CN115826597A - Adaptive neural network-based anti-interference control method and device for rotorcraft - Google Patents
Adaptive neural network-based anti-interference control method and device for rotorcraft Download PDFInfo
- Publication number
- CN115826597A CN115826597A CN202211165741.6A CN202211165741A CN115826597A CN 115826597 A CN115826597 A CN 115826597A CN 202211165741 A CN202211165741 A CN 202211165741A CN 115826597 A CN115826597 A CN 115826597A
- Authority
- CN
- China
- Prior art keywords
- disturbance
- neural network
- adaptive neural
- representing
- attitude
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 96
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000008878 coupling Effects 0.000 claims abstract description 52
- 238000010168 coupling process Methods 0.000 claims abstract description 52
- 238000005859 coupling reaction Methods 0.000 claims abstract description 52
- 230000001629 suppression Effects 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 28
- 239000011159 matrix material Substances 0.000 claims description 18
- 239000002131 composite material Substances 0.000 claims description 11
- 238000013507 mapping Methods 0.000 claims description 11
- 230000005484 gravity Effects 0.000 claims description 10
- 238000012938 design process Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 230000001133 acceleration Effects 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 10
- 230000008569 process Effects 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 239000012636 effector Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000007667 floating Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
A self-adaptive neural network-based rotor craft anti-interference control method and equipment belong to the field of aerial work unmanned aerial vehicle control. The problem of influence of lumped external force and other multi-source interference on system pose control is not considered in the prior art is solved. The method divides various interferences into lumped external interference and internal coupling disturbance for processing respectively, and processes the internal coupling disturbance based on a feedforward compensation mode; and for the extra dynamic disturbance caused by the lumped external force disturbance and various uncertain factors, processing the extra dynamic disturbance as a feedback compensation processing mode based on adaptive neural network estimation, and estimating the extra external disturbance caused by the lumped external force disturbance and various uncertain factors according to the state deviation of the disturbed system. The high-precision pose control of the target is realized in a disturbance suppression mode of accelerating the feedback compensation by the feedforward compensation under the influence of various disturbances existing in the real environment of a rotor aircraft system. The invention is used for the anti-interference control of the rotorcraft.
Description
Technical Field
The invention belongs to the field of aerial operation unmanned aerial vehicle control, and particularly relates to an anti-interference control method of a rotorcraft, a storage medium and equipment.
Background
With the development of the fire speed of the unmanned aerial vehicle technology in more than ten years, the unmanned aerial vehicle has wide application in practical scenes such as aerial photography, surveying and mapping, disaster search and rescue and the like due to the flexible mobility of the unmanned aerial vehicle in a three-dimensional space, and the unmanned aerial vehicle and a three-degree-of-freedom light serial manipulator driven by a steering engine are effectively combined to form a novel unmanned aerial vehicle system which has both flexible mobility in the three-dimensional space and strong manipulation capability, and the unmanned aerial vehicle system has very wide application prospect and research value, so that the unmanned aerial vehicle system is paid more and more attention and widely researched by researchers at home and abroad.
For a rotorcraft system performing aerial work tasks, the actual environment it faces is complex, it may suffer from sudden gust effects, additional gravity interference after performing a grabbing task, and the effects of multi-source interference in its interior due to coupling disturbances between the unmanned aerial vehicle base and the manipulator. Meanwhile, as a typical under-actuated, strongly coupled, nonlinear dynamic system, the rotorcraft is equipped with a manipulator on the basis of being used as a floating base, and then the stable control of the whole system becomes very difficult. Firstly, external force interference applied by external environments such as gust, object grabbing task contact force, extra gravity and the like (when a system is modeled, the disturbance can be called as lumped interference) easily influences the pose control precision of an unmanned aerial vehicle base, and further reduces the execution precision of a manipulator end effector. Then, the coupling disturbance influence of the unmanned aerial vehicle and the manipulator is directly transmitted through rigid connection between the unmanned aerial vehicle and the manipulator, so that the attitude deviation of the unmanned aerial vehicle is caused, and the attitude error further influences the position control deviation of the unmanned aerial vehicle due to the under-actuated characteristic of the unmanned aerial vehicle system, so that the control precision of the manipulator end effector of the rotor craft is seriously influenced, the failure of an aerial operation task is caused, and even the stable operation of the whole system is threatened. The necessary condition for executing aerial operation tasks is that a manipulator end effector carried by the rotor craft can accurately reach an expected pose state, and the premise of the condition is that the unmanned aerial vehicle base can still keep stable and high-precision pose control under the influence of multi-source interference of an actual environment. Therefore, it is very important to research and analyze a high-precision robust pose control method of the rotorcraft under multi-source interference.
