CN-121978916-A - Non-singular terminal sliding mode formation control method of four-rotor unmanned aerial vehicle under input limitation

CN121978916ACN 121978916 ACN121978916 ACN 121978916ACN-121978916-A

Abstract

The invention discloses a non-singular terminal sliding mode formation control method of a four-rotor unmanned aerial vehicle under input limitation, relates to the technical field of unmanned aerial vehicle control, and aims to solve the problems of low track tracking and formation control precision, poor stability, buffeting and the like of the four-rotor unmanned aerial vehicle under the conditions of actuator saturation, uncertain parameters and external interference. The method comprises the steps of 1) establishing a four-rotor unmanned aerial vehicle second order dynamics model under input limitation, considering speed command amplitude limitation, 2) designing a nonsingular terminal sliding mode function based on an exponential approach law, guaranteeing limited time convergence of a system and no singular point, 3) introducing a unit vector control method to inhibit sliding mode switching buffeting, 4) approaching input saturation errors and unknown interference by adopting an RBF neural network, 5) designing an adaptive law on-line updating control parameter, compensating parameter time-varying characteristics, and 6) expanding the adaptive overall rapid nonsingular terminal sliding mode formation control method based on a pilot-following model.

Inventors

TANG XIAOMING
LI ZHEXING
CAI LINQIN
WANG HUIMING
LIU YANGJIE
LI HAOLAN
TANG HUI
QIN YUJIE

Assignees

重庆邮电大学

Dates

Publication Date: 20260505
Application Date: 20260106

Claims (9)

