Method for Controlling Steady Flight of Unmanned Aircraft

Abstract

Disclosed is a method for controlling stable flight of an unmanned aircraft, comprising the following steps: acquiring real-time flight operation data of the aircraft itself by means of an attitude sensor, a position sensor and an altitude sensor mounted to the unmanned aircraft, performing corresponding analysis on a kinematic problem of the aircraft by a processor mounted thereto, and establishing a dynamics model of the aircraft (S1); designing a controller of the unmanned aircraft according to a multi-layer zeroing neurodynamic method (S2); solving output control quantities of motors of the aircraft by the designed multi-layer zeroing neural network controller using the acquired real-time operation data of the aircraft and target attitude data (S3); and transferring solution results to a motor governor of the aircraft, and controlling powers of the motors according to a relationship between the control quantities solved by the controller and the powers of the motors of the multi-rotor unmanned aircraft, so as to control the motion of the unmanned aircraft (S4). Based on the multi-layer zeroing neurodynamic method, a correct solution to the problem can be approached rapidly, accurately and in real time, and a time-varying problem can be significantly solved.

Description

TECHNICAL FIELD

The present invention relates to a flight control method, in particular to a method for controlling stable flight of an unmanned aircraft.

BACKGROUND ART

In recent years, the world's unmanned aircraft technology has developed rapidly. Multi-rotor aircrafts with vertical takeoff and landing, stable hovering, wireless transmission, long-range aerial photography and autonomous cruising capabilities have broad application prospects in military and civil fields. Due to the characteristics of excellent maneuverability, simple mechanical structure, easy deployment and easy maintenance, the small rotor-type aircrafts are widely used in the fields of aerial photography, power inspection, environmental monitoring, forest fire prevention, disaster inspection, anti-terrorism rescue, military reconnaissance, battlefield assessment, etc. With wide application of unmanned aircrafts, the design of stable and fast-responding controllers of the unmanned aircrafts has attracted the attention of many researchers. The conventional controllers of the unmanned aircrafts are all designed on the basis of PID closed-loop control algorithms and corresponding improved control algorithms. PID controllers and feedback closed-loop control systems, which are simple in design and have good control effects, have been widely used in the design of the controllers of the aircrafts. Although the PID controllers are easy to use, the PID controllers and the power allocation schemes obtained according to the PID controllers do not achieve the desired stability of the unmanned aircrafts.

SUMMARY OF THE INVENTION

An object of the present invention is, in order to overcome the deficiencies of the prior art, to provide a method for designing a stable flight controller and a power allocation scheme, which controls stable flight of an unmanned aircraft by acquiring real-time flight operation data of the aircraft by using a sensor, solving output control quantities of motors of the aircraft by means of a multi-layer zeroing neural network, and obtaining a corresponding power allocation scheme.

The object of the present invention is achieved by means of the following technical solution:

a method for controlling stable flight of an unmanned aircraft, comprising the steps of:

1) acquiring real-time flight operation data of the aircraft itself, analyzing a kinematic problem of the aircraft, and establishing a dynamics model of the aircraft;

2) constructing a deviation function according to the real-time flight operation data acquired in step 1) and target attitude data, and constructing neurodynamic equations based on the deviation function by using a multi-layer zeroing neurodynamic method, wherein the neurodynamic equations based on the deviation function corresponding to all parameters together constitute a controller of the unmanned aircraft, and output quantities solved from differential equations of the controller are output control quantities of motors of the aircraft; and

3) controlling powers of the motors according to a relationship between the output control quantities solved in step 2) and the powers of the motors of the multi-rotor unmanned aircraft to complete motion control over the unmanned aircraft, specifically:

according to a power allocation scheme for the unmanned aircraft, the control quantities solved by the controller have the following relationship with the powers of the motors of the multi-rotor unmanned aircraft:

U=WF

where U=[u₁ u₂ u₃ u₄]^T refers to the output control quantities of the unmanned aircraft, F=[F₁ . . . F_j]^T refers to the powers of the motors of the unmanned aircraft, j is the number of the motors of the multi-rotor unmanned aircraft, W is a power allocation matrix of the unmanned aircraft, and the matrix W has different forms depending on different structures and the number of rotors, and needs to be determined according to the structure thereof and the number of the rotors;

the corresponding powers F of the motors are obtained by means of matrix inversion or pseudo-inversion, i.e.:

F=W ⁻¹ U

if the matrix W is a square matrix and is reversible, W⁻¹ is obtained by means of an inverse operation, and if W is not a square matrix, W⁻¹ is solved by means of a corresponding pseudo-inverse operation; and the desired powers F of the motors are finally obtained, input voltages of the motors are controlled according to a relationship between the voltages and powers of the motors to control the rotational speeds of the motors, and the control over the powers of the motors is finally realized to complete stable flight control over the unmanned aircraft.

Further, the step of performing corresponding analysis on a kinematic problem of the aircraft by a processor mounted thereto specifically comprises:

defining a ground coordinate system E and a fuselage coordinate system B, and establishing a relationship E=RB between the ground coordinate system and the fuselage coordinate system by means of a transformation matrix R, where R may be expressed as

$R=\hspace{1em}\hspace{1em}\left[\begin{array}{ccc}\cos \theta \cos \psi & \cos \theta \sin \psi & -s\mathrm{in}\theta \\ \sin \varphi \sin \theta \cos \psi -\cos \varphi \sin \psi & \begin{array}{c}\sin \varphi \sin \theta \sin \psi +\\ \cos \varphi \cos \psi \end{array}& \sin \varphi \cos \theta \\ \cos \varphi \sin \theta \cos \psi +\sin \varphi \sin \psi & \begin{array}{c}\cos \varphi \sin \theta \sin \psi -\\ \sin \varphi \cos \psi \end{array}& \cos \varphi \cos \theta \end{array}\right]$

where ϕ is a roll angle, θ is a pitch angle, and ψ is a yaw angle;

ignoring the effect of an air resistance on the aircraft, the stress analysis (in the form of Newton-Euler) of the aircraft system in the fuselage coordinate system is as follows

$\left[\begin{array}{cc}{\mathrm{mI}}_{3\times 3}& 0\\ 0& I\end{array}\right]\left[\begin{array}{c}\stackrel{.}{V}\\ \stackrel{.}{\omega }\end{array}\right]+\left[\begin{array}{c}\omega \times mV\\ \omega \times I\omega \end{array}\right]=\left[\begin{array}{c}F\\ \tau \end{array}\right]$

where m is the total mass of the aircraft, I_3×3 is a unit matrix, I is an inertia matrix, V is a linear velocity in the fuselage coordinate system, ω is an angular velocity in the fuselage coordinate system, F is a resultant external force, and τ is a resultant torque.