In the existing related research, for example, patent application No. CN202010801707 proposes a method for controlling the attitude of an unmanned aerial vehicle with an arm and a rotor, which aims at centroid shift and base floating, and has two problems, one of which is that the situation that lumped external interference such as gust affects the position control of the system is not considered, so that additional processing is required for the position control in practical application; secondly, when the internal coupling disturbance is considered, only the mass center offset caused by the movement of the mechanical arm is considered, and the condition that the inertia of the system is changed is not considered, so that the effect of inhibiting the internal coupling disturbance is relatively reduced. The patent application number CN201810014083 proposes a flying mechanical arm based on sliding-film PID control, wherein the proposed method does not consider lumped external interference and internal coupling disturbance, and only depends on robustness of a control algorithm, so that the serious influence of multi-source interference on system pose control is difficult to suppress. The patent application number CN201810094313 provides a rotor flight mechanical arm system and an algorithm based on dynamic gravity center compensation, wherein the proposed method also has the internal coupling disturbance suppression only considering gravity center deviation, and does not consider the influence of other multi-source interference such as lumped external force on the system pose control, so that the practical application thereof is difficult to avoid certain limitation.
Disclosure of Invention
The invention aims to solve the problem that influence of lumped external force and other multi-source interference on system pose control is not considered in the prior art. And further provides a rotorcraft anti-interference control method based on the adaptive neural network.
The rotor craft disturbance rejection control method based on the adaptive neural network realizes the disturbance rejection control of the rotor craft by utilizing the composite disturbance rejection multi-source disturbance resisting pose controller based on the adaptive neural network;
the design process of the composite multi-source interference resisting pose controller based on the adaptive neural network comprises the following steps:
the method comprises the following steps: constructing a rotorcraft system dynamics model containing lumped external force interference, coupling disturbance influence and internal coupling disturbance:
wherein,respectively representing a position vector and a velocity vector of the unmanned aerial vehicle under an inertial coordinate system; m is s The total mass of a rotor craft system carrying a three-degree-of-freedom light series manipulator driven by a steering engine is shown; f l Tau respectively representing the generation of propellers of the droneLift and moment; I R B a rotation matrix representing the machine to the inertial system; g represents the acceleration of gravity; e.g. of the type 3 =[0 0 1] T Represents a vertical direction unit vector;respectively representing an internal coupling disturbance force and a moment; f. of add ,τ add Representing lumped external force interference applied by external environments such as gust, object grabbing interaction and the like in the actual environment; phi (phi) of b =[φ,θ,ψ]Representing roll, pitch, and yaw angles of the drone; t (phi) b ) A rotation matrix representing a body angular velocity to an attitude angular velocity; B ω b representing an angular velocity vector of the unmanned aerial vehicle body; x represents a cross product operation;representing a rotary inertia matrix I of a rotorcraft system b The inverse of (1);
step two: converting the constructed dynamic model into a state space form;
step three: designing a backstepping control method with a cascade control characteristic as a basic controller based on the underactuation and cascade characteristics of an unmanned aerial vehicle base, wherein the basic controller comprises a basic position controller aiming at a position ring and a basic attitude controller aiming at an attitude ring;
step four: designing the feedforward compensation quantity of the coupling disturbance based on the internal coupling disturbance brought by the motion of the manipulator and the real-time state quantity mapping of the unmanned aerial vehicle and the manipulator:
wherein f is 1 ,f 2 Respectively representing the mapping relation of each state quantity of the unmanned aerial vehicle and the manipulator to the coupling disturbance force and moment m s ,m man Shows the total mass of a rotor craft system carrying three-degree-of-freedom light serial manipulator driven by a steering engine and the mass of the manipulator, I R B representing the machine to the inertial systemThe matrix of the rotation is then rotated in a direction, B R I representing a rotation matrix of the inertial system to the machine system, B r omc representing the representation of the mass center vector of the manipulator in a body coordinate system, B r oc a representation of a vector representing the center of mass of a rotorcraft system in a body coordinate system, B r o representing the representation of the base centroid vector of the unmanned aerial vehicle in the body coordinate system, B ω b representing the angular velocity vector of the unmanned aerial vehicle body,expression of the rotary inertia matrix of the representative manipulator relative to the origin of the unmanned aerial vehicle base in the system, ge 3 Representing a gravity vector; f dis , B τ dis Respectively representing feedforward compensation force and moment estimated values obtained based on the coupling disturbance and unmanned aerial vehicle and manipulator state mapping relations;
step five: for extra dynamic disturbance caused by lumped external force interference corresponding to the position ring and the attitude ring, estimating as a disturbance suppression mode based on an adaptive neural network, estimating extra external interference according to the disturbed state deviation of the system, and adding the extra external interference as feedback compensation quantity into a basic controller to realize compensation of the external interference;
step six: obtaining a composite anti-multi-source interference pose controller based on the adaptive neural network by calculating feedforward and feedback compensation quantities; the composite multi-source interference resisting pose controller based on the adaptive neural network comprises a multi-source interference resisting position controller based on the adaptive neural network and a multi-source interference resisting attitude controller based on the adaptive neural network.