1. The non-singular terminal sliding mode formation control method of the four-rotor unmanned aerial vehicle under the input limitation is characterized by comprising the following steps of: s1, establishing a four-rotor unmanned aerial vehicle dynamics model under input limitation, wherein the model considers speed instruction amplitude limitation, and the expression is as follows: Wherein, the For the current location of the drone, For the current speed of the vehicle, In order to control the gain of the gain control, For the acceleration of the unmanned aerial vehicle, Is a speed vector of the unmanned aerial vehicle, The acceleration vector is the unmanned aerial vehicle acceleration vector; s2, defining position tracking error , For a desired position, a nonsingular terminal sliding mode function is designed: Wherein, the 、 As a parameter of the sliding mode, the sliding mode is defined, For the termination index to be a function of the termination index, ; S3, designing a sliding mode control law prototype based on an index approach law, and introducing a unit vector control method to inhibit system buffeting, wherein the approach law expression is as follows: Wherein, the In order to approach the gain of the law, For the buffeting suppression parameter, As a sign function by a unit vector Approximating a surrogate sign function; Is a tiny constant; s4, approximating the input saturation error and the unknown interference term by adopting RBF neural network , As a saturation function, the neural network outputs as , As a matrix of weights, the weight values, Is a radial basis function; S5, designing an adaptive law to update RBF neural network weights and control parameters on line, wherein the adaptive law expression is as follows: Wherein, the In order to adapt the gain matrix in a way that, Is the attenuation coefficient; s6, substituting the RBF neural network output and the self-adaptive law into a sliding mode control law, and deriving a final control law through an error dynamics relation.
2. The method for controlling the non-singular terminal sliding mode formation of the four-rotor unmanned aerial vehicle under the input limitation according to claim 1, further comprising the step of expanding the control of the formation of the multiple unmanned aerial vehicles: s7, adopting a pilot-following model to define the number 0 aircraft as the pilot aircraft, namely the track is known in position Speed and velocity of Acceleration of First, the Random position error Speed error Wherein To be spaced from the desired distance of the navigator at random, For the i-th current position vector with random, ; S8, designing a global quick nonsingular terminal sliding mode surface: Wherein, the In order to approximate the error compensation coefficient, The error upper bound estimate is approximated for the RBF neural network, A global fast nonsingular terminal slip plane for the ith follower, The p power term of the position error of the ith follower; S9, designing formation self-adaptive control law based on sliding mode arrival conditions, wherein the formation self-adaptive control law comprises RBF weight updating law and approximation error upper bound estimation law, and the RBF weight updating law is based on sliding mode variables And radial basis function The designed self-adaptive rule is used for online adjustment of the weight matrix of the ith and random RBF neural network The approximation accuracy of the saturation error and the unknown interference is improved, the approximation error upper bound estimation law refers to the self-adaptive rule of the approximation error upper bound of the ith and random RBF neural network, and the stability of the formation system is ensured by compensating the error.
3. The method for controlling the sliding mode formation of the non-singular terminal of the quad-rotor unmanned helicopter under the input limitation according to claim 1, wherein the step S6 is characterized in that the RBF neural network output and the adaptive law are substituted into the sliding mode control law, and the final control law is obtained through the derivation of the error dynamics relation, and the specific derivation process is as follows: first, define by error Solving a second derivative on time to obtain an error dynamics equation: Is arranged to obtain In order to obtain the desired position vector, For the current position vector of the drone, In order for the desired acceleration vector to be present, 、 The desired speed derivative and the error second derivative respectively; Second, the dynamics model in the step S1 is obtained In conjunction with the above-described error dynamics equation, cancellation The method comprises the following steps: Preliminary collation control input The expression of (2) is: ; Third, the sliding mode function designed in the step S2 Deriving to obtain ; Fourth, the approach law based on unit vector improvement in the step S3 is improved Substitution co-ordinates gives: Sorting and solving Related items: ; Fifth, introducing the saturated error compensation term approximated by RBF neural network in step S4 Considering unknown disturbance compensation under input limitation, substituting each relation into preliminary arrangement Expression, finally derived: Wherein, the 、 The desired speed derivative and the error second derivative, respectively.
4. The method for non-singular terminal sliding mode formation control of a quad-rotor unmanned helicopter with limited input according to claim 1, wherein said limited input in step S1 comprises a speed command amplitude limitation The dynamic model corrects the traditional second-order model to match the actual control scene; The unit vector control method in the step S3 is implemented by Substitution of sign function Wherein (0,0.1) To achieve buffeting amplitude suppression Within the range.
5. The method for controlling the sliding mode formation of the non-singular terminal of the four-rotor unmanned aerial vehicle under the input limitation according to claim 1, wherein the radial basis function of the RBF neural network in the step S4 adopts a gaussian function Wherein As a center of the cluster, Network input for width parameters 。
6. The method for controlling the formation of the nonsingular terminal sliding mode of the four-rotor unmanned aerial vehicle under the input limitation according to claim 2, wherein the formation adaptive control law in the step S9 is through a Lyapunov function It is derived that the data of the first cell, For the ith and random Lyapunov functions, For the ith and random slip-form surface variables, For the ith and random RBF neural network weight matrix, For the ith random adaptive gain matrix, The method is characterized in that an i-th and random RBF neural network approximates an error upper bound estimated value, and the specific deduction process is as follows: the first step, the relation of the variable derivative and the error of the formation sliding mode is determined by the formation sliding mode surface Deriving and obtaining Wherein , ; Second step, substituting the random dynamics model , In order to follow the speed of the machine, For controlling and inputting the acceleration of the following machine, combining the track information of the navigation machine Is arranged to obtain ; Third step, design formation slip form approach law , Is the i-th and random RBF neural network approximation term, and is compared with the above The expressions are linked; fourth step, lyapunov function And (3) derivative: Wherein 、 ; Fifth step, in order to make Ensure the asymptotic stability of the system and substituting , 、 For following the input of the mechanical neural network, a random weight update law is designed Upper bound estimation law of approximation error Finally deducing formation self-adaptive control law ; By the above derivation, ensure Negative setting, the system is asymptotically stable.
7. The four-rotor unmanned aerial vehicle nonsingular terminal sliding mode formation control method under the input limitation of claim 6, wherein the four-rotor unmanned aerial vehicle nonsingular terminal sliding mode formation control method is characterized in that a Simulink control module of the indoor unmanned aerial vehicle is established based on an indoor optical positioning system experiment platform, and five-machine track tracking formation control of the indoor unmanned aerial vehicle is realized, and the specific method is as follows: Firstly, performing real-time pose estimation on a plurality of unmanned aerial vehicles by utilizing an optical positioning system, and acquiring three-dimensional position and speed information of the unmanned aerial vehicles. Secondly, designing and realizing a track tracking control algorithm in a Matlab/Simulink environment, taking real-time pose information as input, and outputting a desired speed control instruction; and finally, transmitting a control instruction to the unmanned aerial vehicle through the wireless communication module to complete flight control.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for controlling the non-singular terminal slipform formation of a quad-rotor unmanned helicopter under input limitation as claimed in any one of claims 1 to 7 when executing the program.
9. A computer program product comprising a computer program which, when executed by a processor, implements a method of non-singular terminal slipform formation control of a quad-rotor unmanned helicopter with limited input as claimed in any of claims 1 to 7.

