CN116974199A - Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation - Google Patents
Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation Download PDFInfo
- Publication number
- CN116974199A CN116974199A CN202310969774.4A CN202310969774A CN116974199A CN 116974199 A CN116974199 A CN 116974199A CN 202310969774 A CN202310969774 A CN 202310969774A CN 116974199 A CN116974199 A CN 116974199A
- Authority
- CN
- China
- Prior art keywords
- neural network
- discrete
- disturbance
- control
- aircraft
- 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
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 35
- 238000013461 design Methods 0.000 claims description 38
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 10
- 230000033001 locomotion Effects 0.000 claims description 9
- 230000001133 acceleration Effects 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000012938 design process Methods 0.000 claims description 7
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 4
- 230000015556 catabolic process Effects 0.000 claims description 4
- 238000006731 degradation reaction Methods 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 4
- 230000010355 oscillation Effects 0.000 claims description 4
- 230000000087 stabilizing effect Effects 0.000 claims description 4
- 238000005096 rolling process Methods 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 6
- 230000006870 function Effects 0.000 description 9
- 238000004590 computer program Methods 0.000 description 2
- 238000011217 control strategy Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000593 degrading effect Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000010006 flight Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Abstract
The invention discloses a discrete increment sliding mode four-rotor control method based on a neural network and disturbance observation, which belongs to the field of four-rotor unmanned aerial vehicles and comprises the following steps: s1, establishing a six-degree-of-freedom four-rotor unmanned aerial vehicle model by considering aircraft uncertainty factors; s2, designing a disturbance observer, modeling wind power under the wind gust condition, and designing a wind power online predictor based on the disturbance observer and a neural network; s3, designing a sensor error and time delay fault problem solution based on a neural network; s4, designing discrete incremental sliding mode fault-tolerant controllers in the gesture loop and the position loop respectively. The invention has the advantages of strong robustness, high fault tolerance and high-precision control.
Description
Technical Field
The invention relates to the field of four-rotor unmanned aerial vehicles, in particular to a discrete increment sliding mode four-rotor control method based on a neural network and disturbance observation.
Background
The four-rotor aircraft is a multi-rotor aircraft and consists of four symmetrically distributed rotors. The device has the capability of vertical take-off, landing and hovering, so that the device is widely applied to the fields of unmanned aerial vehicles, aerial photography, logistics distribution and the like. In order to achieve accurate control of a quad-rotor aircraft, a number of control methods and algorithms have been proposed. However, in conventional quad-rotor control methods, continuous controllers are typically used for design. These continuous controllers tend to be based on a continuous time model, ignoring the fact that the controller laws are discrete. The design of the discrete controller is more suitable for the discrete property of an actual system, so that the development of a control method for the discrete system has important significance. Discrete incremental sliding mode control (INDI) has good stability in coping with uncertainty of the aircraft model, sudden actuator failure and structural damage. The discrete increment sliding mode control has the advantages of strong robustness, high-precision control, simple realization, high fault tolerance, suitability for discrete systems and the like, but is not mature enough to be applied to the field of four-rotor control.
Meanwhile, the control performance of the quadrotor aircraft can be influenced by factors such as control sensor errors, time delay, faults and the like. Sensor errors may be caused by noise, non-linear characteristics of the sensor itself, or inaccurate calibration, etc., which can affect accurate perception of the system state. Time delay refers to a time delay in signal transmission or processing, such as the time required for signal transfer from the sensor to the controller, which can lead to controller response delays, thereby degrading control performance. In addition, sensor failure or communication failure and other failure conditions can further destroy the reliability and stability of the system.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a discrete increment sliding mode four-rotor wing control method based on a neural network and disturbance observation, which has the advantages of strong robustness, high fault tolerance and high precision control.
The invention aims at realizing the following scheme:
a discrete increment sliding mode four-rotor wing control method based on a neural network and disturbance observation comprises the following steps:
s1, establishing a six-degree-of-freedom four-rotor unmanned aerial vehicle model by considering aircraft uncertainty factors;
s2, designing a disturbance observer, modeling wind power under the wind gust condition, and designing a wind power online predictor based on the disturbance observer and a neural network;
s3, designing a sensor error and time delay fault problem solution based on a neural network;
s4, designing discrete incremental sliding mode fault-tolerant controllers in the gesture loop and the position loop respectively.
Further, the method for establishing the six-degree-of-freedom four-rotor unmanned aerial vehicle model by considering the uncertainty factor of the aircraft specifically comprises the following sub-steps:
the following aircraft dynamics model is established:
wherein x, y, z represent the position coordinates of the quadrotor relative to the inertial frame; θ, φ, ψ represent the Euler angles of the quadrotor, which are pitch angle, roll angle, yaw angle, respectively; p, q, r represent rotational angular velocity of the quadrotor in the body coordinate system; j (J) xb ,J yb ,J zb Representing the inertial constant about the x-axis, y-axis, z-axis of the quadrotor, m representing the mass of the quadrotor, f r Is the resultant force of four rotors, τ xr ,τ yr ,τ zr For the resultant moment of four rotors under the machine body coordinate system, f dx ,f dy ,f dz For interference force τ dx ,τ dy ,τ dz Is a disturbance moment; s (-) represents sin (-), C (-) represents cos (-),representing the acceleration of the aircraft in the inertial frame, < +.>The representation represents the angular acceleration of rotation of the quadrotor in the body coordinate system.
