Disclosure of Invention
The invention aims to provide a construction method of a decoupling controller (hereinafter referred to as a motor decoupling controller) of an ironless outer rotor bearingless permanent magnet motor, which is based on the generalized inverse of a least square support vector machine, can realize nonlinear decoupling control between two-degree-of-freedom radial suspension force and between the radial suspension force and the rotating speed of the ironless outer rotor bearingless permanent magnet synchronous motor under the load condition, and can effectively improve various control performance indexes of the ironless outer rotor bearingless permanent magnet synchronous motor, such as steady-state tracking precision, dynamic response speed, parameter robustness and disturbance suppression.
The invention provides a construction method of a coreless outer rotor bearingless permanent magnet motor decoupling controller, which adopts the technical scheme that: the structure of a composite controlled object comprising a coreless outer rotor bearingless permanent magnet synchronous motor is characterized in that the composite controlled object uses four reference currents
As input, a displacement x in x, y direction
M,y
MAnd a rotational speed omega
eAs an output, the method further comprises the following steps:
step 1) constructing a least square support vector machine generalized inverse by adopting a least square support vector machine, 2 integrators and 3 transfer functions, wherein the least square support vector machine generalized inverse is based on the displacement x
M、y
MSecond derivative of (2)
And a rotational speed omega
eFirst derivative of
As input, four reference currents are output
Step 2) the least square support vector machine generalized inverse is connected with the composite controlled object in series to form a pseudo linear subsystem consisting of a first displacement integral type pseudo linear subsystem, a second displacement integral type pseudo linear subsystem and a rotating speed integral type pseudo linear subsystem, wherein the input of the first displacement integral type pseudo linear subsystem is
The output being the displacement x
MThe input of the second shift integral pseudo linear subsystem is
The output being the displacement y
M(ii) a The input of the rotating speed integral type pseudo linear subsystem is
The output being the displacement omega
e;
Step 3) a first displacement regulator is connected in series in front of the first displacement integral type pseudo linear subsystem, a second displacement regulator is connected in series in front of the second displacement integral type pseudo linear subsystem, and a rotating speed regulator is connected in series in front of the rotating speed integral type pseudo linear subsystem; the inputs of the first, second and third pre-filters are respectively the reference displacement in x and y directions in one-to-one correspondence
And a reference rotational speed
The output is correspondingly the filtered displacement instruction value
And a rotational speed command value
The displacement instruction value
And a displacement x
MDifferencing to obtain a displacement error e in the x direction
xAs input to the first displacement regulator, the first displacement regulator is derived
The displacement instruction value
And a displacement y
MDifferencing to obtain a y-direction displacement error signal e
yAs an input to the second displacement regulator, the second displacement regulator is fed
The rotating speed instruction value
With the speed of rotation omega
eDifferencing to obtain a rotational speed error signal e
ωAs an input to the speed regulator, the speed regulator is obtained
The first, second and third pre-filters (51, 52, 53) are all
Filter of F
r(s) is the transfer function of the pre-filter, ω
rIs the bandwidth parameter, s is the laplacian operator.
The rotational speed regulator is
PI regulator of (1)
PID(s) is the transfer function of a displacement regulator having a PID kernel; k
PIDAs a PID kernel parameter, ω
cIs the desired bandwidth.
The invention has the beneficial effects that:
1) the use of the pre-filter in the anti-interference closed-loop regulator enables the signal input by the motor decoupling controller to be more gentle, avoids the influence of step signals and impact signals, enables the output to better follow the instruction signal, and enables the bandwidth of the controller to be remarkably improved due to better following effect, thereby improving the control performance of the motor decoupling controller.
2) The anti-interference closed-loop regulator is formed by the prefilter, the displacement regulator and the rotating speed regulator respectively, the anti-interference closed-loop regulator has a parameter adjusting method similar to the traditional PID, and meanwhile, due to the introduction of the bandwidth parameter and the prefilter, after the bandwidth of the controller is improved, the PID nuclear parameter of the anti-interference closed-loop regulator has a wider adjusting range, so the anti-interference closed-loop regulator is better than the PID controller in parameter applicability, and meanwhile, the regulator can also well inhibit external disturbance and internal disturbance, so the anti-interference closed-loop regulator can better resist the disturbance, and the operation control performance of the coreless outer rotor bearingless permanent magnet synchronous motor is better.
3) The motor is subjected to decoupling control in a mode of constructing the generalized inverse system, and the pseudo linear system has a stable open-loop relation by configuring parameters of the generalized inverse system, so that the performance of the system is better than that of the traditional alpha-order inverse system. Meanwhile, a least square support vector machine with the advantages of high learning speed, short training time, easy optimization of a network structure and the like is adopted to establish a model of the motor, and the decoupling control problem of multivariable nonlinear strong coupling of the motor is solved.
4) The control of a multivariable, nonlinear and strong coupling time-varying system, namely the coreless outer rotor bearingless permanent magnet synchronous motor, is converted into the control of two displacement second-order standard linear subsystems and a rotating speed first-order standard linear subsystem by constructing the generalized inverse of a least square support vector machine. The anti-interference closed-loop controller is designed by utilizing a regulator method, so that dynamic decoupling between two-degree-of-freedom radial suspension forces and between the radial suspension force and the rotating speed is realized. Therefore, the control of a two-degree-of-freedom displacement system and the rotating speed of the motor can be independently realized, and the high-performance operation control of the coreless outer rotor bearingless permanent magnet synchronous motor is obtained.
5) The motor decoupling controller constructed by the invention has simple control structure and excellent control performance, is also suitable for controlling other types of bearingless motors and various motors supported by a magnetic bearing.
Detailed Description
Referring to fig. 1, a first PI controller 11, a second PI controller 12, a third PI controller 21, a fourth PI controller 22, a first Park inverse transformation module 13, a second Park inverse transformation module 23, a first Park transformation module 14, a second Park transformation module 24, a first Clarke transformation module 16, a second Clarke transformation module 26, a first SVPWM inverter 15, a second SVPWM inverter 26, and a coreless outer rotor bearingless permanent magnet synchronous motor 27 are combined as a whole to form a composite controlled object 28, the composite controlled object 28 is equivalent to a differential equation model of 5 th order, and the relative order of a system vector is {2,2,1 }. As shown in fig. 2, a least squares support vector machine generalized inverse 3 of a composite controlled object having 8 input nodes and 4 output nodes is constructed by using a least squares support vector machine 31 having 8 input nodes and 4 output nodes, plus 2 integrators, and 3 transfer functions based on input and output. As shown in fig. 3, the least square support vector machine generalized inverse 3 is connected in series before the composite controlled object 28, and the least square support vector machine generalized inverse 3 and the composite controlled object 28 are synthesized into the pseudo-linear system 4 composed of two displacement second-order transfer function standard pseudo-linear subsystems and a speed first-order transfer function standard pseudo-linear subsystem, so that a complex multivariable, nonlinear and strongly coupled control system is converted into the control of the two second-order transfer function standard pseudo-linear subsystems and the first-order transfer function standard pseudo-linear subsystem. Referring to fig. 4, for two second-order subsystems and one first-order subsystem that have been linearly decoupled, an anti-interference PID regulator design method is adopted to design a first pre-filter 51, a second pre-filter 52, a third pre-filter 53, a first displacement regulator 54, a second displacement regulator 55 and a rotation speed regulator 56, respectively, and the anti-interference closed-loop controller 5 is composed of the first pre-filter 51, the second pre-filter 52, the third pre-filter 53, the first displacement regulator 54, the second displacement regulator 55 and the rotation speed regulator 56. Finally, the decoupling controller of the coreless outer rotor bearingless permanent magnet motor is formed by an anti-interference closed-loop controller 5, a least square support vector machine generalized inverse 3, a first PI controller 11, a second PI controller 12, a third PI controller 21, a fourth PI controller 22, a first Park inverse transformation module 13, a second Park inverse transformation module 23, a first Park transformation module 14, a second Park transformation module 24, a first Clarke transformation module 16, a second Clarke transformation module 26, a first SVPWM inverter 15 and a second SVPWM inverter 26, and nonlinear dynamic decoupling control is carried out on the coreless outer rotor bearingless permanent magnet synchronous motor 27. The specific implementation comprises the following 7 steps:
1) as shown in fig. 1, a composite controlled
object 28 including an ironless outer rotor bearingless permanent magnet
synchronous motor 27 is constructed. In the feedback channel, the output end of the coreless outer rotor bearingless permanent magnet
synchronous motor 27 is respectively connected with a first Clarke conversion module 16 and a second Clarke conversion moduleAnd the output end of the second Clarke conversion module 26 is connected with the second Park conversion module 24 in series, and the output end of the first Clarke conversion module 16 is connected with the first Park conversion module 14 in series. The first Park conversion module 14 and the second Park conversion module 24 are fed back to the forward channel. In the forward channel, a first PI controller 11 and a second PI controller 12 are connected in series before a first Park inverse transformation 13, a third PI controller 21 and a fourth PI controller 22 are connected in series before a second Park inverse transformation 23, the first Park inverse transformation 13 is connected in series before an ironless outer rotor bearingless permanent magnet
synchronous motor 27 through a
first SVPWM inverter 15, and the second Park inverse transformation 23 is connected in series before the ironless outer rotor bearingless permanent magnet
synchronous motor 27 through a
second SVPWM inverter 25, so as to jointly form a composite controlled
object 28. The composite controlled
object 28 has four reference currents
With signal as input, with displacement x in x, y direction
M、y
MAnd the rotational speed omega of the motor rotor
eAs an output.
The torque current i is obtained by collecting the torque current and the suspension current of the coreless outer rotor bearingless permanent magnet
synchronous motor 27
Ma、i
MbAnd a levitation current i
Ba、i
BbThe torque current i can be converted by means of a first Clarke transformation 16
Ma、i
MbIs converted into i
Mα、i
MβTo i with
Mα、i
MβAs input to the first Park transformation 14, an output i is obtained
Md、i
Mq. Will obtain i
Md、i
MqAnd torque current command value
The difference is made, and the obtained current difference is input to the corresponding first PI controller 11 and second PI controller 12. The current error can be converted into control voltage through the first PI controller 11 and the second PI controller 12 respectively
And
signal to be converted control voltage
And
as an input signal of the first Park inverse transformation 13, an output voltage can be obtained
And
then the obtained output voltage is used
And
the torque voltage u of the coreless outer rotor bearingless permanent magnet
synchronous motor 27 can be obtained as an input of the
first SVPWM inverter 15
Ma、u
MbAnd u
McInputting a signal. The torque current i can be converted by a second Clarke transformation 26
Ba、i
BbIs converted into i
Bα、i
BβTo i with
Bα、i
BβAs input to the second Park transformation 24, an output i is available
Bd、i
Bq. Will obtain i
Bd、i
BqAnd torque current command value
The difference is made, and the obtained current difference is input to the corresponding third PI controller 21 and fourth PI controller 22. The current error can be converted into control voltage by the third PI controller 21 and the fourth PI controller 22
And
converting the control voltage
And
as an input signal of the second Park inverse transformation 23, an output voltage can be obtained
And
then the obtained output voltage is used
And
as an input of the
second SVPWM inverter 25, the levitation voltage u of the coreless outer rotor bearingless permanent magnet
synchronous motor 27 can be obtained
Ba、u
BbAnd u
BcInputting a signal. The obtained torque voltage u
Ma、u
MbAnd u
McAnd a levitation voltage u
Ba、u
BbAnd u
BcThe displacement x can be obtained as an input signal of the coreless outer rotor bearingless permanent magnet
synchronous motor 27
M、y
MAnd a rotational speed omega
e。
2) Through analysis, equivalence and derivation, a basis on a method is provided for the construction and the learning training of the generalized inverse of the least square support vector machine. Firstly, a mathematical model of a composite controlled object 28 is established, based on the working principle of the coreless outer rotor bearingless permanent magnet synchronous motor 27, the mathematical model of the coreless outer rotor bearingless permanent magnet synchronous motor 27 is established, closed-loop control of torque and suspension is formed through Clarke transformation and Park transformation in a feedback channel and PI controller, Park inverse transformation and linear amplification in a forward channel, and the mathematical model of the composite controlled object 28, namely a 5-order differential equation, is obtained, wherein the relative order of vectors of the equation is {2,2,1 }. Derivation can prove that the 5 th order differential equation is reversible, namely a generalized inverse system exists, three inputs of the inverse system can be determined to be two displacement second-order derivatives and one first-order derivative of the rotating speed, and four outputs are respectively four inputs of the composite controlled object 28. A least squares support vector machine generalized inverse 3 can thus be constructed, as shown in fig. 2. Provides a method basis for learning and training.
3) As shown in fig. 2, a least squares support vector machine generalized inverse 3 is constructed by using a least squares
support vector machine 31, 2 integrators and 3 input-output-based transfer functions, and the least squares support vector machine generalized inverse 3 is displaced by x in x and y directions
M、y
MSecond derivative of (2)
And the rotational speed omega of the motor rotor
eFirst derivative of
As input, four reference currents are output
Wherein the number of input nodes of the least squares
support vector machine 31 is 8, the number of output layer nodes is 4, and the vector coefficient and the threshold of the least squares
support vector machine 31 are determined in the next off-line learning. Using a least squares
support vector machine 31 with 8 input nodes, 4 output nodes plus two integrators s
-1A second order normalized transfer function s/(a)
12s
2+a
11s+a
10) A second order normalized transfer function s/(a)
22s
2+a
21s+a
20) And a first order canonical transfer function s/(a)
31s+a
30) Forming a generalized inverse 3 of a least squares support vector machine, wherein s is a Laplace operator; a is
10、a
11、a
12、a
20、a
21、a
22、a
30、a
31Respectively, constant coefficients of a second-order standard type transfer function, the values of which can be adjusted according to actual operation, wherein the optimal damping coefficient value in engineering application, namely a, is selected
10、a
12、a
20、a
22、a
30、a
31Are all taken to be 1, a
11、a
21All taken to be 1.414.
Wherein: the first input of the generalized inverse 3 of the least squares support vector machine is displacement
Displacement of
As a first input to the least squares
support vector machine 31, a displacement
Through the first input-output based second-order standard type transfer function s/(a)
12s
2+a
11s+a
10) Then, the displacement x is output
MFirst derivative of
As a second input to the least squares
support vector machine 31,
then passes through a first integrator s
-1Rear output displacement x
MBy a displacement x
MAs a third input to the least squares
support vector machine 31. The second input of the generalized inverse 3 of the least squares support vector machine is displacement
Displacement of
As a fourth input to the least squares
support vector machine 31, a displacement
Passing through a second order normalized transfer function s/(a)
22s
2+a
21s+a
20) Rear output y
MFirst derivative of
As a fifth input to the least squares
support vector machine 31,
then passes through a second integrator s
-1Rear output displacement y
M,y
MAs the sixth input to the least squares
support vector machine 31. The third input of the generalized inverse 3 of the least squares support vector machine is the rotation speed
Rotational speed
As a seventh input to the least squares support
vector machine 31,
through a third input-output-based first-order standard type transfer function s/(a)
31s+a
30) After that, the output rotation speed omega
e,ω
eAs the eighth input to the least squares support
vector machine 31. The output of the least squares support
vector machine 31 is four reference currents output by the generalized inverse 3 of the least squares support vector machine
4) The vector coefficients and thresholds of the least squares support
vector machine 31 are adjusted and determined as follows: (1) four reference currents
Is applied to the input end of the composite controlled
object 28 as a step excitation signal to acquire the displacement x of the coreless outer rotor bearingless permanent
magnet synchronous motor 27
M、y
MAnd a rotational speed omega
e. (2) Will bit x
M、y
MRespectively solving the first derivative and the second derivative of the rotation speed omega off line
eThe first derivative is calculated, and the signals are normalized to form a training sample set of the least square support vector machine 31
(3) And selecting a Gaussian kernel function as the kernel function of the least square
support vector machine 31, setting the regularization parameter of the least square
support vector machine 31 to be 996, and setting the kernel width to be 3.5, so as to adjust the vector coefficient and the threshold of the least square
support vector machine 31 off line.
5) As shown in fig. 3, two displacement integral type pseudo-linear subsystems and one rotation speed integral type pseudo-linear subsystem are formed. The least square support vector machine generalized inverse 3 and the composite controlled
object 28 are connected in series to form a pseudo linear system 4, and the pseudo linear system 4 is composed of a first displacement integral type pseudo linear subsystem 41, a second displacement integral type pseudo linear subsystem 42 and a rotating speed integral type pseudo linear subsystem 43. Wherein the input of the shift integral pseudo linear subsystem 41 is
The output being the displacement x
MThe input to the shift integral pseudo linear subsystem 42 is
The output being the displacement y
M(ii) a The input to the rotational speed integral type pseudo linearity subsystem 43 is
The output being the displacement omega
e. Therefore, nonlinear dynamic decoupling between two-degree-of-freedom radial suspension force and between the radial suspension force and the rotating speed is realized, and the control of a complex nonlinear system is converted into the control of three simple univariate linear systems.
6) As shown in fig. 5, the anti-disturbance closed-loop controller 5 is designed. Regulators are respectively designed for two displacement integral type pseudo linear subsystems 41 and 42 and one rotation speed integral type pseudo linear subsystem 43. Corresponding first prefilter 51 and first displacement adjuster 54 are designed for the displacement integral pseudo linear control subsystem 41, and the first displacement adjuster 54 is connected in series in front of the displacement integral pseudo linear control subsystem 41. Designing corresponding second prefilter 52 and second phase shift modulation for phase shift integration pseudowire control subsystem 42The node 55 and the second displacement regulator 55 are connected in series before the displacement integral pseudo linear subsystem 42. A third prefilter 53 and a rotational speed regulator 56 are designed for one rotational speed integral type pseudo linear subsystem 43, and the rotational speed regulator 56 is connected in series before the rotational speed integral type pseudo linear subsystem 43. Wherein the input of the first pre-filter 51 is the reference displacement in the x-direction
The output is the displacement instruction value in the x direction after filtering
Filtering the x-direction displacement instruction value
Displacement x from the detected x direction
MTaking the difference to obtain the displacement error e in the x direction
xAnd apply the displacement error e
xAs input to the first displacement regulator 54, the input is obtained after being regulated by the first displacement regulator 54
A first displacement regulator 54 is connected in series before the displacement integrating pseudo linear subsystem 41,
the displacement x in the x direction is obtained as an input to the displacement integral pseudo linear subsystem 41
M. Similarly, the input to the second prefilter 52 is a reference displacement in the y-direction
The output is the displacement instruction value of the y direction after filtering
The filtered displacement instruction value in the y direction
With the detected displacement signal y in the y-direction
MMaking a difference between the two steps,obtaining a displacement error signal e in the y direction
yAnd the signal is used as the input of the second displacement regulator 55, and after the regulation of the second displacement regulator 55, the output signal is obtained
Will be provided with
As an input of the shift integral pseudo linear subsystem 42, a shift signal y in the y direction is obtained
M. Similarly, the input of the third pre-filter 53 is the reference rotation speed
The output is the filtered rotating speed instruction value
Filtering the speed instruction value
With detected speed signal omega
eMaking difference to obtain rotation speed error signal e
ωAnd sends the signal e
ωAs an input of the rotational speed regulator 56, the output signal is obtained after being regulated by the rotational speed regulator 56
The obtained rotating speed
As an input of the rotational speed integral type pseudo linear subsystem 43, the rotational speed integral type pseudo linear subsystem 43 outputs a rotational speed ω
e. In the implementation process of the invention, an anti-interference closed-loop controller 5 is formed by all the prefilters and all the regulators, wherein a given reference signal is slowed down by the prefilters, so that a tracking signal can better follow the reference signal, thereby improving the dynamic performance of displacement; the regulator adopts a PID core structure, the structure only needs to determine PID core parameters and bandwidth parameters, and the regulator can better regulate according to the engineering regulation experience of PIDAnd guiding engineering application. The parameters of the regulator are selected and adjusted according to the operation parameters of the coreless outer rotor bearingless permanent magnet synchronous motor, and the first pre-filter 51 and the second pre-filter 52 are selected in the form of
Wherein F
r(s) represents the transfer function of the pre-filter, ω
rAs a bandwidth parameter of the filter, 50, x and y displacements were set
And
filtered by the corresponding first pre-filter 51 and second pre-filter 52 to obtain
And
according to setting different bandwidth parameters omega of filter
rThe dynamic performance of the system can be adjusted, thereby improving the bandwidth of the whole displacement control system. The first displacement regulator 54 and the second displacement regulator 55 are both selected in the form of
In which C
PID(s) representing a transfer function of a displacement regulator having a PID kernel; k
PIDFor PID kernel parameter, set to 3, ω
cRepresenting the desired bandwidth, ω
cThe first displacement adjuster 54 and the second displacement adjuster 55 set the displacement error e in the x and y directions to 50, λ represents a weight coefficient and is set to 0.01
xAnd e
yAs input, regulated to obtain output
And
in the process, only the kernel parameter K of PID is needed
PID、ω
cAnd λ, and ω
cThe product of λ and λ is typically less than 1, where the product of the two is set to 0.5. The third pre-filter 53 is also optionally of the form
In which F
r(s) represents the transfer function of the pre-filter, the bandwidth parameter ω of the filter
rSet to 500, rotation speed command value
As the input of the third pre-filter 53, the filtered rotation speed command value is obtained after filtering
By setting different bandwidth parameters omega in the third pre-filter 53
rA corresponding control performance can be obtained, where the bandwidth parameter ω
rSet to 500. The speed regulator 56 takes the form of
The PI regulator of (1), wherein C
PI(s) represents the transfer function of a tachometer regulator with a PI kernel, K
PIAs a PI nuclear parameter, K
PIIs set to 5, omega
cRepresenting the desired bandwidth, set to 500, rotational speed error e
ωAs an input to the speed regulator 56, an output is obtained after regulation by the speed regulator 56
The output of the disturbance rejection closed-loop regulator 5 is
And
it is used as 3 inputs of generalized inverse 3 of least square support vector machineAdjusting generalized inverse 3 of the support vector machine of the least squares to obtain the output reference current
And will be referenced to the current
As an input of the composite controlled
object 28, the displacement x can be detected by the displacement sensor and the rotation speed sensor
M,y
MAnd an output rotation speed omega
eThe whole controller adopts the structure of the anti-interference closed-loop regulator, so that the control parameters of the whole controller are directly related to the bandwidth, and the anti-interference and decoupling capacity of the system is improved by configuring the corresponding bandwidth coefficient.
7) And forming an anti-interference least square support vector machine generalized inverse decoupling controller 6. The anti-interference closed-loop controller 5 is connected in series before the least square support vector machine generalized inverse 3, and the anti-interference closed-loop controller 5, the least square support vector machine generalized inverse 3 and the composite controlled object 28 jointly form the anti-interference least square support vector machine generalized inverse decoupling controller 6 of the coreless outer rotor bearingless permanent magnet synchronous motor, namely the motor decoupling controller, as shown in fig. 6.
From the foregoing, it will be appreciated that those skilled in the art, upon attaining an object of the present invention, may readily produce alterations to, variations of, and departures from the spirit and scope of the present invention.