Further, the state space model in step two is as follows:
wherein x is 1 =x,x 3 =y,x 5 = z represents the position in three dimensions of the drone base in three dimensions respectively,respectively representing the speed x in three directions of the three-dimensional space of the base of the unmanned aerial vehicle 7 =φ,x 9 =θ,x 11 ψ represents the roll, pitch and yaw attitude angles of the drone base, x, respectively 8 =p,x 10 =q,x 12 = r represents roll, pitch and yaw attitude angular velocities of the drone base, respectively, m represents rotorcraft system total mass;respectively representing the disturbance forces and moments of internal coupling, J φ ,J θ ,J ψ Respectively represent three-axis rotational inertia of the unmanned aerial vehicle u x ,u y ,u z Respectively representing control forces in three directions in three dimensions, tau, produced by the lift of the propeller φ ,τ θ ,τ ψ Respectively representing control moments in three directions of the body system, f, generated by the lift of the propeller addx ,f addy ,f addz Respectively representing lumped external disturbance forces f add Components in three directions in three-dimensional space; tau is addφ ,τ addθ ,τ addψ Respectively representing lumped external disturbance moment tau add Components in three directions of the machine system;respectively representing the internal coupling disturbance forcesComponents in three directions in three-dimensional space;respectively representing the internal coupling disturbance torqueThe components in three directions of the machine system.
Further, the design process of the basic position controller for the position loop comprises the following steps:
for the position loop, the position error variables are defined as follows:
wherein x d 、y d 、z d Respectively represent expected position signals in three directions of the unmanned aerial vehicle base three-dimensional space,respectively represent expected speed signals k in three directions of three-dimensional space of unmanned aerial vehicle base 1 、k 3 、k 5 Is a controller parameter;
based on the position error variable, the following basic position controller is designed through a Lyapunov function:
the above k1 to k6 are controller parameters and are all positive numbers.
Further, the design process of the basic attitude controller for the attitude ring comprises the following steps:
for the attitude ring, the attitude error variables are defined as follows:
wherein phi is d ,θ d ,ψ d Respectively representing the expected attitude angle signals of the unmanned aerial vehicle base;are respectively provided withRepresenting an unmanned aerial vehicle base expected attitude angular velocity signal; k is a radical of 7 、k 9 、k 11 Is a controller parameter;
based on the attitude error variable, designing a basic attitude controller by a Lyapunov function as follows:
k7 to k12 are controller parameters and are all positive numbers, phi d ,θ d ,ψ d Representing the desired pose.
Further, the adaptive neural network for suppressing the additional dynamic disturbance caused by the lumped external force interference of the position loop in the step five is as follows:
for position loop lumped external interference, three adaptive neural networks are designedWhereinThe weights and kernel functions of the radial basis function neural networks are respectively represented and are put into the basic position controller to obtain the following form:
separately bringing in position controllers joining an adaptive neural networkThe obtained adaptive neural network approximation error is:
wherein f is 1 ,f 2 ,f 3 Respectively represents f addx ,f addy ,f addz ;
According to an online gradient descent algorithm, the update rate of the neural network in the position ring is obtained as follows:
wherein eta is i Is the learning rate of the radial basis function neural network.
Further, the adaptive neural network for suppressing the additional dynamic disturbance caused by the lumped external force disturbance of the attitude ring in the step five is as follows:
for lumped external interference of attitude ring, three adaptive neural networks are designedWhereinAnd respectively representing the weight and the kernel function of each radial basis function neural network, and putting the weights and the kernel functions into the basic attitude controller to obtain the following form:
also, the attitude controllers added into the adaptive neural network are respectively brought intoObtaining:
wherein, tau 4 ,τ 5 ,τ 6 Respectively represent tau addx ,τ addy ,τ addz ;
Then the obtained update rate of the posture ring neural network is:
W i =η i E i S i (X)
wherein eta is i Is the learning rate of the radial basis function neural network.
Further, the adaptive neural network-based multi-source interference resisting position controller comprises the following steps:
further, the multi-source interference resisting attitude controller based on the adaptive neural network comprises the following steps:
a computer storage medium having stored therein at least one instruction that is loaded into and executed by a processor to implement the adaptive neural network-based rotorcraft anti-interference control method.
An adaptive neural network-based rotorcraft disturbance rejection control apparatus, the apparatus comprising a processor and a memory, the memory having stored therein at least one instruction, the at least one instruction being loaded and executed by the processor to implement the adaptive neural network-based rotorcraft disturbance rejection control method.
Has the advantages that:
the rotor craft anti-interference control method based on the adaptive neural network mainly considers the aerial operation platform which takes the rotor craft as a base and carries a steering engine-driven three-degree-of-freedom light serial manipulator, comprehensively considers the multisource interference conditions of lumped external force interference such as gust, extra gravity after the grabbing task is executed and the like and internal coupling disturbance faced by the system in the real environment, is very close to the actual condition of the system for executing the aerial task, and has very wide practicability. In addition, for the processing of the lumped external interference, a feedback compensation method based on adaptive neural network estimation is provided, the method can quickly and accurately estimate the lumped interference suffered by the system according to the state deviation of the disturbed system, and quickly compensate the lumped interference suffered by the system in a feedback compensation mode; for the processing of internal coupling disturbance, a coupling disturbance feedforward compensation quantity is designed according to the states of a rotor craft and a manipulator, and the method can timely and accurately compensate the coupling disturbance suffered by the system. For various interferences on the aerial operation rotorcraft, the interference rejection control method based on the adaptive neural network can effectively inhibit the influence of multi-source interference on the system, and obviously improve the interference rejection capability, control accuracy and stability of pose control of the system.
Drawings
Fig. 1 is a design flow chart of an adaptive neural network-based disturbance rejection control method for a rotorcraft according to the present invention.