The expression of the forces and moments generated by the rotor is:
wherein ,Ω 1 ,Ω 2 ,Ω 3 ,Ω 4 the rotational speed of rotors 1,2,3,4, k, b are the thrust and moment constants of the quadrotor, L is the length of the arms of the quadrotor, [ ·] T Representing the transpose of the matrix.
Further, in step S2, the neural network is an LSTM network.
Further, in step S2, the model is trained using the historical wind data and related parameters, and the disturbance observer is combined to realize accurate online prediction of the gust wind power.
Further, in step S2, the method includes the sub-steps of: and optimizing the network structure and the training algorithm of the predictor, so as to improve the prediction performance and the robustness.
Further, in step S2, the design perturbation observer comprises the sub-steps of:
consider a MIMO nonlinear control affine system described in the continuous time domain:
wherein ,x∈Rn Is the system state, u e R m Is a system input, y E R m Is the system output, d E R n Is the uncertainty of the model, f 1 :∈R n →R n ，G 1 :∈R m →R n×m ，G 2 :∈R m →R n×m and h:∈Rn →R m Is a continuous function;
the nonlinear disturbance observer NDO is designed using equation (3) to estimate the uncertainty and disturbance of the model in the system, as follows:
wherein ,η, l are the estimates of the disturbance, the internal state of the nonlinear observer, the design observer gain, +.>Representing the derivative of the nonlinear observer internal state eta.
Further, in step S2, the modeling of the wind force under gust conditions comprises the following sub-steps: the wind power model is built as follows:
extending the "1-cos" gust model to symmetrical oscillations described by equationsAnd asymmetry->Gust field:
wherein wm For maximum gust speed lambda x ，λ y Is X E 、Y E Directional gust length.
Further, in step S3, the design of the neural network-based sensor error and time delay fault problem solution specifically includes the following sub-steps: training to obtain a model by utilizing the measured value of the historical sensor and related data, and realizing online compensation of sensor error data and time delay based on online data; when the sensor fails, the neural network model is utilized to predict sensor data for a period of time, so that the aircraft is prevented from being completely out of control due to the sensor failure.
Further, in step S4, designing the attitude loop discrete increment sliding mode fault tolerant controller includes the following sub-steps:
design attitude loop discrete increment sliding mode fault tolerant controller to enable aircraft to track steadilyRoll, yaw and pitch reference values, the control variables y= [ phi, theta, phi are selected] T By x 1 ＝[p,q,r] T ，x 2 ＝[φ,θ,ψ] T ，x＝[x 1 T ,x 2 T ] T Representing the nonlinear model corresponding to the rotational motion of a rigid body aircraft, the control input is represented as u= [ tau ] rx ,τ ry ,τ rz ] T The following expression is obtained:
y＝x 2
wherein ,
considering an actuator failure, the aircraft dynamics is modeled as follows:
wherein, kappa E [0,1]]Designed as a unit step function to indicate sudden actuator failure during flight; y= [ phi, theta, phi ]] T The vector relativity of (2) is ρ= [2,2] T The method comprises the steps of carrying out a first treatment on the surface of the The output dynamics of the aircraft attitude system are:
the method comprises the following steps of:
the new variable kappa is included in equation (6), and y is calculated (2) Is a first order taylor series expansion of (a):
in the formula
The discrete time increment sliding mode control law in the formula (8) is as follows:
in the above-mentioned method, the step of,is a known control matrix; Λ e R m×m Is a diagonal matrix and an unknown time-varying control degradation matrix, and Λ = diag { w } 1 ,w 2 ,...,w m }；/>Is a virtual control for stabilizing an undisturbed system, K epsilon R n×n And design K to A c -B c K is a Hulviz matrix,/and>is->V s (k) Is the disturbance compensation of the design;
design discrete sliding variable sigma (k): R n →R m Is that
σ(k)＝Se(k)-Se(0)+E I (k),E I (k)＝E I (k-1)+K e e(k-1) (10)
in the formula ,Ke e＝-hS(A c -B c K)＝diag{[K i,e ,0,[,0]}，S＝diag{S i }，
Virtual control quantity v of sliding die s (k) The design is as follows:
in the formula ,is a sigmoid function;
next, u=u 0 +Δu indi-s Based on equation (8), the closed loop system dynamics of equation (7) is found under the control input equation (9):
in the formula ,Φs(k) and the following are provided:
further, in step S4, the design position loop discrete incremental sliding mode fault tolerant controller includes the following sub-steps:
when the position controller is designed, compared with the design gesture controller, the following variables are replaced:
selecting a control variable u l ＝[u 1 ,u 2 ,u 3 ] T Selecting the control variable y l ＝[x,y,z] T By usingx l2 ＝[x,y,z] T Representing the translational motion of the nonlinear model corresponding to the rigid body aircraft, the following expression is obtained:
y l ＝x l2
wherein ,
likewise, the obtained position loop discrete increment sliding mode control law is as follows:
the design process of the parameter in the above formula is the same as the design process of the parameter in the gesture loop.