Description

Non-singular terminal sliding mode formation control method of four-rotor unmanned aerial vehicle under input limitation Technical Field The invention relates to the technical field of unmanned aerial vehicle control, in particular to a four-rotor unmanned aerial vehicle nonsingular terminal sliding mode control method under input limitation, which is suitable for single unmanned aerial vehicle high-precision track tracking and multi-unmanned aerial vehicle cluster formation control, and is particularly suitable for complex scenes with actuator saturation, parameter time-varying and external interference. Background The four-rotor unmanned aerial vehicle has been widely applied to the military and civil fields such as battlefield monitoring, fire rescue, environment detection and the like due to the characteristics of strong maneuverability, flexible lifting and the like. However, in an actual control system, the problem of limited input caused by the saturation of an actuator (such as motor rotation speed limitation and thrust amplitude constraint) can seriously affect the control precision and the system stability, meanwhile, the traditional sliding mode control has the defects of obvious buffeting, slow convergence speed, singular points of a terminal sliding mode and the like, and in addition, the control difficulty is further increased due to uncertain unmanned aerial vehicle dynamic parameters and external interference (such as air flow disturbance). In the prior art, unmanned aerial vehicle hover control under input saturation is designed by a back-stepping method, but the rapid convergence of track tracking is not considered, the compensation parameters are uncertain by self-adaptive control in part of the method, but sliding mode buffeting is not effectively restrained, the RBF neural network has universal approximation characteristics, but saturation error compensation application under an input limited scene still needs to be optimized, and in multi-unmanned aerial vehicle formation control, the rapid convergence of pilot-following models and global stability are difficult to be compatible. Therefore, a need exists for a four-rotor unmanned aerial vehicle control method that can simultaneously address input limitations, buffeting suppression, rapid convergence, and parameter adaptation. The invention aims to solve the technical problems that: 1. the input limitation caused by the saturation of the four-rotor unmanned aerial vehicle actuator influences the execution effect of the control instruction; 2. Buffeting problem and singular point problem of terminal sliding mode of traditional sliding mode control; 3. the robustness caused by the time-varying of system parameters and external interference is insufficient; 4. rapid convergence of multi-drone formation control contradicts global stability. Through retrieval, application publication number CN116088548B, a four-rotor unmanned aerial vehicle attitude control method based on a rapid nonsingular terminal sliding mode comprises the following steps of constructing a mathematical model of the four-rotor unmanned aerial vehicle, designing an integral terminal sliding mode function to eliminate steady state errors and realize finite time convergence, designing an online self-adaptive estimation law to compensate parameter uncertainty and unknown external interference, and designing selection criteria of controller parameters gamma φ1、γθ1 and gamma ψ1. The four-rotor unmanned aerial vehicle attitude control method provided by the invention adopts the rapid nonsingular terminal sliding function with the integral element, so that the tracking precision can be effectively improved, and the rapid response speed is kept. The method adopts an adaptive estimation law to update the control gain on line, and the adaptive estimation law eliminates the requirement on disturbance upper bound information. The invention realizes the dynamic adjustment of the control parameters in the sliding mode function, thereby simplifying the parameter adjusting process to obtain the expected tracking performance under the condition of moderately controlling buffeting. 1. The comparison patent CN116088548B only aims at unmanned aerial vehicle attitude control, does not relate to the problem of limited input (actuator saturation), but definitely aims at unmanned aerial vehicle current particle formation control, considers speed instruction amplitude limitation and actuator saturation characteristics, approximates saturation error through RBF neural network, and pertinently solves control performance attenuation caused by limited input; 2. Compared with patent CN116088548B, the method can eliminate steady-state errors by adopting an integral terminal sliding mode function, but does not optimize the buffeting suppression effect, and introduces a unit vector control method to replace a symbol function, so that buffeting amplitude is suppressed in an acceptable engineering range; 3