Fig. 2 is a schematic diagram of a control system framework for an adaptive neural network-based approach to disturbance rejection control for a rotorcraft.
Detailed Description
The invention aims to overcome the defects of the existing related researches and provides a self-adaptive neural network-based anti-interference control method for a rotorcraft, which can effectively cope with the multi-source interference of lumped external force interference, such as gust, extra gravity caused after grabbing objects, internal coupling interference and the like, which exist when a rotorcraft system carrying a steering engine-driven three-degree-of-freedom light series manipulator executes aerial work tasks, and realize the high-precision pose control target of the system under various interferences.
The overall technical scheme of the invention comprises five parts, namely constructing a rotor aircraft dynamic model containing lumped external force interference and internal coupling disturbance, designing a basic controller according to a backstepping control method, designing coupling disturbance feedforward compensation quantity, designing an adaptive neural network and the updating rate thereof, and comprehensively solving a composite anti-multisource interference pose controller. For various interference problems faced by a rotor craft system carrying a three-degree-of-freedom light series manipulator driven by a steering engine in a real environment, the invention mainly divides various interferences into a lumped external interference part and an internal coupling disturbance part for respectively processing. For internal coupling disturbance, the invention provides a processing mode of feedforward compensation; for the extra dynamic disturbance caused by lumped external force interference and various uncertain factors, because the disturbance can not be accurately modeled, the value of the disturbance can be estimated only through the disturbed system state deviation, the invention provides a processing mode based on adaptive neural network estimation as feedback compensation, and the processing mode can quickly and accurately estimate the extra external interference caused by the lumped external force interference and various uncertain factors according to the disturbed system state deviation. By the interference suppression mode of the targeted timely feedforward compensation and the rapid feedback compensation, a rotary wing aircraft system can realize a high-precision pose control target under the influence of various interferences existing in a real environment. The present invention will be further described with reference to the following embodiments.
The first embodiment is as follows: the present embodiment is described in connection with figure 1,
the embodiment is a rotorcraft anti-interference control method based on a self-adaptive neural network, which comprises the following steps:
the method comprises the following steps: a rotorcraft system dynamics model containing lumped external force disturbances and coupling disturbance effects and internal coupling disturbances was constructed as follows:
wherein,respectively representing a position vector and a velocity vector of the unmanned aerial vehicle under an inertial coordinate system; m is s The total mass of a rotor craft system carrying a three-degree-of-freedom light series manipulator driven by a steering engine is shown; f l Tau respectively represents the lifting force and the moment generated by the propeller of the unmanned aerial vehicle; I R B a rotation matrix representing the machine to the inertial system; g represents the gravitational acceleration; e.g. of the type 3 =[0 0 1] T Represents a vertical direction unit vector;respectively representing an internal coupling disturbance force and a moment; f. of add ,τ add Representing lumped external force interference applied by external environments such as gust, object grabbing interaction and the like in the actual environment; phi b =[φ,θ,ψ]Representing roll, pitch, and yaw angles of the drone; t (phi) b ) A rotation matrix representing a body angular velocity to an attitude angular velocity; B ω b representing an angular velocity vector of the unmanned aerial vehicle body; x represents a cross product operation;representing a rotary inertia matrix I of a rotorcraft system b The inverse of (c).
Step two: for the design convenience of the subsequent controller, the constructed dynamic model needs to be converted into a state space form, and the converted state space model has the following form:
wherein x is 1 =x,x 3 =y,x 5 = z represents the position in three dimensions of the drone base in three dimensions respectively,respectively representing the speed x in three directions of the three-dimensional space of the base of the unmanned aerial vehicle 7 =φ,x 9 =θ,x 11 ψ represents the roll, pitch and yaw attitude angles of the drone base, x, respectively 8 =p,x 10 =q,x 12 = r represents roll, pitch and yaw attitude angular velocities of the drone base, respectively, m represents rotorcraft system total mass;respectively representing the disturbance forces and moments of internal coupling, J φ ,J θ ,J ψ Respectively represent three-axis rotational inertia of the unmanned aerial vehicle u x ,u y ,u z Respectively representing the control forces in three directions in three dimensions, tau, generated by the lift of the propeller φ ,τ θ ,τ ψ Respectively representing control moments in three directions of the body system, f, generated by the lift of the propeller addx ,f addy ,f addz Respectively representing lumped external disturbance forces f add Components in three directions in three-dimensional space; tau is addφ ,τ addθ ,τ addψ Respectively representing lumped external disturbance moments tau add Components in three directions of the machine system;respectively representing the internal coupling disturbance forcesComponents in three directions in three-dimensional space;respectively representing the internal coupling disturbance torqueComponents in three directions of the machine system;
step three: based on the under-actuated and cascade characteristics of the unmanned aerial vehicle base, a backstepping control method with the cascade control characteristic is designed to serve as a basic controller. First for the position loop, the position error variables are defined as follows:
wherein x d 、y d 、z d Respectively represent expected position signals in three directions of the unmanned aerial vehicle base three-dimensional space,respectively represent expected speed signals k in three directions of three-dimensional space of unmanned aerial vehicle base 1 、k 3 、k 5 Is a controller parameter;
defining the Lyapunov function and obtaining the derivative:
from the above, the following basic position controller can be designed:
the above k1 to k6 are controller parameters and are all positive numbers.