The beneficial effects of the invention include:
the discrete increment sliding mode control based on the disturbance observer has the following advantages: (1) robustness enhancement: the disturbance observer, the incremental dynamic inversion control and the sliding mode control enable the control system to have strong robustness, and the control system still has better performance under the influence of factors such as uncertainty, parameter change, external interference, model error and the like. (2) fault tolerance improvement: the discrete increment sliding mode control method introduces a discretization strategy into a control system, so that the controller has certain fault tolerance capability for the problems of sensor errors, actuator faults, time delay and the like. When the system is in fault or abnormal condition, the discrete increment sliding mode control method can adjust the control instruction in time, and ensures the stability and safety of the system. (3) high-precision control: the discrete increment sliding mode control method can improve the control precision of the system through a proper discretization strategy. By generating and executing accurate discrete control instructions, more accurate and stable four-rotor control can be realized, and the maneuverability and accuracy of the aircraft are improved.
The scheme based on the LSTM predictor has the following advantages: (1) sensor error compensation: LSTM networks can learn and model patterns and characteristics of sensor errors. By training the LSTM network, the control system can perform error compensation according to real-time sensor data, so that the accuracy and control performance of state estimation are improved. (2) delay compensation: LSTM networks can handle latency issues of sensor data. Through learning and memorizing capability of historical data, the LSTM network can predict future sensor data, so that time delay compensation is realized. This helps to improve the real-time and response performance of the control system. (3) fault detection and tolerance: LSTM networks may be used to monitor sensor failures and anomalies. By learning the pattern of normal sensor data, the LSTM network can detect abnormal behavior of the sensor output. Upon detection of a fault, the control system may take appropriate fault-tolerant measures, such as using backup sensors or adapting the control strategy to ensure the stability and safety of the aircraft.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic view of a miniature quad-rotor aircraft;
FIG. 2 is a schematic representation of the relationship between the body coordinate system and the ground coordinate system of a micro quad-rotor aircraft;
FIG. 3 is an LSTM network internal structure; the background color inclusion region in fig. 3 is of particular interest and represents one calculation unit;
FIG. 4 is a control block diagram of a micro quad-rotor aircraft based on the LSTM-D-DOISMC algorithm, with position and attitude controllers being discrete incremental sliding mode controllers based on disturbance observer, and with neural network predictors based on the LSTM algorithm.
Detailed Description
All of the features disclosed in all of the embodiments of this specification, or all of the steps in any method or process disclosed implicitly, except for the mutually exclusive features and/or steps, may be combined and/or expanded and substituted in any way.
In view of the problems in the background, the inventors of the present invention have creatively thought that quadrotors may be affected by a variety of disturbances during flight, which may have an impact on their attitude, stability and control performance. The following are some of the usual four rotor disturbances: 1. disturbance of external wind; 2. pneumatic disturbance; 3. gravity disturbance; 4. load changes. To cope with these disturbances, it is necessary to design appropriate control strategies and algorithms to enable the quadrotors to maintain stable attitude and flying performance in an unstable environment.
In order to solve the technical problems, the embodiment of the invention provides a discrete incremental sliding mode four-rotor wing control method (self-named as LSTM-DOISMC) based on a neural network (LSTM) and a disturbance observer. In the inventive concept, the method takes into account the fact that the controller laws are discrete during the controller design phase and takes full advantage of the time-series modeling capabilities of the neural network (LSTM). By using LSTM, the system state can be predicted and modeled in discrete time, thus obtaining a more accurate state estimate. In the further inventive concept, the controller is designed by adopting an incremental sliding mode control method, and the method can realize the rapid sliding of the system state on the sliding mode surface and keep the system state stable. Compared with the traditional continuous sliding mode control method, the incremental sliding mode control method is more suitable for controlling a discrete system, and can better handle the nonlinearity and uncertainty characteristics of the discrete system.
In order to improve the robustness of the controller and the suppression capability of external disturbance, a disturbance observer is also introduced into the invention. The disturbance observer estimates and compensates external disturbance of the system in real time by observing and compensating for differences between the system output and the controller output. The influence of disturbance on the system can be effectively reduced by introducing the disturbance observer, so that the performance and the robustness of the controller are improved.
In a further inventive concept, in practical applications, the embodiments of the present invention also contemplate control reversing (control reversing) and abrupt actuator failure. Control inversion (Control inversion) refers to a phenomenon that when a designed controller law is not suitable for a specific situation or a system dynamic changes in a Control system, the output of the controller causes the system to become unstable or respond reversely, which is common in four-rotor Control. A sudden actuator failure refers to a situation where one or more actuators (e.g., motors) in the quadrotor suddenly fail or fail during operation. Such faults may result in sudden changes in the output torque or thrust of the actuators, which in turn have an unpredictable effect on the attitude and motion of the quadrotors, which may lead to unstable flights and even accidents.