Further, the step of establishing a dynamics model of the aircraft specifically comprises:

according to the defined ground coordinate system E and fuselage coordinate system B, the relationship E=RB established between the two by means of the transformation matrix R and the stress analysis of the aircraft system in the fuselage coordinate system, obtaining dynamics equations of the multi-rotor aircraft as follows

$\{\begin{array}{c}\ddot{x}=\frac{u_x{u}_1}{m}\\ \ddot{y}=\frac{u_y{u}_1}{m}\\ \ddot{z}=-g+\frac{\left(\cos \varphi \cos \theta \right)u_1}{m}\\ \ddot{\varphi }=\frac{\left(I_y-I_z\right)\stackrel{.}{\theta }\stackrel{.}{\psi }+l{u}_2}{I_x}\\ \ddot{\theta }=\frac{\left(I_z-I_x\right)\ddot{\psi }\ddot{\varphi }+l{u}_3}{I_y}\\ \ddot{\psi }=\frac{\left(I_x-I_y\right)\varphi \ddot{\theta }+u_4}{I_z}\end{array}$

where l is an arm length; g is a gravitational acceleration; x, y, z are respectively position coordinates of the aircraft in the ground coordinate system; I_x, I_y, I_z are respectively rotational inertia of the aircraft in X, Y and Z axes; u_x=cos ϕ sin θ cos ψ+sin ϕ sin ψ; u_y=cos ϕ sin θ sin ψ−sin ϕ cos ψ; and u₁, u₂, u₃, u₄ are output control quantities.

Further, the step of designing a controller of the unmanned aircraft specifically comprises the steps of:

(1) designing a deviation function regarding the output control quantity u₁ from the vertical altitude z by means of the multi-layer zeroing neurodynamic method, and designing an altitude controller for the unmanned aircraft according to this deviation function;

(2) designing a deviation function regarding u_x and u_y from the horizontal positions x and y by means of the multi-layer zeroing neurodynamic method, designing a position controller for the unmanned aircraft according to this deviation function, and then inversely solving target attitude angles ϕ_T and θ_T; and

(3) designing a deviation function regarding the output control quantities u₂˜u₄ from the roll angle ϕ, the pitch angle θ and the yaw angle ψ by means of the multi-layer zeroing neurodynamic method, and designing an attitude controller according to this deviation function.

Further, the step of designing a deviation function regarding the output control quantity u₁ and a corresponding altitude controller of the unmanned aircraft specifically consists in that:

for the vertical altitude z, according to the target altitude value z_T and the actual altitude value z in the Z axis, a deviation function may be defined as

e _z1 =z−z _T  (1)

and its derivative may be obtained as follows

ė _z1 =ż−ż _T  (2)

in order to converge the actual value z to the target value z_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _z1 =γe _z1  (3)

where γ is a constant;

equations (1) and (2) are substituted into equation (3) and collating is performed to obtain

ż−ż _T+γ(z−z_T)=0  (4)

since equation (4) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _z2 =ż−ż _T+γ(z−z_T)  (5)

and its derivative may be obtained as follows

ė _z2 ={umlaut over (z)}−{umlaut over (z)} _T+γ(ż−ż_T)  (6)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _z2 =−γe _z2  (7)

equations (5) and (6) are substituted into equation (7) and collating is performed to obtain

{umlaut over (z)}−{umlaut over (z)} _T+2γ(ż−ż_T)+γ²(z−z_T)=0  (8)

in this way, a deviation function may be defined as

E _z ={umlaut over (z)}−{umlaut over (z)} _T+2γ(ż−ż_T)+γ²(z−z_T)  (9)

according to the dynamics equations of the aircraft, (9) may be simplified into

E _z =a _z u ₁ +b _z  (10)

where

$a_z=\frac{\cos \varphi \cos \theta }{m},$

and b_z=−g−{umlaut over (z)}_T+2γ(ż−ż_T)+γ² (z−z_T); and its derivative may be obtained as follows

Ė _z =a _z {dot over (u)} ₁ +{dot over (a)} _z u ₁ +{dot over (b)} _z  (11)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _z =−γE _z  (12)

equations (10) and (11) are substituted into equation (12) and collating is performed to obtain

a _z {dot over (u)} ₁=γ(a_zu₁+b_z)−{dot over (b)}_z−{dot over (a)}_zu₁  (13).

Further, the step of designing a deviation function regarding u_x and u_y and a corresponding position controller for the unmanned aircraft specifically consists in that:

for the horizontal position x, according to the target value x_T and the actual value x in the X axis, a deviation function may be defined as

e _x1 =x−x _T  (14)

and its derivative may be obtained as follows

ė _x1 ={dot over (x)}−{dot over (x)} _T  (15)

in order to converge the actual value x to the target value x_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _x1 =−γe _x1  (16)

equations (14) and (15) are substituted into equation (16) and collating is performed to obtain

{dot over (x)}−{dot over (x)} _T+γ(x−x_T)=0  (17)

since equation (17) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _x2 ={dot over (x)}−{dot over (x)} _T+γ(x−x_T)  (18)

and its derivative may be obtained as follows

ė _x2 ={umlaut over (x)}−{umlaut over (x)} _T+γ({dot over (x)}−{dot over (x)}_T)  (19)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _x2 =−γe _x2  (20)

equations (18) and (19) are substituted into equation (20) and collating is performed to obtain

{umlaut over (x)}−{umlaut over (x)} _T+2γ({dot over (x)}−{dot over (x)}_T)+γ²(x−x_T)=0  (21)

in this way, a deviation function may be defined as

E _x ={umlaut over (x)}−{umlaut over (x)} _T+2γ({dot over (x)}−{dot over (x)}_T)+γ²(x−x_T)  (22)

according to the dynamics equations of the aircraft, equation (22) may be simplified into

E _x =a _x u _x +b _x  (23)

where

$a_x=\frac{u_1}{m},$

and b_x=−{umlaut over (x)}_T+2γ({dot over (x)}−{dot over (x)}_T)+γ²(x−x_T); and its derivative may be obtained as follows

Ė _x =a _x {dot over (u)} _x +{dot over (a)} _x u _x +b _x  (24)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _x =−γE _x  (25)

equations (23) and (24) are substituted into equation (25) and collating is performed to obtain

a _x {dot over (u)} _x=−γ(a_xu_x+b_x)−{dot over (b)}_x−{dot over (a)}_xu_x  (26)

for the horizontal position y, according to the target value y_T and the actual value y in the Y axis, a deviation function may be defined as

e _y1 =y−y _T  (27)

and its derivative may be obtained as follows

ė _y1 =y−y _T  (28)

in order to converge the actual value y to the target value y_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _y1 =−γe _y1  (29)

equations (27) and (28) are substituted into equation (29) and collating is performed to obtain

{dot over (y)}−{dot over (y)} _T+γ(y−y_T)=0  (30)

since equation (30) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _y2 ={dot over (y)}−{dot over (y)} _T+γ(y−y_T)  (31)

and its derivative may be obtained as follows

ė _y2 =ÿ−ÿ _T+γ({dot over (y)}−{dot over (y)}_T)  (32)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _y2 =−γe _y2  (33)

equations (31) and (32) are substituted into equation (33) and collating is performed to obtain

ÿ−ÿ _T+2γ({dot over (y)}−{dot over (y)}_T)+γ²(y−y_T)=0  (34)

in this way, a deviation function may be defined as

E _y =ÿ−ÿ _T+2γ({dot over (y)}−{dot over (y)}_T)+γ²(y−y_T)  (35)

according to the dynamics equations of the aircraft, equation (35) may be simplified into

E _y =a _y u _y +b _y  (36)

where

$a_y=\frac{u_1}{m},$

and b_y=ÿ_T+2γ({dot over (y)}−{dot over (y)}_T)+γ²(y−y_T); and its derivative may be obtained as follows

Ė _y =a _y {dot over (u)} _y +{dot over (a)} _y u _y +{dot over (b)} _y  (37)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _y =−γE _y  (38)

equations (36) and (37) are substituted into equation (38) and collating is performed to obtain

a _y {dot over (u)} _y=γ(a_yu_y+b_y)−{dot over (b)}_y−{dot over (a)}_yu_y  (39).