For the attitude ring, the attitude error variables are defined as follows:
wherein phi is d ,θ d ,ψ d Respectively representing the expected attitude angle signals of the unmanned aerial vehicle base;respectively representing the expected attitude angular velocity signals of the unmanned aerial vehicle base; k is a radical of 7 、k 9 、k 11 Is a controller parameter;
since the derivatives of the roll and pitch desired signals are difficult to accurately solve, they are in practice approximated valuesFor ease of illustration, the derivatives of the roll and pitch desired signals are used directlyThe description is given;
the Lyapunov function is defined as follows:
from the above, we can design the basic attitude controller as:
k7 to k12 are controller parameters and are all positive numbers, phi d ,θ d ,ψ d Representing the desired pose.
Step four: the internal coupling disturbance caused by the movement of the manipulator is comprehensively considered and can be obtained by mapping the real-time state quantities of the unmanned aerial vehicle and the manipulator, and the mapping relation can be obtained by modeling or measuring and the like and timely compensated in a feedforward mode. We can design the feedforward compensation amount of the coupling disturbance as follows:
wherein, f 1 ,f 2 Respectively represents the mapping relation of each state quantity of the unmanned aerial vehicle and the manipulator to the coupling disturbance force and moment, m s ,m man Show and carry on rudderThe rotor craft system total mass and manipulator mass of the light serial manipulator with three degrees of freedom of mechanical drive, I R B representing the rotation matrix of the machine to the inertial system, B R I representing a rotation matrix of the inertial system to the machine system, B r omc representing the representation of the mass center vector of the manipulator in a body coordinate system, B r oc a representation of a vector representing the center of mass of a rotorcraft system in a body coordinate system, B r o representing the representation of the base centroid vector of the unmanned aerial vehicle in the body coordinate system, B ω b representing the angular velocity vector of the unmanned aerial vehicle body,expression of the rotary inertia matrix of the representative manipulator relative to the origin of the unmanned aerial vehicle base on the system, ge 3 Representing a gravity vector; f dis , B τ dis Respectively representing feedforward compensation force and moment estimated values obtained based on the state mapping relations of the coupling disturbance, the unmanned aerial vehicle and the manipulator, and being capable of accurately estimating and feedforward compensating the real coupling disturbance force existing in the systemAnd coupling disturbance torque
Step five: for additional dynamic disturbances caused by lumped external disturbances and various uncertainty factors, i.e. f as described above addx ,f addy ,f addz ,τ addφ ,τ addθ ,τ addψ The disturbance suppression mode based on the self-adaptive neural network estimation as the fast feedback compensation is provided, the lumped external force interference and the additional external interference caused by various uncertain factors can be quickly and accurately estimated according to the state deviation after the system is disturbed, and the disturbance suppression mode is used as the feedback compensation quantity to be added into the basic controller to realize the quick and accurate compensation of the external interference.
Firstly, for position loop lumped external interference, three adaptive neural networks are designedWhereinThe weight and the kernel function of each radial basis function are respectively represented, and the method selects the Gaussian kernel function and places the Gaussian kernel function in the basic position controller to obtain the following form:
separately bringing in position controllers joining an adaptive neural networkWe can obtain the adaptive neural network approximation error as:
wherein f is 1 ,f 2 ,f 3 Respectively represent f addx ,f addy ,f addz 。
According to the online gradient descent algorithm, the update rate of the neural network in the position ring can be obtained as follows:
wherein eta is i Is the learning rate of the radial basis function neural network.
Then, for the lumped external interference of the attitude loop, three adaptive neural networks are designed in the same wayWhereinThe weights and kernel functions representing each radial basis function separately, which placed in the basis attitude controller, can be given the following form:
also, the attitude controllers added into the adaptive neural network are respectively brought intoThe following can be obtained:
wherein, tau 4 ,τ 5 ,τ 6 Respectively represent tau addx ,τ addy ,τ addz 。
Then the obtained update rate of the posture loop neural network is:
W i =η i E i S i (X)
wherein eta is i Is the learning rate of the radial basis function neural network.
From the above neural network update rate, the position loop neural network output is the estimated value of the external additional disturbance, and the attitude loop neural network output is the estimated value of the difference between the additional disturbance and the intermediate quantity, because the intermediate quantity α is i The derivative of (a) cannot be accurately obtained, and is processed together by an adaptive neural network.
Step six: by solving the feedforward and feedback compensation quantities, the composite anti-multisource interference pose controller based on the adaptive neural network can be comprehensively solved as follows:
the self-adaptive neural network-based multi-source interference resistant position controller comprises:
the multi-source interference resistant attitude controller based on the adaptive neural network comprises:
and the composite anti-multi-source interference pose controller based on the adaptive neural network is utilized to realize anti-interference control on the rotor wing aircraft.