In summary, the LSTM-DOISMC method provided by the embodiment of the present invention comprehensively considers the discreteness of the controller law, the discrete increment sliding mode control and the design of the disturbance observer, and considers the time-varying characteristics of the control validity matrix, the control inversion, the abrupt actuator failure, and the processing of the problems of sensor error, time delay, failure, and the like. By the method, the four-rotor aircraft can be accurately controlled, and the performance, the robustness and the fault tolerance of the system are improved.
In the specific implementation process, the method comprises the following steps:
(1) Establishing a six-degree-of-freedom four-rotor unmanned aerial vehicle mathematical model, and considering an aircraft uncertainty factor;
(2) Designing a disturbance observer, modeling wind power under the wind gust condition, and designing a wind power online predictor based on the disturbance observer and a neural network (LSTM);
(3) Designing LSTM-based sensor error, time delay and fault problem solutions;
(4) And designing discrete incremental sliding mode fault-tolerant controllers in the gesture loop and the position loop respectively.
The aircraft dynamics model in step (1) is as follows:
wherein x, y, z represent the position coordinates of the quadrotor relative to the inertial frame; θ, φ, ψ represent Euler for a quad-rotor aircraftThe angle is a pitch angle, a roll angle and a yaw angle respectively; p, q, r represent rotational angular velocity of the quadrotor in the body coordinate system; j (J) xb ,J yb ,J zb Representing the inertial constant about the x-axis, y-axis, z-axis of the quadrotor, m representing the mass of the quadrotor, f r Is the resultant force of four rotors, τ xr ,τ yr ,τ zr For the resultant moment of four rotors under the machine body coordinate system, f dx ,f dy ,f dz For interference force τ dx ,τ dy ,τ dz Is a disturbance moment; s (-) represents sin (-), C (-) represents cos (-),representing the acceleration of the aircraft in the inertial frame, < +.>The representation represents the angular acceleration of rotation of the quadrotor in the body coordinate system.
The expression of the forces and moments generated by the rotor is:
wherein ,Ω 1 ,Ω 2 ,Ω 3 ,Ω 4 the rotational speed of rotors 1,2,3,4, k, b are the thrust and moment constants of the quadrotor, L is the length of the arms of the quadrotor, [ ·] T Representing the transpose of the matrix.
The disturbance observer in the step (2) is
Consider a MIMO nonlinear control affine system described in the continuous time domain:
wherein ,x∈Rn Is the system state, u e R m Is a system input, y E R m Is the system output, d E R n Is the uncertainty of the model, f 1 :∈R n →R n ，G 1 :∈R m →R n×m ，G 2 :∈R m →R n×m and h:∈Rn →R m Is a continuous function.
The system (3) designs a Nonlinear Disturbance Observer (NDO) for estimating uncertainty and disturbance of the model in the system, as described below:
wherein ,η, l are the estimates of the disturbance, the internal state of the nonlinear observer, the design observer gain, +.>Representing the derivative of the nonlinear observer internal state eta.
The wind model in step (2) is as follows:
the "1-cos" gust model can be extended to symmetrical oscillations described by the equationAnd asymmetry->A gust field.
wherein wm For maximum gust speed lambda x ，λ y Is X E 、Y E Directional gust length.
The internal structure of the LSTM network in the step (2) is shown in figure 3, and the accurate online prediction of the wind power of the gust is realized by utilizing the historical wind power data and a relevant parameter training model and combining a disturbance observer. And the network structure and the training algorithm of the predictor are optimized, and the prediction performance and the robustness are improved.
The internal structure of the LSTM network in the step (3) is as shown in fig. 3, a model is obtained by training the measured values of the historical sensors and related data, on-line compensation of sensor error data and time delay is realized based on-line data, and when the sensor breaks down, the neural network model can predict sensor data for a period of time, so that the aircraft is prevented from being completely out of control due to the sensor fault.
The attitude loop discrete increment sliding mode fault-tolerant controller designed in the step (4) is as follows:
the attitude loop discrete increment sliding mode fault-tolerant controller is designed to enable the aircraft to stably track rolling, yaw and pitch angle reference values, so that the control variables y= [ phi, theta, phi are selected] T By x 1 ＝[p,q,r] T ，x 2 ＝[φ,θ,ψ] T ，x＝[x 1 T ,x 2 T ] T Representing the nonlinear model corresponding to the rotational motion of a rigid body aircraft, the control input is represented as u= [ tau ] rx ,τ ry ,τ rz ] T The following expression can be obtained:
y＝x 2
wherein ,
considering an actuator failure, aircraft dynamics can be modeled as follows:
wherein κ ε [0,1] is designed as a unit step function to indicate a sudden actuator failure during flight.
y＝[φ,θ,ψ] T The vector relativity of (2) is ρ= [2,2] T . Thus, the output dynamics of the aircraft attitude system are:
thus, it is possible to obtain:
the new variable kappa is included in equation (6), and y is calculated (2) Is a first order taylor series expansion of (a):
in the formula
The discrete time increment sliding mode control law in the system (8) is as follows:
in the above-mentioned method, the step of,is a known control matrix; Λ e R m×m Is a diagonal matrix and an unknown time-varying control degradation matrix, and Λ = diag { w } 1 ,w 2 ,...,w m }。/>Is a virtual control for stabilizing an undisturbed system, K epsilon R n×n And design K to A c -B c K is a Hulviz matrix,/and>is->V s (k) Is the disturbance compensation of the design.