Further, according to the designed position controller, the calculation method of inversely solving the target attitude angles ϕ_T and θ_T is:

u_x and u_y solved from equations (26) and (39) are

$\{\begin{array}{c}{u}_x=\cos \varphi \sin \theta \cos \psi +\sin \varphi \sin \psi \\ u_y=\cos \varphi \sin \theta \sin \psi -\sin \varphi \cos \psi \end{array}$

so that the inversely solved target angle values ϕ_T and θ_T are

$\begin{array}{cc}\{\begin{array}{c}{\varphi }_T={\sin }^{-1}\left(u_x\sin \psi -u_y\cos \psi \right)\\ {\theta }_T={\sin }^{-1}\left(\frac{u_x-\sin {\varphi }_T\sin \psi }{\cos {\varphi }_T\cos \psi }\right)\end{array}.& \left(40\right)\end{array}$

Further, the step of designing a deviation function regarding the output control quantity u₂˜u₄ and a corresponding attitude controller of the unmanned aircraft specifically consists in that:

for the roll angle ϕ, according to the target angle ϕ_T solved in (40) and the actual angle ϕ, a deviation function may be defined as

e _ϕ1=ϕ−ϕ_T  (41)

and its derivative may be obtained as follows

ė _ϕ1={dot over (ϕ)}−{dot over (ϕ)}_T  (42)

in order to converge the actual value ϕ to the target value ϕ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ϕ1 =−γe _ϕ1  (43)

equations (41) and (42) are substituted into equation (43) and collating is performed to obtain

{dot over (ϕ)}−{dot over (ϕ)}_T+γ(ϕ−ϕ_T)=0  (44)

since equation (44) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _ϕ2={dot over (ϕ)}−{dot over (ϕ)}_T+γ(ϕ−ϕ_T)  (45)

and its derivative may be obtained as follows

ė _ϕ2={umlaut over (ϕ)}−{umlaut over (ϕ)}_T+γ({dot over (ϕ)}−{dot over (ϕ)}_T)  (46)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ϕ2 =−γe _ϕ2  (47)

equations (45) and (46) are substituted into equation (47) and collating is performed to obtain

{umlaut over (ϕ)}−{umlaut over (ϕ)}_T+2γ({dot over (ϕ)}−{dot over (ϕ)}_T)+γ²(ϕ−ϕ_T)=0  (48)

in this way, a deviation function may be defined as

E _ϕ={umlaut over (ϕ)}−{umlaut over (ϕ)}_T+2γ({dot over (ϕ)}−{dot over (ϕ)}_T)+γ²(ϕ−ϕ_T)  (49)

according to the dynamics equations of the aircraft, equation (49) may be simplified into

E _ϕ =a _ϕ u ₂ +b _ϕ  (50)

where

$a_{\varphi }=\frac{l}{I_x},\mathrm{and}{b}_{\varphi }=\frac{\left(I_y-I_z\right)\stackrel{.}{\theta }\stackrel{.}{\psi }}{I_x}-{\ddot{\varphi }}_T+2\gamma \left(\stackrel{.}{\varphi }-{\stackrel{.}{\varphi }}_T\right)+{\gamma }^2\left(\varphi -{\varphi }_T\right);$

and its derivative may be obtained as follows

Ė _ϕ =a _ϕ {dot over (u)} ₂ +{dot over (a)} _ϕ u ₂ +{dot over (b)} _ϕ  (51)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _ϕ =γE _ϕ  (52)

equations (50) and (51) are substituted into equation (52) and collating is performed to obtain

a _ϕ {dot over (u)} ₂=−γ(a_ϕu₂+b_ϕ)−{dot over (b)}_ϕ−{dot over (a)}_ϕu₂  (53)

for the pitch angle θ, according to the target angle θ_T solved in (40) and the actual angle θ, a deviation function may be defined as

e _θ1=θ−θ_T  (54)

and its derivative may be obtained as follows

ė _θ1={dot over (θ)}−{dot over (θ)}_r  (55)

in order to converge the actual value θ to the target value θ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _θ1 =−γe _θ1  (56)

equations (54) and (55) are substituted into equation (56) and collating is performed to obtain

{dot over (θ)}−{dot over (θ)}_T+γ(θ−θ_T)=0  (57)

since equation (57) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _θ2={dot over (θ)}−{dot over (θ)}_T+γ(θ−θ_T)  (58)

and its derivative may be obtained as follows

ė _θ2={umlaut over (θ)}−{umlaut over (θ)}_T+γ({dot over (θ)}−{dot over (θ)}_T)  (59)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _θ2 =−γe _θ2  (60)

equations (58) and (59) are substituted into equation (60) and collating is performed to obtain

{umlaut over (θ)}−{umlaut over (θ)}_T+2γ({dot over (θ)}−{dot over (θ)}_T)+γ²(θ−θ_T)=0  (61)

in this way, a deviation function may be defined as

E _θ={umlaut over (θ)}−{umlaut over (θ)}_T+2γ({dot over (θ)}−{dot over (θ)}_T)+γ²(θ−θ_T)  (62)

according to the dynamics equations of the aircraft, equation (62) may be simplified into

E _θ =a _θ u ₃ +b _θ  (63)

where

$a_{\theta }=\frac{l}{I_y},\mathrm{and}{b}_{\theta }=\frac{\left(I_z-I_x\right)\stackrel{.}{\varphi }\stackrel{.}{\psi }}{I_y}-{\ddot{\theta }}_T+2\gamma \left(\stackrel{.}{\theta }-{\stackrel{.}{\theta }}_T\right)+{\gamma }^2\left(\theta -{\theta }_T\right);$

and its derivative may be obtained as follows

Ė _θ =a _θ {dot over (u)} ₃ +{dot over (b)} _θ  (64)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _θ =−γE _θ  (65)

equations (63) and (64) are substituted into equation (65) and collating is performed to obtain

a _θ {dot over (u)} ₃=γ(a_θu₃+b_θ)−{dot over (b)}_θ−{dot over (a)}_θu₃  (66)

for the yaw angle ψ, according to an artificially set angle ψ_T and the actual angle ψ, a deviation function may be defined as

e _ψ1=ψ−ψ_T  (67)

and its derivative may be obtained as follows

ė _ψ1={dot over (ψ)}−{dot over (ψ)}_T  (68)

in order to converge the actual value ψ to the target value ψ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ψ1 =−γe _ψ1  (69)

equations (67) and (68) are substituted into equation (69) and collating is performed to obtain

{dot over (ψ)}−{dot over (ψ)}_T+γ(ψ−ψ_T)=0  (70)

since equation (70) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _ψ2={dot over (ψ)}−{dot over (ψ)}_T+γ(ψ−ψ_T)  (71)

and its derivative may be obtained as follows

ė _ψ2={umlaut over (ψ)}−{umlaut over (ψ)}_T+γ({dot over (ψ)}−{dot over (ψ)}_T)  (72)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ψ2 =−γe _ψ2  (73)

equations (71) and (72) are substituted into equation (73) and collating is performed to obtain