To this end, the detailed description of the specific implementation of the rotorcraft for executing the task of the aerial work by using the adaptive neural network-based rotorcraft disturbance rejection control method according to the present invention in response to the multi-source disturbance in the real environment is completed, and the schematic diagram of the control system framework of the adaptive neural network-based rotorcraft disturbance rejection control method is shown in fig. 2, and the specific implementation process of the present invention is further explained with reference to fig. 2:
firstly, a controlled object is a rotor craft system, and the controlled object is composed of a three-degree-of-freedom light serial manipulator driven by an unmanned aerial vehicle base carrying a steering engine, and can be influenced by internal coupling disturbance and collective external interference in an actual environment;
then, for a control flow, the position controller consists of a basic position controller, a lumped external disturbance force estimation feedback compensation quantity based on an adaptive neural network and an internal coupling disturbance force feedforward compensation quantity, an expected position signal is firstly sent to the position ring controller, control quantities in three directions of a three-dimensional space can be generated, and the control quantities can be used for solving an expected rolling angle and an expected pitch angle of an attitude ring through a pose conversion relation; the attitude controller consists of a basic attitude controller, an integrated external disturbance torque estimation feedback compensation quantity and an internal coupling disturbance torque feedforward compensation quantity based on a self-adaptive neural network, an expected roll angle and an expected pitch angle which are obtained by calculation of the position controller are combined with an expected yaw angle given out from an expected pose signal to form an expected signal of an attitude ring, and the expected signal is given to the attitude ring controller, so that control quantities in three directions under a machine system can be generated;
and finally, the rotorcraft system is controlled by the control quantity generated by the position and attitude ring controller together, so that the rotorcraft system can realize high-precision pose control targets under various interferences.
The second embodiment is as follows:
the present embodiment is a computer storage medium having at least one instruction stored therein, the at least one instruction being loaded and executed by a processor to implement the adaptive neural network-based rotorcraft disturbance rejection control method.
It should be understood that any method described herein, including any methods described herein, may accordingly be provided as a computer program product, software, or computerized method, which may include a non-transitory machine-readable medium having stored thereon instructions, which may be used to program a computer system, or other electronic device. Storage media may include, but is not limited to, magnetic storage media, optical storage media; a magneto-optical storage medium comprising: read only memory ROM, random access memory RAM, erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers; or other type of media suitable for storing electronic instructions.
The third concrete implementation mode:
the present embodiment is a rotorcraft disturbance rejection control device based on an adaptive neural network, the device including a processor and a memory, it being understood that any device described herein including a processor and a memory may also include other units, modules, etc. that display, interact, process, control, etc. and other functions via signals or instructions;
at least one instruction is stored in the memory and loaded and executed by the processor to implement the adaptive neural network-based rotorcraft disturbance rejection control method.
The above-described calculation examples of the present invention are merely to describe the calculation model and the calculation flow of the present invention in detail, and are not intended to limit the embodiments of the present invention. It will be apparent to those skilled in the art that other variations and modifications of the present invention can be made based on the above description, and it is not intended to be exhaustive or to limit the invention to the precise form disclosed, and all such modifications and variations are possible and contemplated as falling within the scope of the invention.
Claims (10)
1. The rotor craft disturbance rejection control method based on the adaptive neural network is characterized in that a composite disturbance rejection multi-source disturbance rejection pose controller based on the adaptive neural network is used for realizing disturbance rejection control of the rotor craft;
the design process of the composite multi-source interference resisting pose controller based on the adaptive neural network comprises the following steps:
the method comprises the following steps: constructing a rotorcraft system dynamics model containing lumped external force interference, coupling disturbance influence and internal coupling disturbance:
wherein,respectively representing a position vector and a velocity vector of the unmanned aerial vehicle under an inertial coordinate system; m is s The total mass of a rotor craft system carrying a three-degree-of-freedom light series manipulator driven by a steering engine is shown; f l Tau respectively represents the lifting force and the moment generated by the propeller of the unmanned aerial vehicle; I R B a rotation matrix representing the machine to the inertial system; g represents the gravitational acceleration; e.g. of the type 3 =[0 0 1] T Represents a vertical direction unit vector;respectively representing the internal coupling disturbance forces and forcesMoment; f. of add ,τ add Representing lumped external force interference applied by external environments such as gust, object grabbing interaction and the like in the actual environment; phi b =[φ,θ,ψ]Representing roll, pitch, and yaw angles of the drone; t (phi) b ) A rotation matrix representing a body angular velocity to an attitude angular velocity; B ω b representing an angular velocity vector of the unmanned aerial vehicle body; x represents a cross product operation;representing a rotary inertia matrix I of a rotorcraft system b The inverse of (1);
step two: converting the constructed dynamic model into a state space form;
step three: designing a backstepping control method with a cascade control characteristic as a basic controller based on the underactuation and cascade characteristics of an unmanned aerial vehicle base, wherein the basic controller comprises a basic position controller aiming at a position ring and a basic attitude controller aiming at an attitude ring;
step four: designing the feedforward compensation quantity of the coupling disturbance based on the internal coupling disturbance brought by the motion of the manipulator and the real-time state quantity mapping of the unmanned aerial vehicle and the manipulator:
wherein f is 1 ,f 2 Respectively representing the mapping relation of each state quantity of the unmanned aerial vehicle and the manipulator to the coupling disturbance force and moment m s ,m man Shows the total mass of a rotor craft system carrying three-degree-of-freedom light serial manipulator driven by a steering engine and the mass of the manipulator, I R B representing the rotation matrix of the machine to the inertial system, B R I representing a rotation matrix of the inertial system to the machine system, B r omc representing the representation of the mass center vector of the manipulator in a body coordinate system, B r oc a representation of a vector representing the center of mass of a rotorcraft system in a body coordinate system, B r o representing unmanned aerial vehicle base centroid vector on machineA representation of the body coordinate system is shown, B ω b representing the angular velocity vector of the unmanned aerial vehicle body,expression of the rotary inertia matrix of the representative manipulator relative to the origin of the unmanned aerial vehicle base on the system, ge 3 Representing a gravity vector; f dis , B τ dis Respectively representing feedforward compensation force and moment estimated values obtained based on the coupling disturbance and unmanned aerial vehicle and manipulator state mapping relations;
step five: for extra dynamic disturbance caused by lumped external force interference corresponding to the position ring and the attitude ring, estimating as a disturbance suppression mode based on an adaptive neural network, estimating extra external interference according to the disturbed state deviation of the system, and adding the extra external interference as feedback compensation quantity into a basic controller to realize compensation of the external interference;
step six: obtaining a composite anti-multi-source interference pose controller based on the adaptive neural network by calculating feedforward and feedback compensation quantities; the composite multi-source interference resisting pose controller based on the adaptive neural network comprises a multi-source interference resisting position controller based on the adaptive neural network and a multi-source interference resisting attitude controller based on the adaptive neural network.