Design discrete sliding variable sigma (k): R n →R m Is that
σ(k)＝Se(k)-Se(0)+E I (k),E I (k)＝E I (k-1)+K e e(k-1)(10)
in the formula ,Ke e＝-hS(A c -B c K)＝diag{[K i,e ,0,...,0]}，S＝diag{S i }，
Thus, the sliding die virtual control amount v s (k) Designed as
in the formula ,is a sigmoid function.
Next, u=u 0 +Δu indi-s Based on equation (8), the closed loop system dynamics of equation (7) is found under the control input equation (9):
in the formula ,Φs(k) and the following is shown:
the position loop discrete increment sliding mode fault-tolerant controller designed in the step (4) is as follows:
the position controller is designed to be similar to the attitude controller and is replaced by the following variables
Selecting a control variable u l ＝[u 1 ,u 2 ,u 3 ] T Selecting the control variable y l ＝[x,y,z] T By usingx l2 ＝[x,y,z] T Representing that the nonlinear model corresponds to translational motion of a rigid body aircraft, the following expression can be obtained:
y l ＝x l2
wherein ,
also, the position loop discrete increment sliding mode control law can be obtained as follows:
note that: the design process of the parameters in the above formula is as a gesture loop.
It should be noted that, within the scope of protection defined in the claims of the present invention, the following embodiments may be combined and/or expanded, and replaced in any manner that is logical from the above specific embodiments, such as the disclosed technical principles, the disclosed technical features or the implicitly disclosed technical features, etc.
Example 1
A discrete increment sliding mode four-rotor wing control method based on a neural network and disturbance observation comprises the following steps:
s1, establishing a six-degree-of-freedom four-rotor unmanned aerial vehicle model by considering aircraft uncertainty factors;
s2, designing a disturbance observer, modeling wind power under the wind gust condition, and designing a wind power online predictor based on the disturbance observer and a neural network;
s3, designing a sensor error and time delay fault problem solution based on a neural network;
s4, designing discrete incremental sliding mode fault-tolerant controllers in the gesture loop and the position loop respectively.
Example 2
Based on embodiment 1, the method for establishing the six-degree-of-freedom four-rotor unmanned aerial vehicle model by taking the uncertainty factor of the aircraft into consideration specifically comprises the following substeps:
the following aircraft dynamics model is established:
/>
wherein x, y, z represent the position coordinates of the quadrotor relative to the inertial frame; θ, φ, ψ represent the Euler angles of the quadrotor, which are pitch angle, roll angle, yaw angle, respectively; p, q, r represent rotational angular velocity of the quadrotor in the body coordinate system; j (J) xb ,J yb ,J zb Representing the inertial constant about the x-axis, y-axis, z-axis of the quadrotor, m representing the mass of the quadrotor, f r Is the resultant force of four rotors, τ xr ,τ yr ,τ zr For the resultant moment of four rotors under the machine body coordinate system, f dx ,f dy ,f dz For interference force τ dx ,τ dy ,τ dz Is a disturbance moment; s (-) represents sin (-), C (-) represents cos (-),representing the acceleration of the aircraft in the inertial frame, < +.>The representation represents the angular acceleration of rotation of the quadrotor in the body coordinate system.
The expression of the forces and moments generated by the rotor is:
wherein ,Ω 1 ,Ω 2 ,Ω 3 ,Ω 4 the rotational speed of rotors 1,2,3,4, k, b are the thrust and moment constants of the quadrotor, L is the length of the arms of the quadrotor, [ ·] T Representing the transpose of the matrix.
Example 3
On the basis of embodiment 1, in step S2, the neural network is an LSTM network.
Example 4
Based on embodiment 1, in step S2, the model is trained using the historical wind data and related parameters, and the accurate online prediction of the gust wind power is achieved in combination with the disturbance observer.
Example 5
On the basis of embodiment 1, in step S2, the sub-steps are included: and optimizing the network structure and the training algorithm of the predictor, so as to improve the prediction performance and the robustness.
Example 6
On the basis of embodiment 1, in step S2, the design perturbation observer comprises the sub-steps of:
consider a MIMO nonlinear control affine system described in the continuous time domain:
wherein ,x∈Rn Is the system state, u e R m Is a system input, y E R m Is the system output, d E R n Is the uncertainty of the model, f 1 :∈R n →R n ，G 1 :∈R m →R n×m ，G 2 :∈R m →R n×m and h:∈Rn →R m is a continuous function;
the nonlinear disturbance observer NDO is designed using equation (3) to estimate the uncertainty and disturbance of the model in the system, as follows:
wherein ,η, l are the estimates of the disturbance, the internal state of the nonlinear observer, the design observer gain, +.>Representing the derivative of the nonlinear observer internal state eta.
Example 7
On the basis of embodiment 1, in step S2, the modeling of the wind force in gust conditions comprises the following sub-steps: the wind power model is built as follows:
extending the "1-cos" gust model to symmetrical oscillations described by equationsAnd asymmetry->Gust field:
wherein wm For maximum gust speed lambda x ，λ y Is X E 、Y E Directional gust length.
Example 8
On the basis of embodiment 1, in step S3, the design of a neural network-based sensor error, time delay fault problem solution specifically includes the following sub-steps: training to obtain a model by utilizing the measured value of the historical sensor and related data, and realizing online compensation of sensor error data and time delay based on online data; when the sensor fails, the neural network model is utilized to predict sensor data for a period of time, so that the aircraft is prevented from being completely out of control due to the sensor failure.
Example 9
On the basis of embodiment 1, in step S4, the design gesture loop discrete incremental sliding mode fault tolerant controller comprises the following sub-steps:
designing a discrete increment sliding mode fault-tolerant controller of a gesture loop, enabling an aircraft to stably track rolling, yaw and pitch angle reference values, and selecting control variables y= [ phi, theta and phi] T By x 1 ＝[p,q,r] T ，x 2 ＝[φ,θ,ψ] T ，x＝[x 1 T ,x 2 T ] T Representing the nonlinear model corresponding to the rotational motion of a rigid body aircraft, the control input is represented as u= [ tau ] rx ,τ ry ,τ rz ] T The following expression is obtained:
y＝x 2
wherein ,
considering an actuator failure, the aircraft dynamics is modeled as follows:
wherein, kappa E [0,1]]Designed as a unit step function to indicate sudden actuator failure during flight; y= [ phi, theta, phi ]] T The vector relativity of (2) is ρ= [2,2] T The method comprises the steps of carrying out a first treatment on the surface of the The output dynamics of the aircraft attitude system are:
the method comprises the following steps of:
the new variable kappa is included in equation (6), and y is calculated (2) Is a first order taylor series expansion of (a):
in the formula
The discrete time increment sliding mode control law in the formula (8) is as follows:
in the above-mentioned method, the step of,is a known control matrix; Λ e R m×m Is a diagonal matrix and an unknown time-varying control degradation matrix, and Λ = diag { w } 1 ,w 2 ,...,w m }；/>Is a virtual control for stabilizing an undisturbed system, K epsilon R n×n And design K to A c -B c K is a Hulviz matrix,/and>is->V s (k) Is the disturbance compensation of the design;
design discrete sliding variable sigma (k): R n →R m Is that
σ(k)＝Se(k)-Se(0)+E I (k),E I (k)＝E I (k-1)+K e e(k-1) (10)
in the formula ,Ke e＝-hS(A c -B c K)＝diag{[K i,e ,0,...,0]}，S＝diag{S i }，
Virtual control quantity v of sliding die s (k) The design is as follows:
in the formula ,is a sigmoid function;
next, u=u 0 +Δu indi-s Based on equation (8), the closed loop system dynamics of equation (7) is found under the control input equation (9):
in the formula ,Φs(k) and the following are provided:
example 10
On the basis of embodiment 1, in step S4, the design position loop discrete incremental sliding mode fault tolerant controller comprises the following sub-steps:
when the position controller is designed, compared with the design gesture controller, the following variables are replaced:
selecting a control variable u l ＝[u 1 ,u 2 ,u 3 ] T Selecting the control variable y l ＝[x,y,z] T By usingx l2 ＝[x,y,z] T Representing the translational motion of the nonlinear model corresponding to the rigid body aircraft, the following expression is obtained:
y l ＝x l2
wherein ,
likewise, the obtained position loop discrete increment sliding mode control law is as follows:
the design process of the parameter in the above formula is the same as the design process of the parameter in the gesture loop.
The units involved in the embodiments of the present invention may be implemented by software, or may be implemented by hardware, and the described units may also be provided in a processor. Wherein the names of the units do not constitute a limitation of the units themselves in some cases.
According to an aspect of embodiments of the present invention, there is provided a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The computer instructions are read from the computer-readable storage medium by a processor of a computer device, and executed by the processor, cause the computer device to perform the methods provided in the various alternative implementations described above.
As another aspect, the embodiment of the present invention also provides a computer-readable medium that may be contained in the electronic device described in the above embodiment; or may exist alone without being incorporated into the electronic device. The computer-readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to implement the methods described in the above embodiments.
Claims (10)
1. A discrete increment sliding mode four-rotor control method based on a neural network and disturbance observation is characterized by comprising the following steps:
s1, establishing a six-degree-of-freedom four-rotor unmanned aerial vehicle model by considering aircraft uncertainty factors;
s2, designing a disturbance observer, modeling wind power under the wind gust condition, and designing a wind power online predictor based on the disturbance observer and a neural network;
s3, designing a sensor error and time delay fault problem solution based on a neural network;
s4, designing discrete incremental sliding mode fault-tolerant controllers in the gesture loop and the position loop respectively.
2. The discrete incremental sliding-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein the four-rotor unmanned aerial vehicle model with six degrees of freedom is built by considering the uncertainty factors of the aircraft, and specifically comprises the following sub-steps:
the following aircraft dynamics model is established:
wherein x, y, z represent the position coordinates of the quadrotor relative to the inertial frame; θ, φ, ψ represent a quad-rotor flightEuler angles of the device are pitch angle, roll angle and yaw angle respectively; p, q, r represent rotational angular velocity of the quadrotor in the body coordinate system; j (J) xb ,J yb ,J zb Representing the inertial constant about the x-axis, y-axis, z-axis of the quadrotor, m representing the mass of the quadrotor, f r Is the resultant force of four rotors, τ xr ,τ yr ,τ zr For the resultant moment of four rotors under the machine body coordinate system, f dx ,f dy ,f dz For interference force τ dx ,τ dy ,τ dz Is a disturbance moment; s (·) represents sin (·), C () represents cos (·),representing the acceleration of the aircraft in the inertial frame, < +.>The representation represents the angular acceleration of rotation of the quadrotor in the body coordinate system.
The expression of the forces and moments generated by the rotor is:
wherein ,Ω 1 ,Ω 2 ,Ω 3 ,Ω 4 the rotational speed of rotors 1,2,3,4, k, b are the thrust and moment constants of the quadrotor, L is the length of the arms of the quadrotor, [ ·] T Representing the transpose of the matrix.
3. The discrete delta slip-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S2, the neural network is an LSTM network.
4. The discrete incremental sliding-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S2, the model is trained by using historical wind data and related parameters, and the disturbance observer is combined to realize accurate online prediction of wind power of gusts.
5. The discrete delta slip-mode four-rotor control method based on neural network and disturbance observation according to claim 1, comprising the sub-steps of, in step S2: and optimizing the network structure and the training algorithm of the predictor, so as to improve the prediction performance and the robustness.
6. The discrete delta slip-mode four-rotor control method based on neural network and disturbance observer according to claim 1, wherein in step S2, the design disturbance observer comprises the sub-steps of:
consider a MIMO nonlinear control affine system described in the continuous time domain:
wherein ,x∈Rn Is the system state, u e R m Is a system input, y E R m Is the system output, d E R n Is the uncertainty of the model, f 1 :∈R n →R n ，G 1 :∈R m →R n×m ，G 2 :∈R m →R n×m and h:∈Rn →R m Is a continuous function;
the nonlinear disturbance observer NDO is designed using equation (3) to estimate the uncertainty and disturbance of the model in the system, as follows:
wherein ,η and l are disturbance respectivelyIs a non-linear observer, designs observer gain, +.>Representing the derivative of the nonlinear observer internal state eta.
7. The discrete delta slip-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S2, the modeling of wind power under gust conditions comprises the sub-steps of: the wind power model is built as follows:
extending the "1-cos" gust model to symmetrical oscillations described by equationsAnd asymmetry->Gust field:
wherein wm For maximum gust speed lambda x ，λ y Is X E 、Y E Directional gust length.
8. The discrete incremental sliding-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S3, the design of a neural network-based sensor error, time delay fault problem solution specifically comprises the sub-steps of: training to obtain a model by utilizing the measured value of the historical sensor and related data, and realizing online compensation of sensor error data and time delay based on online data; when the sensor fails, the neural network model is utilized to predict sensor data for a period of time, so that the aircraft is prevented from being completely out of control due to the sensor failure.
9. The discrete delta slip-form four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S4, designing a gesture loop discrete delta slip-form fault-tolerant controller comprises the following sub-steps:
designing a discrete increment sliding mode fault-tolerant controller of a gesture loop, enabling an aircraft to stably track rolling, yaw and pitch angle reference values, and selecting control variables y= [ phi, theta and phi] T By x 1 ＝[p,q,r] T ，x 2 ＝[φ,θ,ψ] T ，x＝[x 1 T ,x 2 T ] T Representing the nonlinear model corresponding to the rotational motion of a rigid body aircraft, the control input is represented as u= [ tau ] rx ,τ ry ,τ rz ] T The following expression is obtained:
y＝x 2
wherein ,
considering an actuator failure, the aircraft dynamics is modeled as follows:
wherein, kappa E [0,1]]Designed as a unit step function to indicate sudden actuator failure during flight; y= [ phi, theta, phi ]] T The vector relativity of (2) is ρ= [2,2] T The method comprises the steps of carrying out a first treatment on the surface of the The output dynamics of the aircraft attitude system are:
the method comprises the following steps of:
the new variable kappa is included in equation (6), and y is calculated (2) Is a first order taylor series expansion of (a):
in the formula
The discrete time increment sliding mode control law in the formula (8) is as follows:
in the above-mentioned method, the step of,is a known control matrix; Λ e R m×m Is a diagonal matrix and an unknown time-varying control degradation matrix, and Λ = diag { w } 1 ,w 2 ,...,w m }；/>Is a virtual control for stabilizing an undisturbed system, K epsilon R n×n And design K to A c -B c K is a Hulviz matrix,/and>is->V s (k) Is the disturbance compensation of the design;
design discrete sliding variable sigma (k): R n →R m Is that
σ(k)＝Se(k)-Se(0)+E I (k),E I (k)＝E I (k-1)+K e e(k-1) (10)
in the formula ,Ke e＝-hS(A c -B c K)＝diag{[K i,e ,0,...,0]}，S＝diag{S i }，
Virtual control quantity v of sliding die s (k) The design is as follows:
in the formula ,is a sigmoid function;
next, u=u 0 +Δu indi-s Based on equation (8), the closed loop system dynamics of equation (7) is found under the control input equation (9):
in the formula ,Φs(k) and the following are provided:
10. the discrete delta slip-mode four-rotor control method based on neural network and disturbance observation according to claim 1, wherein in step S4, designing a position loop discrete delta slip-mode fault-tolerant controller comprises the following sub-steps:
when the position controller is designed, compared with the design gesture controller, the following variables are replaced:
selecting a control variable u l ＝[u 1 ,u 2 ,u 3 ] T Selecting the control variable y l ＝[x,y,z] T By usingx l2 ＝[x,y,z] T Representing the translational motion of the nonlinear model corresponding to the rigid body aircraft, the following expression is obtained:
y l ＝x l2
wherein ,
likewise, the obtained position loop discrete increment sliding mode control law is as follows:
the design process of the parameter in the above formula is the same as the design process of the parameter in the gesture loop.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310969774.4A CN116974199A (en) | 2023-08-02 | 2023-08-02 | Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310969774.4A CN116974199A (en) | 2023-08-02 | 2023-08-02 | Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116974199A true CN116974199A (en) | 2023-10-31 |
Family
ID=88481167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310969774.4A Pending CN116974199A (en) | 2023-08-02 | 2023-08-02 | Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116974199A (en) |
-
2023
- 2023-08-02 CN CN202310969774.4A patent/CN116974199A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hou et al. | Nonsingular terminal sliding mode control for a quadrotor UAV with a total rotor failure | |
| Yang et al. | Adaptive NN backstepping control design for a 3-DOF helicopter: Theory and experiments | |
| Gao et al. | Active fault tolerant control design for reusable launch vehicle using adaptive sliding mode technique | |
| Derafa et al. | Super twisting control algorithm for the attitude tracking of a four rotors UAV | |
| Jiang et al. | Adaptive neural observer‐based backstepping fault tolerant control for near space vehicle under control effector damage | |
| Dong et al. | Adaptive disturbance observer‐based finite‐time continuous fault‐tolerant control for reentry RLV | |
| CN110347170B (en) | Reusable carrier reentry segment robust fault-tolerant guidance control system and working method | |
| CN111880410B (en) | Four-rotor unmanned aerial vehicle fault-tolerant control method for motor faults | |
| Lin et al. | Neural‐network‐based robust terminal sliding‐mode control of quadrotor | |
| Xu et al. | Decentralized asymptotic fault tolerant control of near space vehicle with high order actuator dynamics | |
| Tripathi et al. | Finite‐time super twisting sliding mode controller based on higher‐order sliding mode observer for real‐time trajectory tracking of a quadrotor | |
| Huang et al. | Generic adaptive sliding mode control for a quadrotor UAV system subject to severe parametric uncertainties and fully unknown external disturbance | |
| CN113296525B (en) | Data-driven-based non-linear control method for tilting three-rotor unmanned aerial vehicle | |
| CN111781942B (en) | Fault-tolerant flight control method based on self-constructed fuzzy neural network | |
| Huang et al. | Finite-time control for a UAV system based on finite-time disturbance observer | |
| Ortiz-Torres et al. | An actuator fault detection and isolation method design for planar vertical take-off and landing unmanned aerial vehicle modelled as a qLPV system | |
| Azid et al. | Wind gust estimation for precise quasi-hovering control of quadrotor aircraft | |
| CN113568423B (en) | Intelligent fault-tolerant control method of four-rotor unmanned aerial vehicle considering motor faults | |
| CN115248553A (en) | Event triggering adaptive PID track tracking fault-tolerant control method for under-actuated ship | |
| Xian et al. | Data driven adaptive robust attitude control for a small size unmanned helicopter | |
| Meng et al. | Adaptive non‐singular fault‐tolerant control for hypersonic vehicle with unexpected centroid shift | |
| CN116974199A (en) | Discrete increment sliding mode four-rotor wing control method based on neural network and disturbance observation | |
| Okada et al. | Neuroadaptive Observer-Based Fault-Diagnosis and Fault-Tolerant Control for Quadrotor UAV | |
| Li et al. | Fault-tolerant aircraft control based on self-constructing fuzzy neural network for quadcopter | |
| Emami et al. | Disturbance observer-based adaptive neural guidance and control of an aircraft using composite learning |
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 |