{umlaut over (ψ)}−{umlaut over (ψ)}_T+2γ({dot over (ψ)}−{dot over (ψ)}_T)+γ²(ψ−ψ_T)=0  (74)

in this way, a deviation function may be defined as

E _ψ={umlaut over (ψ)}−{umlaut over (ψ)}_T+2γ({dot over (ψ)}−{dot over (ψ)}_T)+γ²(ψ−ψ_T)  (75)

according to the dynamics equations of the aircraft, equation (75) may be simplified into

E _ψ =a _ψ u ₄ +b _ψ  (76)

where

$a_{\psi }=\frac{1}{I_z},\mathrm{and}{b}_{\psi }=\frac{\left(I_x-I_y\right)\stackrel{.}{\varphi }\stackrel{.}{\theta }}{I_z}-{\ddot{\psi }}_T+2\gamma \left(\stackrel{.}{\psi }-{\stackrel{.}{\psi }}_T\right)+{\gamma }^2\left(\psi -{\psi }_T\right);$

and its derivative may be obtained as follows

Ė _ψ =a _ψ {dot over (u)} ₄ +{dot over (a)} _ψ u ₄ +{dot over (b)} _ψ  (77)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _ψ =−γE _ψ  (78)

equations (76) and (77) are substituted into equation (78) and collating is performed to obtain

a _ψ {dot over (u)} ₄=−γ(a_ψu₄+b_ψ)−{dot over (b)}_ψ−{dot over (a)}_ψu₄  (79).

Further, the step in which the altitude controller designed according to the altitude variable z, the position controller designed according to the position variables x and y, and the attitude controller designed according to the attitude control quantities ϕ, θ and ψ together constitute a stable flight controller of the multi-rotor unmanned aircraft specifically consists in that:

a controller of the unmanned aircraft may be obtained according to equations (13), (53), (66) and (79), wherein the controller can be implemented by a network structure; the controller of the unmanned aircraft is capable of controlling the stable flight of the unmanned aircraft; and the controller may be written in the following form:

$\hspace{1em}\{\begin{array}{c}{\stackrel{.}{u}}_1=\frac{-\gamma \left(a_z{u}_1+b_z\right)-{\stackrel{.}{b}}_z-{\stackrel{.}{a}}_z{u}_1}{a_z}\\ {\stackrel{.}{u}}_2=\frac{-\gamma \left(a_{\varphi }{u}_2+b_{\varphi }\right)-{\stackrel{.}{b}}_{\varphi }-{\stackrel{.}{a}}_{\varphi }{u}_2}{a_{\varphi }}\\ {\stackrel{.}{u}}_3=\frac{-\gamma \left(a_{\theta }{u}_3+b_{\theta }\right)-{\stackrel{.}{b}}_{\theta }-{\stackrel{.}{a}}_{\theta }{u}_3}{a_{\theta }}\\ {\stackrel{.}{u}}_4=\frac{-\gamma \left(a_{\psi }{u}_4+b_{\psi }\right)-{\stackrel{.}{b}}_{\psi }-{\stackrel{.}{a}}_{\psi }{u}_4}{a_{\psi }}\end{array}$

a zeroing neural network is constructed from the differential equations of the controller, and the control quantities of the unmanned aircraft are solved by means of the zeroing neural network.

Compared with the prior art, the present invention has the following beneficial effects:

1. The multi-layer zeroing neural network has better convergence characteristics, can realize real-time response of the aircraft and has a strong robustness, and the controller system designed according to the neural network is stable and has a good control effect.

2. The present invention is based on the multi-layer zeroing neurodynamic method, the method is described by using a ubiquitous implicit dynamics model, derivative information of various time-varying parameters can be fully utilized from the method and system level, and the method has a certain predictive ability for solving problems, can quickly, accurately and timely approach correct solutions of the problems, and can solve a variety of time-varying problems such as matrices, vectors, algebras and optimization.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a flow chart of a method for controlling stable flight of a multi-rotor aircraft according to an embodiment of the present invention;

FIG. 2 is a side view showing the structure of the multi-rotor aircraft according to the present invention;

FIG. 3 is a top view showing the structure of the multi-rotor aircraft according to the present invention;

FIG. 4 is a three-dimensional view showing the structure of the multi-rotor aircraft according to the present invention; and

FIG. 5 is a diagram showing a fuselage coordinate system of the multi-rotor aircraft.

DETAILED DESCRIPTION OF EMBODIMENTS

Hereafter the present invention will be further described in detail in conjunction with embodiments and appended drawings, but the embodiments of the present invention are not limited thereto.

Embodiment

As shown in FIG. 1, the present embodiment provides a method for controlling stable flight of an unmanned aircraft. The method comprises the following steps:

S1: acquiring real-time flight operation data of the aircraft itself by means of an attitude sensor, a position sensor and an altitude sensor mounted to the unmanned aircraft, performing corresponding analysis on a kinematic problem of the aircraft by a processor mounted thereto, and establishing a dynamics model of the aircraft;

One type of rotor flight structure in the multi-rotor aircraft is shown in FIGS. 2, 3 and 4. The structure is a six-rotor aircraft mechanism model consisting of multi-rotor aircraft propellers, brushless motors, rotor arms and a fuselage. Arrows in FIGS. 3 and 4 indicate the directions of rotation of the motors, and the combination of the illustrated clockwise and counterclockwise directions of rotation is to achieve mutual offsetting of torques the motors so as to achieve stable steering control.

Real-time attitude data θ(t), ϕ(t) and ψ(t) of the aircraft may be acquired by sensors such as gyros and accelerometers mounted to the multi-rotor aircraft by means of quaternion algebra, Kalman filtering and other algorithms, and position data x(t), y(t) and z(t) of the aircraft in the three-dimensional space is acquired by using altitude sensors and position sensors.

The definition of aircraft attitude variables is shown in FIG. 5.

The multi-rotor aircraft in FIG. 5 is defined as follows based on the fuselage coordinate system:

(1) six motors of the six-rotor aircraft are defined No. 1 to No. 6 in the clockwise direction;

(2) X axis extends in the direction of No. 1 rotor arm and points to the forward direction of the aircraft through the center of gravity of the fuselage;

(3) Y axis extends in the direction of the axis of symmetry of No. 2 and No. 3 rotor arms and points to the right motion direction of the aircraft through the center of gravity of the fuselage;

(4) Z axis extends upwardly perpendicular to the plane of six rotors and points to the climbing direction of the aircraft through the center of gravity of the fuselage;

(5) the pitch angle θ is an angle between the X axis of the fuselage and the horizontal plane, and is set to be positive when the fuselage is downward;

(6) the roll angle φ is an angle between the Z axis of the fuselage and the vertical plane passing through the X axis of the fuselage, and is set to be positive when the fuselage is rightward; and

(7) the yaw angle ψ is an angle between the projection of the X axis of the fuselage on the horizontal plane and the X axis of a geodetic coordinate system, and is set to be positive when the head of the aircraft is leftward.