2. The adaptive neural network-based rotorcraft disturbance rejection control method according to claim 1, wherein the state space model of step two is as follows:
wherein x is 1 =x,x 3 =y,x 5 = z represents the position in three dimensions of the drone base in three dimensions respectively,respectively representing the speed x in three directions of the three-dimensional space of the base of the unmanned aerial vehicle 7 =φ,x 9 =θ,x 11 ψ represents the roll, pitch and yaw attitude angles of the drone base, x, respectively 8 =p,x 10 =q,x 12 = r represents roll, pitch and yaw attitude angular velocities of the drone base, respectively, m represents rotorcraft system total mass;respectively representing the disturbance forces and moments of internal coupling, J φ ,J θ ,J ψ Respectively represent three-axis rotational inertia of the unmanned aerial vehicle u x ,u y ,u z Respectively representing the control forces in three directions in three dimensions, tau, generated by the lift of the propeller φ ,τ θ ,τ ψ Respectively representing control moments in three directions of the body system, f, generated by the lift of the propeller addx ,f addy ,f addz Respectively representing lumped external disturbance forces f add Components in three directions in three-dimensional space; tau is addφ ,τ addθ ,τ addψ Respectively representing lumped external disturbance moment tau add Components in three directions of the machine system;respectively representing the internal coupling disturbance forcesComponents in three directions in three-dimensional space;respectively representing the internal coupling disturbance torqueThe components in three directions of the machine system.
3. The adaptive neural network-based rotorcraft disturbance rejection control method of claim 2, wherein the design process for the base position controller for the position loop comprises the steps of:
for the position loop, the position error variables are defined as follows:
wherein x d 、y d 、z d Respectively represent expected position signals in three directions of the unmanned aerial vehicle base three-dimensional space,respectively represent expected speed signals k in three directions of three-dimensional space of unmanned aerial vehicle base 1 、k 3 、k 5 Is a controller parameter;
based on the position error variable, the following basic position controller is designed through a Lyapunov function:
the above k1 to k6 are controller parameters and are all positive numbers.
4. The adaptive neural network-based rotorcraft disturbance rejection control method of claim 3, wherein the design process for the fundamental attitude controller for the attitude ring comprises the steps of:
for the attitude ring, the attitude error variables are defined as follows:
wherein phi is d ,θ d ,ψ d Respectively representing the expected attitude angle signals of the unmanned aerial vehicle base;respectively representing the expected attitude angular velocity signals of the unmanned aerial vehicle base; k is a radical of 7 、k 9 、k 11 Is a controller parameter;
based on the attitude error variable, designing a basic attitude controller by a Lyapunov function as follows:
k7 to k12 are controller parameters and are all positive numbers, phi d ,θ d ,ψ d Representing a desired pose.
5. A method for adaptive neural network-based rotorcraft disturbance rejection control according to claim 4, wherein the adaptive neural network that suppresses the additional dynamic disturbance caused by the lumped external force disturbance of the position loop in step five is as follows:
for position loop lumped external interference, three adaptive neural networks are designedWhereinS j (x j ) j=1,2,3 The weights and kernel functions of the radial basis function neural networks are respectively represented and put into the basic position controller to obtain the following form:
location to be added to adaptive neural networkThe controllers being brought intoThe obtained adaptive neural network approximation error is:
wherein f is 1 ,f 2 ,f 3 Respectively represents f addx ,f addy ,f addz ;
According to an online gradient descent algorithm, the update rate of the neural network in the position ring is obtained as follows:
wherein eta is i Is the learning rate of the radial basis function neural network.
6. The adaptive neural network-based rotorcraft disturbance rejection control method of claim 5, wherein the adaptive neural network that suppresses the additional dynamic disturbance caused by the lumped external force disturbance of the attitude ring in step five is as follows:
for lumped external interference of attitude ring, three adaptive neural networks are designedWhereinS j (x j ) j=4,5,6 Weight and kernel function respectively representing each radial basis function are put in basic attitude controlThe following form is obtained:
also, the attitude controllers added into the adaptive neural network are respectively brought intoObtaining:
wherein, tau 4 ,τ 5 ,τ 6 Respectively represent tau addx ,τ addy ,τ addz ;
Then the obtained update rate of the posture ring neural network is as follows:
W i =η i E i S i (X)
wherein eta is i Is the learning rate of the radial basis function neural network.
9. a computer storage medium having stored therein at least one instruction that is loaded and executed by a processor to implement an adaptive neural network-based rotorcraft disturbance rejection control method according to any one of claims 1-8.
10. A rotorcraft disturbance rejection control apparatus based on an adaptive neural network, the apparatus comprising a processor and a memory, the memory having stored therein at least one instruction, the at least one instruction being loaded and executed by the processor to implement the method of adaptive neural network-based rotorcraft disturbance rejection control according to any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211165741.6A CN115826597A (en) | 2022-09-23 | 2022-09-23 | Adaptive neural network-based anti-interference control method and device for rotorcraft |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211165741.6A CN115826597A (en) | 2022-09-23 | 2022-09-23 | Adaptive neural network-based anti-interference control method and device for rotorcraft |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115826597A true CN115826597A (en) | 2023-03-21 |
Family
ID=85523891
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211165741.6A Pending CN115826597A (en) | 2022-09-23 | 2022-09-23 | Adaptive neural network-based anti-interference control method and device for rotorcraft |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115826597A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117666332A (en) * | 2024-02-02 | 2024-03-08 | 北京航空航天大学 | Self-learning anti-interference control method for multi-rotor aircraft in dynamic disturbance environment |
CN117742156A (en) * | 2023-12-28 | 2024-03-22 | 哈尔滨工业大学 | Four-rotor unmanned aerial vehicle control method and system based on RBF neural network |
-
2022
- 2022-09-23 CN CN202211165741.6A patent/CN115826597A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117742156A (en) * | 2023-12-28 | 2024-03-22 | 哈尔滨工业大学 | Four-rotor unmanned aerial vehicle control method and system based on RBF neural network |
CN117742156B (en) * | 2023-12-28 | 2024-06-28 | 哈尔滨工业大学 | Four-rotor unmanned aerial vehicle control method and system based on RBF neural network |
CN117666332A (en) * | 2024-02-02 | 2024-03-08 | 北京航空航天大学 | Self-learning anti-interference control method for multi-rotor aircraft in dynamic disturbance environment |
CN117666332B (en) * | 2024-02-02 | 2024-04-05 | 北京航空航天大学 | Self-learning anti-interference control method for multi-rotor aircraft in dynamic disturbance environment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Robust control of an aerial manipulator based on a variable inertia parameters model | |
CN111766899B (en) | Interference observer-based quad-rotor unmanned aerial vehicle cluster anti-interference formation control method | |
CN112241125B (en) | Unmanned aerial vehicle trajectory tracking method based on differential flatness characteristic | |
CN107562068B (en) | Dynamic surface output regulation control method for attitude of four-rotor aircraft | |
CN115826597A (en) | Adaptive neural network-based anti-interference control method and device for rotorcraft | |
CN110427043B (en) | Pose controller design method based on gravity center offset of operation flying robot | |
CN112346470A (en) | Four-rotor attitude control method based on improved active disturbance rejection control | |
CN112558621A (en) | Decoupling control-based flying mechanical arm system | |
CN109696830A (en) | The reinforcement learning adaptive control method of small-sized depopulated helicopter | |
CN107491083B (en) | Four-rotor-wing autonomous landing method based on saturation self-adaptive sliding mode control | |
CN115556111B (en) | Flight mechanical arm coupling disturbance control method based on variable inertia parameter modeling | |
Ou et al. | Adaptive backstepping tracking control for quadrotor aerial robots subject to uncertain dynamics | |
CN111338369B (en) | Multi-rotor flight control method based on nonlinear inverse compensation | |
Lai et al. | Image dynamics-based visual servo control for unmanned aerial manipulatorl with a virtual camera | |
CN115480583A (en) | Visual servo tracking and impedance control method of flying operation robot | |
CN115431271A (en) | Anti-interference pointing control method for tail end of aircraft mechanical arm | |
CN110275542B (en) | Four-rotor aircraft control method based on self-adaptive finite time control | |
Pedro et al. | Direct adaptive neural control of a quadrotor unmanned aerial vehicle | |
CN111650836B (en) | Control method for dynamically gliding and grabbing object based on operation flying robot | |
CN113580127A (en) | Multi-rotor aircraft humanoid double-mechanical-arm system and dynamic self-balancing control design method thereof | |
CN115657474B (en) | Flexible interaction control method for flying mechanical arm for man-machine cooperative transportation | |
CN116679548A (en) | Three-degree-of-freedom helicopter robust output feedback control method based on time-varying observer | |
CN112034872B (en) | Integral compensation deterministic strategy gradient control method for quad-rotor unmanned aerial vehicle | |
Jiao et al. | Control of quadrotor equipped with a two dof robotic arm | |
Afhami et al. | Updating LQR control for full dynamic of a quadrotor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |