A method of aircraft attitude prediction control with adaptive disturbance observation compensation

Abstract

A method for predictive control of aircraft attitude with adaptive disturbance observation compensation belongs to the field of aircraft control technology. The method is as follows: constructing an aircraft attitude dynamics and kinematics model, and forming an attitude control error model by subtracting it from the attitude command; designing disturbance observers with adaptive observation gain for the three channels of aircraft pitch, yaw and roll respectively; using the disturbance observation value, performing variable gain compensation control to obtain the compensation control quantity; defining an integral performance index, using a Critic network with a variable prediction period to predict the index, and updating the Actor network based on the predicted value to obtain an approximate optimal control quantity; adding the compensation control quantity to the approximate optimal control quantity to obtain the total attitude control quantity of the aircraft. The variable gain disturbance observation compensation method can improve the aircraft's accurate observation and compensation of various types of interference during flight, and improve flight stability.

Description

Technical Field

[0001] The invention belongs to the technical field of aircraft control, and relates to an aircraft attitude prediction control method with adaptive disturbance observation compensation. Background Art

[0002] During the execution of the mission, the aircraft will be subject to various types of disturbances such as aerodynamic interference and wind interference due to the complexity of the flight environment, which will seriously affect the accuracy and stability of attitude control. In order to achieve high-precision control, the aircraft needs to be able to adaptively overcome the influence of various disturbances to ensure flight stability. At the same time, in order to achieve high-precision control as much as possible under limited control resources, the aircraft's attitude control law needs to be optimal. Accurately predicting and evaluating the pros and cons of the current control law based on flight data is the key to guiding the online optimization of flight control. Summary of the invention

[0003] In order to solve the problem in the background technology of adaptively overcoming the influence of disturbances and accurately predicting and evaluating the quality of control laws, the present invention provides an aircraft attitude prediction control method with adaptive disturbance observation compensation.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] An aircraft attitude prediction control method with adaptive disturbance observation compensation, the method comprising:

[0006] Step 1: Construct the aircraft attitude dynamics and kinematics model, and form the attitude control error model by subtracting it from the attitude command;

[0007] Step 2: For the three channels of aircraft pitch, yaw and roll, design disturbance observers with adaptive observation gains respectively;

[0008] Step 3: Use the disturbance observation value to perform variable gain compensation control to obtain the compensation control amount;

[0009] Step 4: Define an integral performance indicator, use a Critic network with a variable prediction period to predict the indicator, and update the Actor network based on the predicted value to obtain an approximate optimal control amount;

[0010] Step 5: Add the compensation control amount to the approximate optimal control amount to obtain the total attitude control amount of the aircraft. The total attitude control amount of the aircraft is u = U a +U b .

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: the variable gain disturbance observation compensation method can improve the aircraft's accurate observation and compensation of various types of interference during flight, thereby improving flight stability; the predictive control method with a variable prediction period can integrate the aircraft's current data and data at historical sampling times to improve the prediction accuracy of performance indicators, thereby improving the optimality of control, thereby reducing control deviations and saving control energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flow chart of an aircraft attitude prediction control method with adaptive disturbance observation compensation. DETAILED DESCRIPTION

[0013] The technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0014] Embodiment 1:

[0015] An adaptive disturbance observation compensation method for aircraft attitude prediction control, such as Figure 1 As shown, the method is:

[0016] Step 1: construct the aircraft attitude dynamics and kinematics model, and form an attitude control error model by subtracting it from the attitude command; the specific steps of step 1 are:

[0017] Step 1: The aircraft attitude dynamics and kinematics model is:

[0018]

[0019] In formula (1), is the first-order derivative of Ω with respect to time; is the attitude angle vector of the aircraft; is the pitch angle; ψ is the yaw angle; γ is the roll angle;

[0020] R is the attitude transformation matrix,

[0021] is the first-order derivative of ω with respect to time; ω=[ω x ,ω y ,ω z ] is the attitude angular velocity vector of the aircraft; ω x is the rolling angular velocity; ω y is the yaw angular velocity; ω zis the pitch angular velocity; J represents the moment of inertia matrix of the aircraft; ω × represents the cross product matrix of ω; B 1 represents the control torque coefficient matrix; δ=[δ x ,δ y ,δ z ] represents the control input; δ x is the deflection angle of the aileron; δ y is the deflection angle of the rudder; δ z is the deflection angle of the elevator; d = [d x ,d y ,d z ] is the aerodynamic torque and disturbance torque term; d x is the aerodynamic torque and disturbance torque term acting in the rolling direction; d y is the aerodynamic torque and disturbance torque term acting in the yaw direction; d z are the aerodynamic moment and disturbance moment terms acting in the pitch direction;

[0022] Step 1 and 2: Set the aircraft attitude angle change command

[0023]

[0024] In formula (2): Ω cx is the roll channel command; Ω cy is the yaw channel command; Ω cz It is the pitch channel command;

[0025] The attitude angle tracking error is defined as

[0026]

[0027] In formula (3): is the pitch angle tracking error; x 1ψ is the yaw angle tracking error; x 1γ is the rolling angle tracking error;

[0028] By continuously calculating the second-order derivatives of equation (3), we can get the attitude control error model:

[0029]

[0030] In formula (4): Represents x 1 The first derivative with respect to time; x 2 represents the first-order derivative of the attitude angle tracking error, and is the first-order derivative of the pitch angle tracking error; x 2ψ is the first-order derivative of the yaw angle tracking error; x 2γ is the first-order derivative of the roll angle tracking error; Represents x 1 The second-order derivative with respect to time; u is the control variable, and u=RJ -1 B 1 δ=[u x ,u y ,u z ];u x is the rolling direction control quantity, which will be given by step 5; u y is the yaw direction control quantity, which will be given by step 5; u z is the pitch direction control quantity, which will be given by step 5; H is the total disturbance, and H x is the total disturbance in the rolling direction; H y is the total disturbance in the yaw direction; H z is the total disturbance in the pitch direction;

[0031] Step 2: For the three channels of pitch, yaw and roll of the aircraft, a disturbance observer with adaptive observation gain is designed respectively; the specific step 2 is as follows:

[0032] Step 21: Taking the pitch channel as an example, design an extended state observer of the following form

[0033]

[0034] In formula (5): e z For the observer The estimation error of For estimates; For estimates; For H z Estimate of β 01z is the adaptive gain coefficient of the first-order correction term of the pitch channel extended state observer; β 02z is the adaptive gain coefficient of the second-order correction term of the pitch channel expansion state observer; β 03z is the adaptive gain coefficient of the third-order correction term of the pitch channel expanded state observer;

[0035] β 01z ,β 02z With β 03z The expression is

[0036]

[0037] In formula (6): k sz >0 is the proportional coefficient, ò∈(1,1.5) is the power coefficient, δ 0iz >0, i=1,2,3 are threshold values;

[0038] The expanded state observer used in the yaw and roll channels is the same as that in the pitch channel. The expanded state observer of the yaw channel is

[0039]

[0040] In formula (7): e y is the observer pair x 1ψ The estimation error of z 1ψ For x 1ψ Estimates of 2ψ For x 2ψ Estimate of 3ψ For H y Estimate of β 01y is the adaptive gain coefficient of the first-order correction term of the yaw channel extended state observer; β 02y is the adaptive gain coefficient of the second-order correction term of the yaw channel extended state observer; β 03y is the adaptive gain coefficient of the third-order correction term of the expanded state observer of the yaw channel;

[0041] β 01y ,β 02y With β 03y The expression is

[0042]

[0043] In formula (8): k sy >0 is the proportionality coefficient, δ 0iy >0, i=1,2,3 are threshold values;

[0044] The extended state observer of the rolling channel is

[0045]

[0046] In formula (9): e x is the observer pair x 1γ The estimation error of z 1γ For x 1γ Estimate of 2γ For x 2γ Estimate of 3γ For H x Estimate of β 01x is the adaptive gain coefficient of the first-order correction term of the roll channel expansion state observer; β 02x is the adaptive gain coefficient of the second-order correction term of the roll channel expansion state observer; β 03x is the adaptive gain coefficient of the third-order correction term of the roll channel expansion state observer;

[0047] The outputs of the expanded state observer for the final pitch, yaw, and roll channels are defined as:

[0048] Step 3: Using the disturbance observation value, variable gain compensation control is performed to obtain the compensation control amount; the step 3 is specifically as follows:

[0049] Step 31: Let the gain matrix of compensation control be in is the compensation control gain of the pitch channel, K bψ is the compensation control gain of the yaw channel, K bγ is the compensation control gain of the roll channel;

[0050] K bψ With K bγ The expression is

[0051]

[0052] In formula (10): ε∈(0.5,1) is the power coefficient;

[0053] Step 32: Multiply the gain of the compensation control by the disturbance observation value to obtain the compensation control amount U b =K b Z 3 ;

[0054] Step 4: Define an integral performance index, use a Critic network with a variable prediction period to predict the index, and update the Actor network based on the predicted value to obtain an approximate optimal control amount; Step 4 is specifically as follows:

[0055] Step 41: Define the integral performance index V

[0056]

[0057] In formula (11): t 0 =0 is the initial time of integration; τ is the integration variable; Q x With R u are all positive definite matrices; x 1 (τ) represents the time x at τ 1 The value of; U(τ) represents the value of U at time τ;

[0058] Step 42: Critic network design

[0059]

[0060] In formula (12): It is the estimated value of the integral performance indicator by the Critic network; is the critic network weight; φ(x1 ) is the activation function;

[0061] Step 43: Design the weight update law of the variable prediction cycle of the Critic network as

[0062]

[0063] In formula (13): is the derivative of the Critic network weight; η c >0 is the learning rate; Γ is the positive definite gain matrix; ω c is the estimated residual of the Critic network at the current sampling moment; ω ci is the estimated residual of the Critic network in the i-th prediction cycle; υ>0 is the regularization gain; The subscript i represents the value of the variable at the i-th prediction period; k is the adaptive prediction period;

[0064] Step 44: Define the regression matrix and adaptively adjust the prediction period according to the rank of the regression matrix

[0065] Define the regression matrix At each sampling moment, calculate whether the regression matrix is ​​full rank; if it is full rank, k does not need to be adjusted; if it is not full rank, at the next sampling moment, k = k + 1;

[0066] Step 4 and 5: Design the Actor network and iteratively update it to obtain the approximate optimal control amount

[0067] Actor networks are designed to in is the weight of the Actor network; the iterative update target of the Actor network is to minimize the normalized prediction error E c (t):

[0068]

[0069] Approximate optimal control quantity U a for

[0070]

[0071] Step 5: Add the compensation control amount to the approximate optimal control amount to obtain the total attitude control amount of the aircraft. The total attitude control amount of the aircraft is u = U a +U b .

[0072] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0073] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. An aircraft attitude prediction control method with adaptive disturbance observation compensation, characterized in that: The method is: Step 1: Construct the aircraft attitude dynamics and kinematics model, and form the attitude control error model by subtracting it from the attitude command; Step 2: For the three channels of aircraft pitch, yaw and roll, design disturbance observers with adaptive observation gains respectively; Step 3: Use the disturbance observation value to perform variable gain compensation control to obtain the compensation control amount; Step 4: Define an integral performance indicator, use a Critic network with a variable prediction period to predict the indicator, and update the Actor network based on the predicted value to obtain an approximate optimal control amount; Step 5: Add the compensation control amount to the approximate optimal control amount to obtain the total attitude control amount u = U of the aircraft a +U b .

2. The method of aircraft attitude prediction control with adaptive disturbance observation compensation according to claim 1, characterized in that: The step 1 is specifically as follows: Step 1: The aircraft attitude dynamics and kinematics model is: In formula (1), is the first-order derivative of Ω with respect to time; is the attitude angle vector of the aircraft; is the pitch angle; ψ is the yaw angle; γ is the roll angle; R is the attitude transformation matrix, is the first-order derivative of ω with respect to time; ω=[ω x ,ω y ,ω z ] is the attitude angular velocity vector of the aircraft; ω x is the rolling angular velocity; ω y is the yaw angular velocity; ω z is the pitch angular velocity; J represents the moment of inertia matrix of the aircraft; ω × represents the cross product matrix of ω; B1 represents the control torque coefficient matrix; δ=[δ x ,δ y ,δ z ] represents the control input; δ x is the deflection angle of the aileron; δ y is the deflection angle of the rudder; δ z is the deflection angle of the elevator; d = [d x ,d y ,d z ] is the aerodynamic torque and disturbance torque term; d x are the aerodynamic torque and disturbance torque terms acting in the rolling direction; d y are the aerodynamic torque and disturbance torque terms acting in the yaw direction; d z are the aerodynamic moment and disturbance moment terms acting in the pitch direction; Step 1 and 2: Set the aircraft attitude angle change command Oh c =[Oh cz ,Oh cy ,Oh cx ] (2) In formula (2): Ω cx is the roll channel command; Ω cy is the yaw channel command; Ω cz It is the pitch channel command; The attitude angle tracking error is defined as In formula (3): is the pitch angle tracking error; x 1ψ is the yaw angle tracking error; x 1γ is the rolling angle tracking error; By continuously calculating the second-order derivatives of equation (3), we can get the attitude control error model: In formula (4): represents the first-order derivative of x1 with respect to time; x2 represents the first-order derivative of the attitude angle tracking error, and is the first-order derivative of the pitch angle tracking error; x 2ψ is the first-order derivative of the yaw angle tracking error; x 2γ is the first-order derivative of the roll angle tracking error; represents the second-order derivative of x1 with respect to time; u is the control variable, and u=RJ -1 B1δ=[u x ,u y ,u z ];u x is the rolling direction control amount; u y is the yaw direction control quantity; u z is the pitch direction control quantity; H is the total disturbance, and H x is the total disturbance in the rolling direction; H y is the total disturbance in the yaw direction; H z is the total disturbance in the pitch direction.

3. The method for predictive control of an aircraft attitude with adaptive disturbance observation compensation according to claim 1, characterized in that: The step 2 is specifically as follows: Step 21: Taking the pitch channel as an example, design an extended state observer In formula (5): e z For the observer The estimation error of For estimates; For estimates; For H z Estimate of β 01z is the adaptive gain coefficient of the first-order correction term of the pitch channel extended state observer; β 02z is the adaptive gain coefficient of the second-order correction term of the pitch channel extended state observer; β 03z is the adaptive gain coefficient of the third-order correction term of the pitch channel expanded state observer; β 01z ,β 02z With β 03z The expression is In formula (6): k sz >0 is the proportional coefficient, ò∈(1,1.5) is the power coefficient, δ 0iz >0, i=1,2,3 are threshold values; The expanded state observer used in the yaw and roll channels is the same as that in the pitch channel. The expanded state observer of the yaw channel is In formula (7): e y is the observer pair x 1ψ The estimation error of z 1ψ For x 1ψ Estimate of 2ψ For x 2ψ Estimate of 3ψ For H y Estimate of β 01y is the adaptive gain coefficient of the first-order correction term of the yaw channel extended state observer; β 02y is the adaptive gain coefficient of the second-order correction term of the yaw channel extended state observer; β 03y is the adaptive gain coefficient of the third-order correction term of the expanded state observer of the yaw channel; β 01y ,β 02y With β 03y The expression is In formula (8): k sy >0 is the proportionality coefficient, δ 0iy >0, i=1,2,3 are threshold values; The extended state observer of the rolling channel is In formula (9): e x is the observer pair x 1γ The estimation error of z 1γ For x 1γ Estimate of 2γ For x 2γ Estimates of 3γ For H x Estimate of β 01x is the adaptive gain coefficient of the first-order correction term of the roll channel expansion state observer; β 02x is the adaptive gain coefficient of the second-order correction term of the roll channel expansion state observer; β 03x is the adaptive gain coefficient of the third-order correction term of the roll channel expansion state observer; The outputs of the expanded state observer for the final pitch, yaw, and roll channels are defined as:

4. The method for predictive control of an aircraft attitude with adaptive disturbance observation compensation according to claim 1, characterized in that: The step three is specifically as follows: Step 31: Let the gain matrix of compensation control be in is the compensation control gain of the pitch channel, K bψ is the compensation control gain of the yaw channel, K bγ is the compensation control gain of the roll channel; K bψ With K bγ The expression is In formula (10): ε∈(0.5,1) is the power coefficient; Step 32: Multiply the gain of the compensation control by the disturbance observation value to obtain the compensation control amount U b =K b Z3.

5. The method of aircraft attitude prediction control with adaptive disturbance observation compensation according to claim 1, characterized in that: The step 4 is specifically as follows: Step 41: Define the integral performance index V In formula (11), t0=0 is the initial time of integration; τ is the integration variable; Q x With R u They are all positive definite matrices; x1(τ) represents the value of x1 at time τ; U(τ) represents the value of U at time τ; Step 42: Critic network design In formula (12): It is the estimated value of the integral performance indicator by the Critic network; is the Critic network weight; φ(x1) is the activation function; Step 43: Design the weight update law of the variable prediction cycle of the Critic network as In formula (13): is the derivative of the Critic network weight; η c >0 is the learning rate; Γ is the positive definite gain matrix; ω c is the estimated residual of the Critic network at the current sampling moment; ω ci is the estimated residual of the Critic network in the i-th prediction cycle; υ>0 is the regularization gain; The subscript i represents the value of the variable at the i-th prediction period; k is the adaptive prediction period; Step 44: Define the regression matrix and adaptively adjust the prediction period according to the rank of the regression matrix Define the regression matrix At each sampling moment, calculate whether the regression matrix is ​​full rank; if it is full rank, k does not need to be adjusted; if it is not full rank, at the next sampling moment, k = k + 1; Step 4 and 5: Design the Actor network and iteratively update it to obtain the approximate optimal control amount Actor networks are designed to in is the weight of the Actor network; the iterative update target of the Actor network is to minimize the normalized prediction error E c (t): Approximate optimal control quantity U a for

Images