According to different rotor-type aircraft models, physical model equations and dynamics equations for the aircraft are established, and dynamics analysis may be completed by means of the following aircraft dynamics modeling steps:

defining a ground coordinate system E and a fuselage coordinate system B, and establishing a relationship E=RB between the ground coordinate system and the fuselage coordinate system by means of a transformation matrix R, where R may be expressed as

$R=\left[\begin{array}{ccc}\cos \mathrm{\theta cos}\psi & \cos \mathrm{\theta sin}\psi & -\sin \theta \\ \sin \mathrm{\varphi sin}\theta \cos \psi -\cos \mathrm{\varphi sin}\psi & \sin \mathrm{\varphi sin}\theta \sin \psi +\cos \mathrm{\varphi cos}\psi & \sin \mathrm{\varphi cos}\theta \\ \cos \mathrm{\varphi sin}\theta \cos \psi +\sin \mathrm{\varphi sin}\psi & \cos \mathrm{\varphi sin}\theta \sin \psi -\sin \mathrm{\varphi cos}\psi & \cos \mathrm{\varphi cos}\theta \end{array}\right]$

where ϕ is a roll angle, θ is a pitch angle, and ψ is a yaw angle;

ignoring the effect of an air resistance on the aircraft, the stress analysis (in the form of Newton-Euler) of the aircraft system in the fuselage coordinate system is as follows

$\left[\begin{array}{cc}{\mathrm{mI}}_{3\times 3}& 0\\ 0& I\end{array}\right]\left[\begin{array}{c}\stackrel{.}{V}\\ \stackrel{.}{\omega }\end{array}\right]+\left[\begin{array}{c}\omega \times mV\\ \omega \times I\omega \end{array}\right]=\left[\begin{array}{c}F\\ \tau \end{array}\right]$

where m is the total mass of the aircraft, I_3×3 is a unit matrix, I is an inertia matrix, V is a linear velocity in the fuselage coordinate system, ω is an angular velocity in the fuselage coordinate system, F is a resultant external force, and τ is a resultant torque.

According to the above equation, the dynamics equations of the aircraft can be obtained as follows

$\hspace{1em}\{\begin{array}{c}\ddot{x}=\frac{u_x{u}_1}{m}\\ \ddot{y}=\frac{u_y{u}_1}{m}\\ \ddot{z}=-g+\frac{\left(\cos \varphi \cos \theta \right)u_1}{m}\\ \ddot{\varphi }=\frac{\left(I_y-I_z\right)\stackrel{.}{\theta }\stackrel{.}{\psi }+l{u}_2}{I_x}\\ \ddot{\theta }=\frac{\left(I_z-I_x\right)\ddot{\psi }\ddot{\varphi }+l{u}_3}{I_y}\\ \ddot{\psi }=\frac{\left(I_x-I_y\right)\ddot{\varphi }\ddot{\theta }+u_4}{I_z}\end{array}$

where l is an arm length; g is a gravitational acceleration; x, y, z are respectively position coordinates of the aircraft in the ground coordinate system; I_x, I_y, I_z are respectively rotational inertia of the aircraft in X, Y and Z axes; u_x=cos ϕ sin θ cos ψ+sin ϕ sin ψ; u_y=cos ϕ sin θ sin ψ−sin ϕ cos ψ; and u₁, u₂, u₃, u₄ are output control quantities.

S2: designing a controller of the unmanned aircraft according to a multi-layer zeroing neurodynamic method;

A deviation function regarding the output control quantity u₁ is designed from the vertical altitude z, an altitude controller of the multi-rotor unmanned aircraft is designed according to this deviation function, u₁ is solved; a deviation function regarding u_x and u_y and a corresponding position controller for the multi-rotor unmanned aircraft are designed from the horizontal positions x and y, and target attitude angles ϕ_T and θ_T are inversely solved; and a deviation function regarding the output control quantities u₂˜u₄ is designed from the roll angle ϕ, the pitch angle θ and the yaw angle ψ according to the target attitude angles, and a corresponding multi-layer zeroing neural network controller is designed. The specific steps are as follows:

for the vertical altitude z, according to the target altitude value z_T and the actual altitude value z in the Z axis, a deviation function may be defined as

e _z1 =z−z _T  (1)

and its derivative may be obtained as follows

ė _z1 =ż−ż _T  (2)

in order to converge the actual value z to the target value z_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _z1 =−γe _z1  (3)

where γ is a constant;

equations (1) and (2) are substituted into equation (3) and collating is performed to obtain

ż−ż _T+γ(z−z_T)=0  (4)

since equation (4) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _z2 =ż−ż _T+γ(z−z_T)  (5)

and its derivative may be obtained as follows

ė _z2 ={umlaut over (z)}−{umlaut over (z)} _T+γ(ż−ż_T)  (6)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _z2 =−γe _z2  (7)

equations (5) and (6) are substituted into equation (7) and collating is performed to obtain

{umlaut over (z)}−{umlaut over (z)} _T+2γ(ż−ż_T)+γ²(z−z_T)=0  (8)

in this way, a deviation function may be defined as

E _z ={umlaut over (z)}−{umlaut over (z)} _T+2γ(ż−ż_T)+γ²(z−z_T)  (9)

according to the dynamics equations of the aircraft, (9) may be simplified into

E _z =a _z u ₁ +b _z  (10)

where

$a_z=\frac{\cos \varphi \cos \theta }{m},$

and b_z=−g−{umlaut over (z)}_T+2γ(ż−ż_T)+γ² (z−z_T); and its derivative may be obtained as follows

Ė _z =a _z {dot over (u)} ₁ +{dot over (a)} _z u ₁ +{dot over (b)} _z  (11)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _z =−γE _z  (12)

equations (10) and (11) are substituted into equation (12) and collating is performed to obtain

a _z {dot over (u)} ₁=−γ(a_zu₁+b_z)−{dot over (b)}_z−{dot over (a)}_zu₁  (13).

for the horizontal position x, according to the target value x_T and the actual value x in the X axis, a deviation function may be defined as

e _x1 =x−x _T  (14)

and its derivative may be obtained as follows

ė _x1 ={dot over (x)}−{dot over (x)} _T  (15)

in order to converge the actual value x to the target value x_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _x1 =−γe _x1  (16)

equations (14) and (15) are substituted into equation (16) and collating is performed to obtain

{dot over (x)}−{dot over (x)} _T+γ(x−x_T)=0  (17)

since equation (17) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _x2 ={dot over (x)}−{dot over (x)} _T+γ(x−x_T)  (18)

and its derivative may be obtained as follows

ė _x2 ={umlaut over (x)}−{umlaut over (x)} _T+γ({dot over (x)}−{dot over (x)}_T)  (19)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _x2 =−γe _x2  (20)

equations (18) and (19) are substituted into equation (20) and collating is performed to obtain

{umlaut over (x)}−{umlaut over (x)} _T+2γ({dot over (x)}−{dot over (x)}_T)+γ²(x−x_T)=0  (21)

in this way, a deviation function may be defined as

E _x ={umlaut over (x)}−{umlaut over (x)} _T+2γ({dot over (x)}−{dot over (x)}_T)+γ²(x−x_T)  (22)

according to the dynamics equations of the aircraft, equation (22) may be simplified into

E _x =a _x u _x +b _x  (23)

where

$a_x=\frac{u_1}{m},$

and b_x=−{umlaut over (x)}_T+2γ({dot over (x)}−{dot over (x)}_T)+γ² (x−x_T); and its derivative may be obtained as follows

Ė _x =a _x {dot over (u)} _x +{dot over (a)} _x u _x +b _x  (24)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _x =−γE _x  (25)

equations (23) and (24) are substituted into equation (25) and collating is performed to obtain

a _x {dot over (u)} _x=−γ(a_xu_x+b_x)−{dot over (b)}_x−{dot over (a)}_xu_x  (26)

for the horizontal position y, according to the target value y_T and the actual value y in the Y axis, a deviation function may be defined as

e _y1 =y−y _T  (27)

and its derivative may be obtained as follows

ė _y1 =y−y _T  (28)

in order to converge the actual value y to the target value y_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _y1 =−γe _y1  (29)

equations (27) and (28) are substituted into equation (29) and collating is performed to obtain

{dot over (y)}−{dot over (y)} _T+γ(y−y_T)=0  (30)

since equation (30) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _y2 ={dot over (y)}−{dot over (y)} _T+γ(y−y_T)  (31)

and its derivative may be obtained as follows

ė _y2 =ÿ−ÿ _T+γ({dot over (y)}−{dot over (y)}_T)  (32)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _y2 =−γe _y2  (33)

equations (31) and (32) are substituted into equation (33) and collating is performed to obtain

ÿ−ÿ _T+2γ({dot over (y)}−{dot over (y)}_T)+γ²(y−y_T)=0  (34)

in this way, a deviation function may be defined as

E _y =ÿ−ÿ _T+2γ({dot over (y)}−{dot over (y)}_T)+γ²(y−y_T)  (35)

according to the dynamics equations of the aircraft, equation (35) may be simplified into

E _y =a _y u _y +b _y  (36)

where

$a_y=\frac{u_1}{m},$

b_y=−ÿ_T+2γ({dot over (y)}−{dot over (y)}_T)+γ² (y−y_T); and its derivative may be obtained as follows

Ė _y =a _y {dot over (u)} _y +{dot over (a)} _y u _y +{dot over (b)} _y  (37)

it is possible to use the multi-layer zeroing neurodynamic method to design

Ė _y =−γE _y  (38)

equations (36) and (37) are substituted into equation (38) and collating is performed to obtain

a _y {dot over (u)} _y=γ(a_yu_y+b_y)−{dot over (b)}_y−{dot over (a)}_yu_y  (39)

u_x and u_y may be solved from equations (26) and (39),

$\hspace{1em}\{\begin{array}{c}{u}_x=\cos \varphi \sin \theta \cos \psi +\sin \varphi \sin \psi \\ u_y=\cos \varphi \sin \theta \sin \psi -\sin \varphi \cos \psi \end{array}$

so that the inversely solved target angle values ϕ_T and θ_T may be

$\begin{array}{cc}\{\begin{array}{c}{\varphi }_T={\sin }^{-1}\left(u_x\sin \psi -u_y\cos \psi \right)\\ {\theta }_T={\sin }^{-1}\left(\frac{u_x-\sin {\varphi }_T\sin \psi }{\cos {\varphi }_T\cos \psi }\right)\end{array}& \left(40\right)\end{array}$

for the roll angle ϕ, according to the target angle ϕ_T solved in (40) and the actual angle ϕ, a deviation function may be defined as

e _ϕ1=ϕ−ϕ_T  (41)

and its derivative may be obtained as follows

ė _ϕ1={dot over (ϕ)}−{dot over (ϕ)}_T  (42)

in order to converge the actual value ϕ to the target value ϕ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ϕ1 =−γe _ϕ1  (43)

equations (41) and (42) are substituted into equation (43) and collating is performed to obtain

{dot over (ϕ)}−{dot over (ϕ)}_T+γ(ϕ−ϕ_T)=0  (44)

since equation (44) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _ϕ2={dot over (ϕ)}−{dot over (ϕ)}_T+γ(ϕ−ϕ_T)  (45)

and its derivative may be obtained as follows

ė _ϕ2={umlaut over (ϕ)}−{umlaut over (ϕ)}_T+γ({dot over (ϕ)}−{dot over (ϕ)}_T)  (46)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ϕ2 =−γe _ϕ2  (47)

equations (45) and (46) are substituted into equation (47) and collating is performed to obtain

{umlaut over (ϕ)}−{umlaut over (ϕ)}_T+2γ({dot over (ϕ)}−{dot over (ϕ)}_T)+γ²(ϕ−ϕ_T)=0  (48)

in this way, a deviation function may be defined as

E _ϕ={umlaut over (ϕ)}−{umlaut over (ϕ)}_T+2γ({dot over (ϕ)}−{dot over (ϕ)}_T)+γ²(ϕ−ϕ_T)  (49)

according to the dynamics equations of the aircraft, equation (49) may be simplified into

E _ϕ =a _ϕ u ₂ +b _ϕ  (50)

where

$a_{\varphi }=\frac{l}{I_x},\mathrm{and}{b}_{\varphi }=\frac{\left(I_y-I_z\right)\stackrel{.}{\theta }\stackrel{.}{\psi }}{I_x}-{\ddot{\varphi }}_T+2\gamma \left(\stackrel{.}{\varphi }-{\stackrel{.}{\varphi }}_T\right)+{\gamma }^2\left(\varphi -{\varphi }_T\right);$

and its derivative may be obtained as follows

Ė _ϕ =a _ϕ {dot over (u)} ₂ +{dot over (a)} _ϕ u ₂ +{dot over (b)} _ϕ  (51)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _ϕ =γE _ϕ  (52)

equations (50) and (51) are substituted into equation (52) and collating is performed to obtain

a _ϕ {dot over (u)} ₂=γ(a_ϕu₂+b_ϕ)−{dot over (b)}_ϕ−{dot over (a)}_ϕu₂  (53)

for the pitch angle θ, according to the target angle θ_T solved in (40) and the actual angle θ, a deviation function may be defined as

e _θ1=θ−θ_T  (54)

and its derivative may be obtained as follows

ė _θ1={dot over (θ)}−{dot over (θ)}_r  (55)

in order to converge the actual value θ to the target value θ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _θ1 =−γe _θ1  (56)

equations (54) and (55) are substituted into equation (56) and collating is performed to obtain

{dot over (θ)}−{dot over (θ)}_T+γ(θ−θ_T)=0  (57)

since equation (57) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _θ2={dot over (θ)}−{dot over (θ)}_T+γ(θ−θ_T)  (58)

and its derivative may be obtained as follows

ė _θ2={umlaut over (θ)}−{umlaut over (θ)}_T+γ({dot over (θ)}−{dot over (θ)}_T)  (59)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _θ2 =−γe _θ2  (60)

equations (58) and (59) are substituted into equation (60) and collating is performed to obtain

{umlaut over (θ)}−{umlaut over (θ)}_T+2γ({dot over (θ)}−{dot over (θ)}_T)+γ²(θ−θ_T)=0  (61)

in this way, a deviation function may be defined as

E _θ={umlaut over (θ)}−{umlaut over (θ)}_T+2γ({dot over (θ)}−{dot over (θ)}_T)+γ²(θ−θ_T)  (62)

according to the dynamics equations of the aircraft, equation (62) may be simplified into

E _θ =a _θ u ₃ +b _θ  (63)

where

$a_{\theta }=\frac{l}{I_y},\mathrm{and}{b}_{\theta }=\frac{\left(I_z-I_x\right)\stackrel{.}{\varphi }\stackrel{.}{\psi }}{I_y}-{\ddot{\theta }}_T+2\gamma \left(\stackrel{.}{\theta }-{\stackrel{.}{\theta }}_T\right)+{\gamma }^2\left(\theta -{\theta }_T\right);$

and its derivative may be obtained as follows

Ė _θ =a _θ {dot over (u)} ₃ +{dot over (b)} _θ  (64)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _θ =−γE _θ  (65)

equations (63) and (64) are substituted into equation (65) and collating is performed to obtain

a _θ {dot over (u)} ₃=−γ(a_θu₃+b_θ)−{dot over (b)}_θ−{dot over (a)}_θu₃  (66)

for the yaw angle ψ, according to the target angle ψ_T solved in (40) and the actual angle ψ, a deviation function may be defined as

e _ψ1=ψ−ψ_T  (67)

and its derivative may be obtained as follows

ė _ψ1={dot over (ψ)}−{dot over (ψ)}_T  (68)

in order to converge the actual value ψ to the target value ψ_T, according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ψ1 =−γe _ψ1  (69)

equations (67) and (68) are substituted into equation (69) and collating is performed to obtain

{dot over (ψ)}−{dot over (ψ)}_T+γ(ψ−ψ_T)=0  (70)

since equation (70) is generally not established in the initial situation and does not contain relevant information of the output control quantities, and the control quantities cannot be solved, a further design is needed, and a definition is then made

e _ψ2={dot over (ψ)}−{dot over (ψ)}_T+γ(ψ−ψ_T)  (71)

and its derivative may be obtained as follows

ė _ψ2={umlaut over (ψ)}−{umlaut over (ψ)}_T+γ({dot over (ψ)}−{dot over (ψ)}_T)  (72)

according to the multi-layer zeroing neurodynamic method, a neurodynamic equation based on the deviation function may be designed as

ė _ψ2 =−γe _ψ2  (73)

equations (71) and (72) are substituted into equation (73) and collating is performed to obtain

{umlaut over (ψ)}−{umlaut over (ψ)}_T+2γ({dot over (ψ)}−{dot over (ψ)}_T)+γ²(ψ−ψ_T)=0  (74)

in this way, a deviation function may be defined as

E _ψ={umlaut over (ψ)}−{umlaut over (ψ)}_T+2γ({dot over (ψ)}−{dot over (ψ)}_T)+γ²(ψ−ψ_T)  (75)

according to the dynamics equations of the aircraft, equation (75) may be simplified into

E _ψ =a _ψ u ₄ +b _ψ  (76)

where

$a_{\psi }=\frac{1}{I_z},\mathrm{and}{b}_{\psi }=\frac{\left(I_x-I_y\right)\stackrel{.}{\varphi }\stackrel{.}{\theta }}{I_z}-{\ddot{\psi }}_T+2\gamma \left(\stackrel{.}{\psi }-{\stackrel{.}{\psi }}_T\right)+{\gamma }^2\left(\psi -{\psi }_T\right);$

and its derivative may be obtained as follows

Ė _ψ =a _ψ {dot over (u)} ₄ +{dot over (a)} _ψ u ₄ +{dot over (b)} _ψ  (77)

according to the multi-layer zeroing neurodynamic method, it is possible to design

Ė _ψ =−γE _ψ  (78)

equations (76) and (77) are substituted into equation (78) and collating is performed to obtain

a _ψ {dot over (u)} ₄=−γ(a_ψu₄+b_ψ)−{dot over (b)}_ψ−{dot over (a)}_ψu₄  (79).

S3: solving output control quantities of motors of the aircraft by the designed multi-layer zeroing neural network controller using the acquired real-time operation data of the aircraft and target attitude data; and

A controller of the unmanned aircraft may be obtained according to multi-layer zeroing neural network equations (13), (53), (66) and (79), wherein the controller can be implemented by a network structure; the controller of the unmanned aircraft is capable of controlling the stable flight of the unmanned aircraft; and the controller may be written in the following form:

$\hspace{1em}\{\begin{array}{c}{\stackrel{.}{u}}_1=\frac{-\gamma \left(a_z{u}_1+b_z\right)-{\stackrel{.}{b}}_z-{\stackrel{.}{a}}_z{u}_1}{a_z}\\ {\stackrel{.}{u}}_2=\frac{-\gamma \left(a_{\varphi }{u}_2+b_{\varphi }\right)-{\stackrel{.}{b}}_{\varphi }-{\stackrel{.}{a}}_{\varphi }{u}_2}{a_{\varphi }}\\ {\stackrel{.}{u}}_3=\frac{-\gamma \left(a_{\theta }{u}_3+b_{\theta }\right)-{\stackrel{.}{b}}_{\theta }-{\stackrel{.}{a}}_{\theta }{u}_3}{a_{\theta }}\\ {\stackrel{.}{u}}_4=\frac{-\gamma \left(a_{\psi }{u}_4+b_{\psi }\right)-{\stackrel{.}{b}}_{\psi }-{\stackrel{.}{a}}_{\psi }{u}_4}{a_{\psi }}\end{array}$

a zeroing neural network is constructed from the differential equations of the controller, and the control quantities of the unmanned aircraft are solved by means of the zeroing neural network.

S4: transferring solution results of step S3 to a motor governor of the aircraft, and controlling powers of the motors according to a relationship between the control quantities solved by the controller and the powers of the motors of the multi-rotor unmanned aircraft, so as to control the motion of the unmanned aircraft;

according to a power allocation scheme for the unmanned aircraft, the control quantities solved by the controller have the following relationship with the powers of the motors of the multi-rotor unmanned aircraft:

U=WF

where U=[u₁ u₂ u₃ u₄]T refers to the control quantities of the unmanned aircraft, F=[F₁ . . . F_j]^T refers to the powers of the motors of the unmanned aircraft, j is the number of the motors of the multi-rotor unmanned aircraft, and W is a power allocation matrix of the unmanned aircraft.

In order to obtain the power required by the corresponding motor, the corresponding powers of the motors F may be obtained by means of matrix inversion or pseudo-inversion, i.e.

F=W ⁻¹ U

if the matrix W is a square matrix and is reversible, W⁻¹ is obtained by means of an inverse operation, and if W is not a square matrix, W⁻¹ is solved by means of a corresponding pseudo-inverse operation; and the desired powers F of the motors are finally obtained, input voltages of the motors are controlled according to a relationship between the voltages and powers of the motors to control the rotational speeds of the motors, and the control over the powers of the motors is finally realized to complete stable flight control over the unmanned aircraft. Since different numbers and structures of the rotors affect the control mode of the multi-rotor unmanned aircraft, the matrix W has different forms depending on the structure and the number of the rotors.

Taking the six-rotor unmanned aircraft as an example, the power allocation thereof has the following relationship:

$\{\begin{array}{c}{u}_1=F_1+F_2+F_3+F_4+F_5+F_6\\ u_2=F_2+F_3-F_5-F_6\\ u_3=F_1-F_4\\ u_4=-F_1-F_3-F_5+F_2+F_4+F_6\end{array}$

The relationship may be further written as

$\left[\begin{array}{c}{u}_1\\ u_2\\ u_3\\ u_4\end{array}\right]=\left[\begin{array}{cccccc}1& 1& 1& 1& 1& 1\\ 0& 1& 1& 0& -1& -1\\ 1& 0& 0& -1& 0& 0\\ -1& 1& -1& 1& -1& 1\end{array}\right]\left[\begin{array}{c}{F}_1\\ F_2\\ F_3\\ F_4\\ F_5\\ F_6\end{array}\right]$

Since W is not a square matrix in the above relationship, W⁻¹ may be obtained by means of pseudo-inversion, i.e.

$w^{-1}=\left[\begin{array}{cccc}0.1667& 0& 0.5& 0\\ 0.1667& 0.25& 0.25& 0.25\\ 0.1667& 0.25& -0.25& -0.25\\ 0.1667& 0& -0.5& 0\\ 0.1667& -0.25& -0.25& -0.25\\ 0.1667& -0.25& 0.25& 0.25\end{array}\right]$

In this way, the power allocation of the six-rotor unmanned aircraft and the corresponding actual motor control quantity may be obtained to control the operation of the motor.

The foregoing description is merely illustrative of preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Equivalent replacements or modifications made to the inventive concept or technical solution of the present invention by a person skilled in the art within the scope of the disclosure of the present invention fall into the scope of protection of the present invention.

Claims

1. A method for controlling stable flight of an unmanned aircraft, characterized by comprising the steps of: 1) acquiring real-time flight operation data of the aircraft itself, analyzing a kinematic problem of the aircraft, and establishing a dynamics model of the aircraft;2) constructing a deviation function according to the real-time flight operation data acquired in step 1) and target attitude data, and constructing neurodynamic equations based on the deviation function by using a multi-layer zeroing neurodynamic method, wherein the neurodynamic equations based on the deviation function corresponding to all parameters together constitute a controller of the unmanned aircraft, and output quantities solved from differential equations of the controller are output control quantities of motors of the aircraft; and3) controlling powers of the motors according to a relationship between the output control quantities solved in step 2) and the powers of the motors of the multi-rotor unmanned aircraft to complete motion control over the unmanned aircraft, specifically:according to a power allocation scheme for the unmanned aircraft, the control quantities solved by the controller have the following relationship with the powers of the motors of the multi-rotor unmanned aircraft: U=WFwhere U=[u1 u2 u3 u4]T refers to the output control quantities of the unmanned aircraft, F=[F1 . . . Fj]T refers to the powers of the motors of the unmanned aircraft, j is the number of the motors of the multi-rotor unmanned aircraft, W is a power allocation matrix of the unmanned aircraft, and the matrix W has different forms depending on different structures and the number of rotors, and needs to be determined according to the structure thereof and the number of the rotors;the corresponding powers F of the motors are obtained by means of matrix inversion or pseudo-inversion, i.e.: F=W−1U if the matrix W is a square matrix and is reversible, W−1 is obtained by means of an inverse operation, and if W is not a square matrix, W−1 is solved by means of a corresponding pseudo-inverse operation; and the desired powers F of the motors are finally obtained, input voltages of the motors are controlled according to a relationship between the voltages and powers of the motors to control the rotational speeds of the motors, and the control over the powers of the motors is finally realized to complete stable flight control over the unmanned aircraft.

2. The method for controlling stable flight of an unmanned aircraft according to claim 1, characterized in that the step of analyzing a kinematic problem of the aircraft by a processor mounted thereto specifically comprises: defining a ground coordinate system E and a fuselage coordinate system B, and establishing a relationship E=RB between the ground coordinate system and the fuselage coordinate system by means of a transformation matrix R, where R is expressed as

3. The method for controlling stable flight of an unmanned aircraft according to claim 2, characterized in that the step of establishing a dynamics model of the aircraft specifically comprises: according to the defined ground coordinate system E and fuselage coordinate system B, the relationship E=RB established between the two by means of the transformation matrix R and the stress analysis of the aircraft system in the fuselage coordinate system, obtaining dynamics equations of the multi-rotor aircraft as follows

4. The method for controlling stable flight of an unmanned aircraft according to claim 1, characterized in that the step of designing a controller of the unmanned aircraft according to a multi-layer zeroing neurodynamic method specifically comprises the steps of: (2-1) designing a deviation function regarding the output control quantity u1 from the vertical altitude z by means of the multi-layer zeroing neurodynamic method, and designing an altitude controller for the unmanned aircraft according to this deviation function;(2-2) designing a deviation function regarding ux and uy from the horizontal positions x and y by means of the multi-layer zeroing neurodynamic method, designing a position controller for the unmanned aircraft according to this deviation function, and then inversely solving target attitude angles ϕT and θT; and(2-3) designing a deviation function regarding the output control quantities u2˜u4 from the roll angle ϕ, the pitch angle θ and the yaw angle ψ by means of the multi-layer zeroing neurodynamic method, and designing an attitude controller according to this deviation function.

5. The method for controlling stable flight of an unmanned aircraft according to claim 4, characterized in that the step of designing a deviation function regarding the output control quantity u1 from the vertical altitude z by means of the multi-layer zeroing neurodynamic method, and designing an altitude controller for the unmanned aircraft according to this deviation function specifically comprises: for the vertical altitude z, according to the target altitude value zT and the actual altitude value z in the Z axis, defining a deviation function as Ez={umlaut over (z)}−{umlaut over (z)}T+2γ(ż−żT)+γ2(z−zT)  (9)according to the dynamics equations of the aircraft, simplifying (9) into Ez=azu1+bz  (10)where

6. The method for controlling stable flight of an unmanned aircraft according to claim 4, characterized in that the step of designing a deviation function regarding ux and uy and a position controller for the unmanned aircraft specifically comprises: for the horizontal position x, according to the target value xT and the actual value x in the X axis, defining a deviation function as Ex={umlaut over (x)}−{umlaut over (x)}T+2γ({dot over (x)}−{dot over (x)}T)+γ2(x−xT)  (22)according to the dynamics equations of the aircraft, simplifying equation (22) into Ex=axux+bx  (23)where

7. The method for controlling stable flight of an unmanned aircraft according to claim 6, characterized in that the calculation formulas of inversely solving the target attitude angles ϕT and θT are: ux and uy solved from position controller equations (26) and (39) are

8. The method for controlling stable flight of an unmanned aircraft according to claim 4, characterized in that the step of designing a deviation function regarding the output control quantities u2˜u4 from the roll angle ϕ, the pitch angle θ and the yaw angle ψ by means of the multi-layer zeroing neurodynamic method, and designing an attitude controller according to this deviation function specifically comprises: for the roll angle ϕ, according to the target angle ϕT solved in (40) and the actual angle ϕ, defining a deviation function as Eϕ={umlaut over (ϕ)}−{umlaut over (ϕ)}T+2γ({dot over (ϕ)}−{dot over (ϕ)}T)+γ2(ϕ−ϕT)  (49)according to the dynamics equations of the aircraft, simplifying equation (49) into Eϕ=aϕu2+bϕ  (50)where

9. The method for controlling stable flight of an unmanned aircraft according to claim 4, characterized in that the step in which the designed altitude controller, position controller and attitude controller together constitute a stable aircraft of the multi-rotor unmanned aircraft specifically comprises: az{dot over (u)}1=−γ(azu1+bz)−{dot over (b)}z−{dot over (a)}zu1  (13)aϕ{dot over (u)}2=−γ(aϕu2+bϕ)−{dot over (b)}ϕ−{dot over (a)}ϕu2  (53)aθ{dot over (u)}3=−γ(aθu3+bθ)−{dot over (b)}θ−{dot over (a)}θu3  (66)aψ{dot over (u)}4=−γ(aψu4+bγ)−{dot over (b)}ψ{dot over (a)}ψu4  (79)obtaining a controller of the unmanned aircraft according to equations (13), (53), (66) and (79), wherein the controller can be implemented by a network structure; the controller of the unmanned aircraft is capable of controlling the stable flight of the unmanned aircraft; and the controller is written in the following form: