get the intermediate variable is m i The estimated value of the control force f i It is expressed as:

[0036]

[0037] The adaptation rule is:

[0038]

[0039] In the formula, κ2, β i and c i are all positive numbers. The dynamic system in formula (12) is pi The states at each moment are and in μ1 is the corresponding time when the dynamic system in equation (12) reaches the maximum value; the estimation error is defined as The closed-loop position system is obtained as:

[0040]

[0041] Based on this closed-loop system, a complete UAV position control system is established, and the position formation tracking of multiple UAVs is achieved based on the complete UAV position control system.

[0042] Among them, the design of the distributed formation control scheme described in step 3, the closed-loop attitude system establishment process is as follows:

[0043] Moment of inertia J i is an unknown symmetric positive definite constant matrix. In order to obtain an accurate estimate of the moment of inertia, the dynamic system of equation (18) is introduced:

[0044]

[0045] in, and The dynamic system state in equation (18) is represented by zero initial state. The filter is defined as and filter ζ i and ρ i is the filter state, and its initial state is zero; constant η a >0; define variable is a linear operator, for x = [x1, x2, x3] T , definition l1, l2 and is a constant, l1<l2;

[0046] The attitude error is It is expressed as:

[0047]

[0048] Among them, ⊙ is the quaternion multiplication, It can be obtained based on the drone position obtained in step 2 and the method described in step 1. express The inverse, Dynamics meet:

[0049]

[0050] In the formula, is the angular velocity error, is the command angular velocity, and the attitude error dynamic equation is expressed as:

[0051]

[0052] Define variable φ i =[j x ,j y ,j z ,j yz ,j xz ,j xy ] T , define a new variable For i Multiply by J i And the derivative is:

[0053]

[0054] Combined with formula (4), we get the intermediate variable κ3 is a positive constant, is φ i The estimated value of , the control torque is:

[0055]

[0056] The law of change is:

[0057]

[0058] Among them, κ4, α i and γ i are all positive numbers. The dynamic system in formula (18) is ai The states at each moment are and in μ2 is the corresponding time when the dynamic system in equation (18) reaches the maximum value; the estimated error The final closed-loop posture system is:

[0059]

[0060] Based on this closed-loop system, a complete UAV attitude control system is established, and the attitude tracking of multiple UAVs is achieved based on the complete UAV attitude control system.

[0061] Beneficial effects:

[0062] 1. The present invention discloses a distributed control method based on relative measurement information, which directly uses the UAV airborne sensors to obtain local and relative measurement values, and uses the relative position information to perform distributed control, thereby completing a complete control system, rather than using absolute information exchanged through communication equipment. This method can improve the anti-interference and anti-attack capabilities of the UAV collaborative control system, thereby resisting network interference and network attacks and reducing the communication burden.

[0063] 2. The present invention discloses a data-based adaptive parameter identification method, which utilizes data-based system identification to respectively solve the unknown problems of mass and moment of inertia, and can solve the problem of parameter changes of UAVs during flight, and can make the parameters converge to true values, thereby improving control performance, so that the UAV formation system can complete collaborative control flight and multi-UAV formation control during parameter changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Schematic diagram of setting coordinate system for multiple quadrotor drones;

[0065] Figure 2 This is a flow chart of a multi-UAV adaptive distributed control method based on relative measurement information disclosed in the present invention;

[0066] Figure 3 Schematic diagram of mass estimation error;

[0067] Figure 4 This is a schematic diagram of the moment of inertia estimation error;

[0068] Figure 5 This is a schematic diagram of the flight process. DETAILED DESCRIPTION

[0069] In order to better illustrate the purpose and advantages of the present invention, the invention is further described below with reference to the accompanying drawings and examples.

[0070] Embodiment 1:

[0071] like Figure 2 As shown, the present embodiment discloses a multi-UAV adaptive distributed control method based on relative measurement information, and the specific implementation steps are as follows:

[0072] (1) Establishing a quadrotor UAV dynamics model

[0073] According to the Euler-Newton formula, the kinematic and dynamic equations of the quadrotor drone are expressed as follows:

[0074]

[0075] In the formula, p i is the position of the drone, v i and ω i are velocity and angular velocity, m i is the mass, g is the acceleration due to gravity, is the unit vector in the z-axis direction, T i It is along The thrust applied in the direction is the unit quaternion expressing the attitude of the quadrotor drone, q i0 is the real part of the quaternion, q ir is the imaginary part of the quaternion. The imaginary part specifically contains three items and can be expressed as q ir =[q ir1 ;q ir2 ;q ir3 ],function I3 is the third-order unit matrix; J i is the moment of inertia, τ i is the torque, R i ∈SO(3) is the rotation matrix, which is expressed as follows:

[0076]

[0077] Further based on the kinematic equation, the command quaternion is calculated according to the position

[0078] According to the cascade system control design, divide both sides of the kinematics formula by m i , thus defining the right-hand side formula as an intermediate variable where u ix 、u iy and u iz are intermediate variables u i There are three specific items in the command rotation matrix is the instruction unit quaternion, then:

[0079]

[0080] In the formula, is a unit vector. To ensure the consistency of the above formula, It can be further deduced as:

[0081]

[0082] Then the command thrust T i for:

[0083]

[0084] To solve the redundancy of quaternions, let Thus we get:

[0085]

[0086] Then the instruction quaternion can be extracted as:

[0087]

[0088] The mass and moment of inertia of the drone are unknown. The number of quad-rotor drones to be designed is 4, such as Figure 1 As shown, the coordinate distribution of multiple quad-rotor drones can be seen.

[0089] Given the mass of the drone is m i =0.85kg, moment of inertia is J i =diag[0.050,0.042,0.086]kgm 2 , but these two data are unknown in the experiment and require parameter identification. Let the expected value of the displacement difference between each follower and the leader at the end of the flight be δ i , then set δ1 = [4,4,0] T m,δ2=[4,-4,0] T m,δ3=[-4,-4,0] T m,δ4=[-4,4,0] T m, that is to say, at the end, the four followers should be distributed in a square with the leader as the center and a side length of 8m.

[0090] Set the initial position of the navigator p0 = [0,0,0] T m, the initial speed of the navigator is v0 = [0.4, 0.4, 0.25] T ms -1 , the initial positions of the followers are p1(0)=[5,3.5,0] T m,p2(0)=[4,-4.5,0] T m, p3(0) = [-4.5, -3.5, 0] T m, p4(0) = [-4.5, 5, 0] T m; the initial postures of the followers are q1(0) = [0.9848, 0, 0, 0.1736] T , q2(0)=[0.9962,0,0,-0.0872] T , q3(0)=[0.9914,0,0,0.1305] T , q4(0)=[0.9659,0,0,-0.2588] T .

[0091] (2) Design the applied force to achieve a closed-loop position system

[0092] Based on the design of the distributed formation control scheme described in step 2, the detailed steps of force control are as follows:

[0093] m i is an unknown constant. To realize parameter identification, two dynamic systems are designed as follows:

[0094]

[0095] In the formula, Q i and P i Represents the state of the dynamic system, and the initial state is zero; define the filter and filter θ i and Represents the filter state, and the initial state is zero; η p =0.06; f i To control; define variables definition

[0096] Assistance Systems:

[0097]

[0098] in, Indicates the state of the auxiliary system, and the initial state is zero; represents the set of drones adjacent to drone i; κ r =0.4; relative position error p ij represents the relative position of UAV i and UAV j, δ ij Define a new variable for the expected position difference between drones κ1=0.2,for z i Multiply both sides by m i And derive it to get:

[0099]

[0100] In the formula,

[0101] f i For control. is m i The estimated value of the design control force f i It is expressed as:

[0102]

[0103] The adaptation rule is:

[0104]

[0105] In the formula, κ2, β i and c i are all positive numbers, κ2=2.4, β1=β2=β3=β4=0.01, c1=0.5, c2=10, c3=2, c4=10; the dynamic system is pi The states at each moment are and in μ1 is the corresponding time when the dynamic system reaches the maximum value; the estimation error is defined The closed-loop position system is obtained as:

[0106]

[0107] Based on this closed-loop system, a complete UAV position control system can be established, thereby realizing the position formation tracking of multiple UAVs.

[0108] (3) Design the applied torque to realize the closed-loop attitude system

[0109] Based on the design of the distributed formation control scheme described in step 3, the detailed steps of torque control are as follows:

[0110] Moment of inertia J i Is an unknown symmetric positive definite constant matrix. In order to obtain an accurate estimate of the inertia matrix, a dynamic system is introduced:

[0111]

[0112] in, and Represents the state of the dynamic system, and its initial state is zero; define the filter and filter ζ i and ρ i is the filter state, and its initial state is zero; η a =0.06; define variables is a linear operator, for x=[x1,x2,x3] T , definition 1, 2 and is a constant, l1<l2, l1=0.1, l2=10,

[0113] The attitude error is It is expressed as:

[0114]

[0115] Among them, ⊙ is the quaternion multiplication, It can be obtained based on the drone position obtained in step 2 and the method described in step 1. express The inverse, Dynamics meet:

[0116]

[0117] In the formula, is the angular velocity error, is the command angular velocity, and the attitude error dynamic equation is expressed as:

[0118]

[0119] Define variable φ i =[j x ,j y ,j z ,j yz ,j xz ,j xy ] T , define a new variable For i Multiply by J i And taking the derivative we get:

[0120]

[0121] Among them, κ3 is a positive constant, κ3=0.5, is φ i The estimated value of , the control torque is:

[0122]

[0123] The law of change is:

[0124]

[0125] Among them, κ4, α i and γ i All are positive numbers, κ4=0.5, α1=α2=α3=α4=0.5, γ1=γ2=γ3=1000, γ4=4000, the dynamic system is ai The states at each moment are and in μ2 is the corresponding time when the dynamic system reaches the maximum value; the estimated error The final closed-loop posture system is:

[0126]

[0127] Based on this closed-loop system, a complete UAV attitude control system can be established to achieve attitude tracking of multiple UAVs.

[0128] According to the above parameters into the corresponding formula to solve, the flight verification is carried out; the error between the estimated value and the actual value of the parameter during the flight is recorded, such as Figure 3 and Figure 4 As shown in Figure 2, the error can gradually drop to zero during the flight, that is, the parameter estimation value can converge to the true value; while recording the flight process, the experimental results are shown in Figure 2. Figure 5 As shown, multiple UAVs can complete the formation of following the leader while maintaining the formation and can achieve the desired position and posture.

[0129] The specific description above further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-UAV adaptive distributed control method based on relative measurement information, characterized in that: The following steps are included: Step 1: Establish the kinematics and dynamics equations of the quadrotor drone and construct the position p i and thrust T i The connection between them and construct the unit quaternion q i and torque τ i The connection between them; get the command quaternion according to the position Thus expressing the command attitude of the quadrotor; Step 2: Due to the mass m i Unknown, parameter m is realized by designing two dynamic systems i Identification, further propose auxiliary systems and variables z i , combined with the kinematic equation established in step 1, the adaptive distributed control force is designed to obtain a closed-loop position system, thereby realizing position formation tracking; Step 3: Due to the moment of inertia J i Unknown, realized by designing two dynamic systems i Identify, and obtain the command quaternion based on the specific position obtained in step 2 and the method of calculating the command quaternion from the position mentioned in step 1 Further propose variable s i , combined with the dynamic equation established in step one, the adaptive control torque is designed to obtain a closed-loop attitude system, and finally attitude tracking is achieved based on the position tracking in step two.

2. The method according to claim 1, characterized in that: The specific implementation method of step one is: According to the Euler-Newton formula, the kinematic and dynamic equations of the quadrotor drone are expressed as follows: Among them, p i is the position of the drone, v i and ω i are velocity and angular velocity, m i is the mass, g is the acceleration due to gravity, is the unit vector in the z-axis direction, T i It is along The thrust applied in the direction is the unit quaternion expressing the attitude of the quadrotor drone, q i0 is the real part of the quaternion, q ir is the imaginary part of the quaternion, which specifically contains three items, expressed as q ir =[q ir1 ;q ir2 ;q ir3 ],function I3 is the third-order unit matrix; J i is the moment of inertia, τ i is the torque, ω i × =[0,-ω i3 ,ω i2 ;ω i3 ,0,-ω i1 ; -ω i2 ,ω i1 ,0],R i ∈SO(3) is the rotation matrix, which is expressed as follows: Further based on the kinematic equation proposed in formula (2), the command quaternion is calculated according to the position According to the cascade system control design, divide both sides of formula (2) by m i , define the right side of the formula as the intermediate variable where u ix 、u iy and u iz are intermediate variables u i There are three specific items in the command rotation matrix is the instruction unit quaternion, then: In the formula, is a unit vector. To ensure the consistency of formula (6), Formula (6) is further derived as: Then the command thrust T i for: To solve the redundancy of quaternions, let get: Then the instruction quaternion is extracted as:

3. The method according to claim 2, characterized in that: The specific implementation method of step 2 is: m i is an unknown constant. To realize parameter identification, two dynamic systems are designed as follows: Among them, Q i and P i The dynamic system state in equation (12) is represented by zero initial state. The filter is defined as and filter θ i and Represents the filter state, and the initial state is zero; the constant η p >0;f i To control; define variables definition ι1, ι2 and All are positive numbers, ι1<ι2; Assistance Systems: in, Represents the state of the auxiliary system (13), and the initial state is zero; represents the set of drones adjacent to drone i; constant Constant κ r >0; relative position error p ij represents the relative position of UAV i and UAV j, δ ij Define a new variable for the expected position difference between drones where κ1 is a positive constant, i Multiply both sides by m i And derive it to get: Combined with formula (2), we get the intermediate variable f i =m i u i ; is m i The estimated value of the control force f i It is expressed as: The adaptation rule is: In the formula, κ2, β i and c i are all positive numbers. The dynamic system in formula (12) is pi The states at each moment are and in μ1 is the corresponding time when the dynamic system in equation (12) reaches the maximum value; the estimation error is defined as The closed-loop position system is obtained as: Based on this closed-loop system, a complete UAV position control system is established, and the position formation tracking of multiple UAVs is achieved based on the UAV position control system.

4. The method according to claim 3, characterized in that: The specific implementation method of step three is: Moment of inertia J i is an unknown symmetric positive definite constant matrix. In order to obtain an accurate estimate of the moment of inertia, the dynamic system of equation (18) is introduced: in, and a i The dynamic system state in equation (18) is represented by zero initial state. The filter is defined as and filter ζ i and ρ i is the filter state, and its initial state is zero; constant η a >0; define variable is a linear operator, for x = [x1, x2, x3] T , definition l1, l2 and is a constant, l1<l2; The attitude error is It is expressed as: Among them, ⊙ is the quaternion multiplication, According to the drone position obtained in step 2 and the method described in step 1, express The inverse, Dynamics meet: In the formula, is the angular velocity error, is the command angular velocity, and the attitude error dynamic equation is expressed as: Define variable φ i =[j x ,j y ,j z ,j yz ,j xz ,j xy ] T , define a new variable For i Multiply by J i And the derivative is: Combined with formula (4), we get the intermediate variable κ3 is a positive constant, is φ i The estimated value of , the control torque is: The law of change is: Among them, κ4, α i and γ i are all positive numbers. The dynamic system in formula (18) is ai The states at each moment are and in μ2 is the corresponding time when the dynamic system in equation (18) reaches the maximum value; the estimated error The final closed-loop attitude system is: Based on this closed-loop system, a complete UAV attitude control system is established, and the attitude tracking of multiple UAVs is achieved based on the complete UAV attitude control system.

Images