CN110048657A - Induction-type bearingless motor Neural Network Inverse System decoupling control method - Google Patents
Induction-type bearingless motor Neural Network Inverse System decoupling control method Download PDFInfo
- Publication number
- CN110048657A CN110048657A CN201910431278.7A CN201910431278A CN110048657A CN 110048657 A CN110048657 A CN 110048657A CN 201910431278 A CN201910431278 A CN 201910431278A CN 110048657 A CN110048657 A CN 110048657A
- Authority
- CN
- China
- Prior art keywords
- induction
- bearingless motor
- type bearingless
- neural network
- rotor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
- H02P23/0004—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
- H02P23/0018—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control using neural networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
- H02P23/0004—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
- H02P23/0022—Model reference adaptation, e.g. MRAS or MRAC, useful for control or parameter estimation
Abstract
Induction-type bearingless motor Neural Network Inverse System decoupling control method establishes induction-type bearingless motor system state equation on the basis of considering torque wound stator electric current dynamic characteristic;Then, according to the nonlinear coupling relationship between inverse system input and output variable, the identification of induction-type bearingless motor inverse system mathematical model is carried out with neural network;Secondly, the Neural Network Inverse System mathematical model of induction-type bearingless motor is connected with its original system using Inverse Decoupling method, induction-type bearingless motor is decoupled into the independent second-order linearity subsystem of rotor flux, revolving speed and two radial displacement components etc. four;Finally, configuring closed-loop regulator to each subsystem, the inverse system dynamic Decoupling Control of Load Torque of induction-type bearingless motor is completed, new special motor driven and control field are belonged to, is particularly suitable for the high-performance magnetism suspension operation control application of induction-type bearingless motor.
Description
Technical field
The present invention relates to extraordinary alternating current generator detection control technology field more particularly to induction-type bearingless motor neural networks
Inverse Decoupling control method.
Background technique
Bearing-free motor is the similitude based on magnetic bearing Yu alternating-current motor stator structure, in developed in recent years suitable
Together in the New-type electric machine to run at high speed, before the fields such as aerospace, material sealing transmission, advanced manufacture have a wide range of applications
Scape.Find there is complicated electromagnetic coupling relationship inside induction-type bearingless motor, both at home and abroad to existing literature and patent retrieval
Inverse Decoupling control research had been carried out to it, but has considered that the parsing inverse system model after electric current dynamic is more complicated, it is inconvenient
In engineer application.Neural network, which has, approaches complex nonlinear function and self-learning capability with arbitrary accuracy, according to nerve net
Network recognizes the inverse system model of induction-type bearingless motor, and the realization that can effectively solve " parsing inverse system " is difficult;But existing nothing
The control research of bearing asynchronous motor neural network inverse system, does not all account for stator current dynamic characteristic, limits to a certain extent
The raising of its dynamic control performance is made.For this purpose, this patent proposes shaftless under conditions of considering stator current dynamic characteristic
Hold asynchronous motor neural network inverse system decoupling control method.
Summary of the invention
In order to solve the above technical problems, the present invention provides a kind of decouplings of induction-type bearingless motor Neural Network Inverse System to control
Method processed.
Used technical solution is the present invention to solve above-mentioned technical problem: induction-type bearingless motor nerve network reverse system
System decoupling control method, comprising the following steps:
Step 1: establishing the non-linear relation model of induction-type bearingless motor
Alpha-beta is set as static two-phase orthogonal coordinate system, the axis of α reference axis and three-phase induction-type bearingless motor A phase torque winding
Line direction is consistent, and it is synchronous as torque system rotor flux linkage orientation to set d-q in the counterclockwise vertical direction of α reference axis for β reference axis
Rotating coordinate system establishes induction-type bearingless motor original system model:
In formula (8),,,,,;R s1 For torque winding resistance;R r1 For rotor resistance;L m1It is in d-q coordinate system equivalent two
The mutual inductance of phase torque system;L s1 For the self-induction of the equivalent two-phase torque winding in d-q coordinate system,L s1 = L m1 +L s1l ;L r1 For d-q
The self-induction of equivalent two-phase rotor windings in coordinate system,L r1 =L m1 +L r1l ;L s1l 、L r1l The respectively stator and rotor leakage of torque system
Sense;For magnetic leakage factor;T L For load torque;p 1For the number of magnetic pole pairs of torque winding;
According to implicit function theorem, induction-type bearingless motor inverse system non-linear relation model is established:
;
Step 2: establishing induction-type bearingless motor Neural Network Inverse System model
Three_layer planar waveguide is selected, the excitation function using tanh S type transfer function as hidden layer neuron:
The input variable and output variable of induction-type bearingless motor are sampled, the original of input variable and output variable is obtained
Data sample, and the disposal of gentle filter is carried out to the initial data that sampling obtains, it acquires and turns further according to the data after smothing filtering
Sub- magnetic linkage, revolving speed, the one of α radial displacement component and β radial displacement component, second dervative, to constitute neural metwork training sample
This collectionWith;
Calculating is optimized to train samples collection by genetic algorithm, using training data prediction Error Absolute Value
Judgement numerical value of the summation numerical value as train samples collection individual adaptation degree is chosen and determines that numerical value is the smallest as individual
The optimum individual of fitness carries out assignment using initial weight and threshold value of the optimum individual to neural network, is then calculated by LM
Method carries out identification training to neural network, guarantees the output mean square error of neural network less than 0.001, so that it is determined that neural network
Each layer neuron between connection weight coefficient, can be obtained according to induction-type bearingless motor inverse system non-linear relation model
Induction-type bearingless motor Neural Network Inverse System model;
Step 3: establishing induction-type bearingless motor nerve network reverse decoupling and controlling system
The output of induction-type bearingless motor Neural Network Inverse System is concatenated with the input of induction-type bearingless motor original system, and will
Four closed-loop regulators are concatenated with the input of induction-type bearingless motor Neural Network Inverse System respectively, by induction-type bearingless motor original
The numerical value of rotor flux, revolving speed, α radial displacement component and β radial displacement component that system exports is respectively respective one with its
Given value is compared operation, then four comparison operation differences are respectively fed to a respective closed-loop regulator using as shaftless
The input quantity of asynchronous motor neural network inverse system is held, i.e. composition induction-type bearingless motor nerve network reverse decoupling and controlling system,
To realize induction-type bearingless motor Neural Network Inverse System decoupling control.
Preferably, the method for building up of induction-type bearingless motor original system model are as follows:
Establish torque system nonlinear dynamic mathematical model:
In formula (1):Respectively indicate voltage, electric current, resistance, magnetic linkage;Subscript s1Withr 1Respectively indicate torque system
The stator and rotor of system;Subscriptd、qIt respectively indicates in rotor flux coordinate systemd、qAxis component;It respectively indicates
Mutual inductance between the rotor self-induction of torque system, stator self inductance, rotor;L s1l 、L r1l Respectively torque system existsd-qIn coordinate system
Stator and rotor leakage inductance;p 1For the number of magnetic pole pairs of torque winding;T L For load torque;T r1 For rotor time constant;ω rFor rotor rotation
Angular speed;ω 1For the synchronization angular rate of motor.
According to magnetic suspension control principle, the controllable radial magnetic suspension force model of induction-type bearingless motor is established:
In formula (2):K m It is the magnetic suspension force coefficient determined by electric machine structure,F α、F βFor along staticα、βCoordinate it is axial can control diameter
To magnetic suspension force component;i s2d 、i s2q Respectively the two poles of the earth suspending windingsd、qShaft current component;Torque system respectively
Air gap flux linkaged、qAxis component,Expression formula are as follows:
Rotor radial suspended motion equation:
In formula (4):mFor rotor quality;f α、f βRespectively rotor eccentricity when because air-gap field be distributed it is uneven due to generate it is unilateral
Electromagnet pull component,, thereink sFor the radial displacement rigidity for being decided by electric machine structure and magnetic field strength
Coefficient,α、βFor the rotor radial displacement component in alpha-beta coordinate system;
Consider torque winding current dynamic differential equation, respectively control variable, state variable and the output variable of selecting system are as follows:
Wherein:For torque windingd、qShaft voltage component;i s2d、i s2qRespectively the two poles of the earth suspending windingsd、qAxis
Current component;α, β are the rotor radial displacement component in alpha-beta coordinate system;For the rotor flux of torque system;ω rFor rotor
Angular velocity of rotation;
(5), (6), (7) formula are substituted into (1)~(4) formula, obtain eight rank state equations of induction-type bearingless motor original system model.
Preferably, in step 2, select input number of nodes be 12, output node number is 4, node in hidden layer is 23 three
Layer feedforward neural network.
Preferably, in step 3, the closed-loop regulator uses the PD adjuster with low-pass filtering link, closed loop control
The transmission function of device processed are as follows:
In formula (17),For advanced derivative time constant;For low-pass filtering link time constant.
According to the above technical scheme, the beneficial effects of the present invention are:
1, the present invention has fully taken into account objective inside motor deposit in considering the dynamic Neural Network Inverse System inversion model of electric current
Stator current dynamic characteristic influence, thus the case where compared to when not considering stator current dynamic characteristic, can obtain higher
Magnetic suspension operation state decoupling control performance.
2, given induction-type bearingless motor Neural Network Inverse System decoupling control method can simplify whole to a certain extent
Control system architecture.When training recognizes induction-type bearingless motor Neural Network Inverse System model, stator current dynamic is had contemplated that
And the cross-coupling characteristics between component, and no longer needed in inverse system model it is difficult to predict load torque variable, therefore
The real-time identification link that the torque winding current in original system adjusts closed loop and load torque can be omitted.
3, using neural network to nonlinear function approach and learning ability recognizes to obtain and considers that electric current is dynamically shaftless
After holding asynchronous machine inverse system model, by Inverse Decoupling, the excellent solution after considering stator current dynamic can be not only obtained
Coupling performance, moreover it is possible to overcome the problems, such as that complicated parsing inverse system model is not easy to realize, and with also have that overshoot is small, tracking
Performance is good, to the strong robustness of Parameters variation the features such as.
Detailed description of the invention
Fig. 1 is the topology diagram that neural metwork training recognizes induction-type bearingless motor inverse system model;
Fig. 2 is the Neural Network Inverse System decoupling principle figure of induction-type bearingless motor;
Fig. 3 is the principle assumption diagram of induction-type bearingless motor nerve network reverse decoupling and controlling system.
Specific embodiment
Induction-type bearingless motor Neural Network Inverse System decoupling control method is considering that torque wound stator electric current dynamic is special
On the basis of property, induction-type bearingless motor system state equation is established;Then, according between inverse system input and output variable
Nonlinear coupling relationship, with neural network carry out induction-type bearingless motor inverse system mathematical model identification;Secondly, using inverse system
System decoupling method, the Neural Network Inverse System mathematical model of induction-type bearingless motor is connected with its original system, and bearing-free is different
Step motor is decoupled into the independent second-order linearity subsystem of rotor flux, revolving speed and two radial displacement components etc. four;Finally, giving
Each subsystem configures closed-loop regulator, completes the inverse system dynamic Decoupling Control of Load Torque of induction-type bearingless motor, belongs to new special electricity
Machine driving and control field are particularly suitable for the high-performance magnetism suspension operation control application of induction-type bearingless motor.
Patented invention principle foundation:
1, induction-type bearingless motor is a multivariable, non-linear, close coupling complex object, wherein torque system and magnetic suspension
There are complicated mechanical, electrical, magnetic coupling relation between system to be parsed according to the working principle of induction-type bearingless motor by mechanism,
It can derive the non-linear relation between correlated variables.
2, neural network has the ability that complex nonlinear function and self study are approached with arbitrary accuracy;If can be by offline
Training, according to the coupled relation between correlated variables, identification study obtains considering the dynamic induction-type bearingless motor of stator current
Inverse system mathematical model, the inverse system dynamic Decoupling Control of Load Torque of induction-type bearingless motor can be carried out, and can effectively solve the problem that
Complicated parsing inverse system mathematical model realizes difficult problem.
To achieve the goals above, a kind of technological means that the present invention takes are as follows: induction-type bearingless motor nerve network reverse
System decoupling control method, comprising the following steps:
1) the non-linear relation mathematical model of induction-type bearingless motor is established
Set: alpha-beta is static two-phase orthogonal coordinate system, the axis of α reference axis and three-phase induction-type bearingless motor A phase torque winding
Line direction is consistent, and β reference axis is in the counterclockwise vertical direction of α reference axis;D-q is torque system rotor flux linkage orientation synchronous rotary
Coordinate system.In view of stator current dynamical equation, rotor field-oriented constraint condition "" etc.,
Torque system nonlinear dynamic mathematical model can be obtained:
In formula (1):Respectively indicate voltage, electric current, resistance, magnetic linkage;Subscript s1Withr 1Respectively indicate torque system
The stator and rotor of system;Subscriptd、qIt respectively indicates in rotor flux coordinate systemd、qAxis component;It respectively indicates
Mutual inductance between the rotor self-induction of torque system, stator self inductance, rotor;L s1l 、L r1l Respectively torque system existsd-qIn coordinate system
Stator and rotor leakage inductance;p 1For the number of magnetic pole pairs of torque winding;T L For load torque;T r1 For rotor time constant;ω rFor rotor rotation
Angular speed;ω 1For the synchronization angular rate of motor.
According to magnetic suspension control principle, the controllable radial magnetic suspension force model of induction-type bearingless motor can be obtained:
In formula (2):K m It is the magnetic suspension force coefficient determined by electric machine structure,F α、F βFor along staticα、βCoordinate it is axial can control diameter
To magnetic suspension force component;i s2d 、i s2q Respectively the two poles of the earth suspending windingsd、qShaft current component;Torque system respectively
Air gap flux linkaged、qAxis component,Expression formula are as follows:
Rotor radial suspended motion equation:
In formula (4):mFor rotor quality;f α、f βRespectively rotor eccentricity when because air-gap field be distributed it is uneven due to generate it is unilateral
Electromagnet pull component,, thereink sFor the radial displacement rigidity for being decided by electric machine structure and magnetic field strength
Coefficient,α、βFor the rotor radial displacement component in alpha-beta coordinate system;
Consider torque winding current dynamic differential equation, respectively control variable, state variable and the output variable of selecting system are as follows:
Wherein:, for torque windingd、qShaft voltage component;i s2d、i s2qRespectively the two poles of the earth suspending windingsd、qShaft current component;
α, β are the rotor radial displacement component in alpha-beta coordinate system;For the rotor flux of torque system;ω rFor rotor rotation angle speed
Degree.
(5), (6), (7) formula are substituted into (1)~(4) formula, the eight rank state equations that can obtain system are as follows:
In formula (8),,, , , ;R s1 For torque winding resistance;R r1 For rotor resistance;L m1It is in d-q coordinate system equivalent two
The mutual inductance of phase torque system;L s1 For the self-induction of the equivalent two-phase torque winding in d-q coordinate system,L s1 = L m1 +L s1l ;L r1 For d-q
The self-induction of equivalent two-phase rotor windings in coordinate system,L r1 =L m1 +L r1l ;L s1l 、L r1l The respectively stator and rotor leakage of torque system
Sense;For magnetic leakage factor;T L For load torque;p 1For the number of magnetic pole pairs of torque winding.
System invertibility analysis is carried out using Interactor algorithm.To outputWhen gradually asking
Between derivative, until in the derivative of output variable show contain input control quantityUntil.Specific derivation process is such as
Under:
Meanwhile according to selected state variable, formula (3) becomes:
It enables, then according to (9)~(12) formula, system Jacobi matrix can be obtained:
During system operates normally,dThe rotor flux of axis, air gap flux linkage value are not 0.Therefore det (A) ≠ 0,rank(A)=4, it is refined
It gram is full rank than matrix.The Relative order of system are as follows: α=(α1, α2, α3, α4)=(2,2,2,2), and meet:, i.e., the sum of Relative order of system be equal to system state equation order, then according to Inverse System Theory it is found that by formula
(8) the induction-type bearingless motor system described is reversible;
According to implicit function theorem, and convolution (9)~(12), it can be the induction-type bearingless motor of consideration stator current dynamic characteristic
Inverse system non-linear relation model indicates are as follows:
2) the training identification of Neural Network Inverse System model
As shown in Figure 1, select input number of nodes be 12, output node number is 4, node in hidden layer be selected as 23 three layers of feedforward net
Network carries out inverse system Model Distinguish, therein Respectively
For the connection weight between input layer and hidden layer, hidden layer and output layer neuron;
Excitation function using tanh S type transfer function as hidden layer neuron, specifically:
Consider the dynamic induction-type bearingless motor Neural Network Inverse System mathematical model of stator current, specific training identification step
It is rapid as follows:
(1) sampling obtains input, output primary data sample.Based on the dynamic induction-type bearingless motor parsing of consideration stator current
Adverse control system chooses the random quantity by normal distribution of OK range as sharp according to the actual physics operation area of motor
Signal is encouraged, the entire actuation duration is 10s.To the input variable of induction-type bearingless motor (suspending windings biphase current and torque around
Group two-phase voltage) and export (two radial displacement components, rotor flux, revolving speeds) sampled, sampling step length is taken as 0.001s;
(2) trained and test sample collection is obtained.After carrying out smothing filtering to the initial data sampled, using " high-precision five
Point value differential " method, can acquire two radial displacement components, rotor flux and revolving speed one, second dervative, to constitute mind
Through training sample collectionWith.?
The 10000 groups of data equal intervals obtained choose 6000 groups and are used as training set, and remaining 4000 groups are used as test set.In practical application
In, sample the initial data of acquisition, and " derivative " data handled through numerical differentiation may not in an order of magnitude, because
This is often normalized training sample set.The convergence for being conducive to neural metwork training is done so, it can be to avoid nerve net
Network is especially sensitive or insensitive to a certain input quantity;
(3) the training identification of Neural Network Inverse System model.Firstly, using " the initial power of genetic algorithm optimization BP neural network
Value " and " threshold value " recognize induction-type bearingless motor inverse system model training to avoid neural network initial weight and threshold value and tie
The influence of fruit.Each individual in Population in Genetic Algorithms all includes a network ownership value and threshold value, and individual is by " adapting to
Spend function " individual adaptation degree is calculated, genetic algorithm finds adaptive optimal control angle value corresponding by selection, intersection and mutation operation
Body;Wherein, individual adaptation degree is used as using " sum of training data prediction Error Absolute Value ", the value of individual adaptation degree is smaller, should
Individual is more excellent.Then, with optimum individual to network initial weight and threshold value assignment.Finally, using LM algorithm to quiet in Fig. 1
State neural network is trained.After training by 1000 times or so, the output mean square error of neural network less than 0.001,
It meets the requirements, so that it is determined that each weight coefficient in the static neural network topological structure in Fig. 1, obtains and consider stator current
Dynamic induction-type bearingless motor Neural Network Inverse System model.
3) the nerve network reverse decoupling and controlling system of induction-type bearingless motor is constructed
Firstly, carrying out the Neural Network Inverse System decoupling of induction-type bearingless motor.To output variable
Acceleration variableAdditional credits device, to obtainEqual variables, as nerve network reverse
The input quantity of system model;Before Neural Network Inverse System is serially connected in original system again, induction-type bearingless motor system decoupling
Subsystem, including rotor flux subsystem, revolving speed subsystem, α radial displacement component subsystem are integrated at four (puppet) linear second-orders
System and β radial displacement component subsystem, the equivalent transfer function of each subsystem are 1/s2, it is illustrated in figure 2 nerve network reverse system
System decoupling principle figure.
Then, band closed-loop regulator is configured to subsystems.From lineary system theory: being 1/ for transmission function
s2Subsystem, good control effect can get using PD adjuster.But it makes an uproar to reduce high band system gain, resisting high frequency
Sound shadow is rung, using " the modified PD adjuster " structure for having low-pass filtering link, the transmission function of specific closed loop controller
Are as follows:
In formula (17),For advanced derivative time constant;For low-pass filtering link time constant.
After adding closed-loop regulator, each linear subsystem is corrected as typical type-Ⅱ system, i.e. the open loop transmitting of subsystem
Function is corrected as:
According to the dynamic performance index requirement of each subsystem and formula (17) controller structure, a rotor flux can be separately designed
Adjuster, a speed regulator, a α radial displacement component adjuster and a β radial displacement component adjuster.
Finally, the nerve network reverse decoupling and controlling system of building induction-type bearingless motor, Fig. 3, which is shown, considers electric current dynamic
Induction-type bearingless motor nerve network reverse decoupling and controlling system structure chart.First the given value of rotor flux and its value of feedback
It is comprehensively compared, then rotor flux error is sent into flux regulating device, using the output quantity of flux regulating device as the two of rotor flux
Order derivative Setting signal, it is sent to the correspondence signal input part of Neural Network Inverse System;The given value and its value of feedback of revolving speed
It is comprehensively compared, then speed error is sent into speed regulator, given using the output quantity of speed regulator as the second dervative of revolving speed
Determine signal, it is sent to the correspondence signal input part of Neural Network Inverse System;The given value of α radial displacement component is fed back with it
Value is comprehensively compared, then radial displacement error is sent into α to displacement governor, using output quantity from α to displacement governor as α to position
The second dervative Setting signal of shifting, it is sent to the correspondence signal input part of Neural Network Inverse System;Company of the β to displacement governor
It is similar to displacement governor with α to connect method.The closed-loop regulator of rotor flux, revolving speed, α to displacement component β to displacement component with
Neural Network Inverse System mathematical model, collectively forms the composite controller of induction-type bearingless motor, to realize that bearing-free is asynchronous
The Neural Network Inverse System dynamic Decoupling Control of Load Torque of motor.
Claims (4)
1. induction-type bearingless motor Neural Network Inverse System decoupling control method, which comprises the following steps:
Step 1: establishing the non-linear relation model of induction-type bearingless motor
Alpha-beta is set as static two-phase orthogonal coordinate system, the axis of α reference axis and three-phase induction-type bearingless motor A phase torque winding
Line direction is consistent, and it is synchronous as torque system rotor flux linkage orientation to set d-q in the counterclockwise vertical direction of α reference axis for β reference axis
Rotating coordinate system establishes induction-type bearingless motor original system model:
In formula (8),,, , ,;R s1 For torque winding resistance;R r1 For rotor resistance;L m1For the mutual of two-phase torque system equivalent in d-q coordinate system
Sense;L s1 For the self-induction of the equivalent two-phase torque winding in d-q coordinate system,L s1 = L m1 +L s1l ;L r1 It is equivalent in d-q coordinate system
The self-induction of two-phase rotor windings,L r1 =L m1 +L r1l ;L s1l 、L r1l The respectively stator and rotor leakage inductance of torque system;For magnetic leakage factor;T L For load torque;p 1For the number of magnetic pole pairs of torque winding;
According to implicit function theorem, induction-type bearingless motor inverse system non-linear relation model is established:
Step 2: establishing induction-type bearingless motor Neural Network Inverse System model
Three_layer planar waveguide is selected, the excitation function using tanh S type transfer function as hidden layer neuron:
The input variable and output variable of induction-type bearingless motor are sampled, the original of input variable and output variable is obtained
Data sample, and the disposal of gentle filter is carried out to the initial data that sampling obtains, it acquires and turns further according to the data after smothing filtering
Sub- magnetic linkage, revolving speed, the one of α radial displacement component and β radial displacement component, second dervative, to constitute neural metwork training sample
This collectionWith;
Calculating is optimized to train samples collection by genetic algorithm, using training data prediction Error Absolute Value
Judgement numerical value of the summation numerical value as train samples collection individual adaptation degree is chosen and determines that numerical value is the smallest as individual
The optimum individual of fitness carries out assignment using initial weight and threshold value of the optimum individual to neural network, is then calculated by LM
Method carries out identification training to neural network, guarantees the output mean square error of neural network less than 0.001, so that it is determined that neural network
Each layer neuron between connection weight coefficient, can be obtained according to induction-type bearingless motor inverse system non-linear relation model
Induction-type bearingless motor Neural Network Inverse System model;
Step 3: establishing induction-type bearingless motor nerve network reverse decoupling and controlling system
The output of induction-type bearingless motor Neural Network Inverse System is concatenated with the input of induction-type bearingless motor original system, and will
Four closed-loop regulators are concatenated with the input of induction-type bearingless motor Neural Network Inverse System respectively, by induction-type bearingless motor original
The numerical value of rotor flux, revolving speed, α radial displacement component and β radial displacement component that system exports is respectively respective one with its
Given value is compared operation, then four comparison operation differences are respectively fed to a respective closed-loop regulator using as shaftless
The input quantity of asynchronous motor neural network inverse system is held, i.e. composition induction-type bearingless motor nerve network reverse decoupling and controlling system,
To realize induction-type bearingless motor Neural Network Inverse System decoupling control.
2. induction-type bearingless motor Neural Network Inverse System decoupling control method according to claim 1, it is characterised in that:
The method for building up of induction-type bearingless motor original system model are as follows:
Establish torque system nonlinear dynamic mathematical model:
In formula (1):Respectively indicate voltage, electric current, resistance, magnetic linkage;Subscript s1Withr 1Respectively indicate torque system
The stator and rotor of system;Subscriptd、qIt respectively indicates in rotor flux coordinate systemd、qAxis component;It respectively indicates
Mutual inductance between the rotor self-induction of torque system, stator self inductance, rotor;L s1l 、L r1l Respectively torque system existsd-qIn coordinate system
Stator and rotor leakage inductance;p 1For the number of magnetic pole pairs of torque winding;T L For load torque;T r1 For rotor time constant;For rotor rotation angle
Speed;ω 1For the synchronization angular rate of motor;
According to magnetic suspension control principle, the controllable radial magnetic suspension force model of induction-type bearingless motor is established:
In formula (2):K m It is the magnetic suspension force coefficient determined by electric machine structure,F α、F βFor along staticα、βCoordinate it is axial can control diameter
To magnetic suspension force component;i s2d 、i s2q Respectively the two poles of the earth suspending windingsd、qShaft current component;Torque system respectively
Air gap flux linkaged、qAxis component,Expression formula are as follows:
Rotor radial suspended motion equation:
In formula (4):mFor rotor quality;f α、f βRespectively rotor eccentricity when because air-gap field be distributed it is uneven due to generate it is unilateral
Electromagnet pull component,, thereink sFor the radial displacement rigidity for being decided by electric machine structure and magnetic field strength
Coefficient,α、βFor the rotor radial displacement component in alpha-beta coordinate system;
Consider torque winding current dynamic differential equation, respectively control variable, state variable and the output variable of selecting system are as follows:
Wherein:For torque windingd、qShaft voltage component;i s2d、i s2qRespectively the two poles of the earth suspending windingsd、qAxis
Current component;α, β are the rotor radial displacement component in alpha-beta coordinate system;For the rotor flux of torque system;For rotor rotation
Tarnsition velocity;
(5), (6), (7) formula are substituted into (1)~(4) formula, obtain eight rank state equations of induction-type bearingless motor original system model.
3. induction-type bearingless motor Neural Network Inverse System decoupling control method according to claim 1, it is characterised in that:
In step 2, select input number of nodes be 12, the three_layer planar waveguide that output node number is 4, node in hidden layer is 23.
4. induction-type bearingless motor Neural Network Inverse System decoupling control method according to claim 1, it is characterised in that:
In step 3, the closed-loop regulator uses the PD adjuster with low-pass filtering link, the transmission function of closed loop controller
Are as follows:
In formula (17),For advanced derivative time constant;For low-pass filtering link time constant.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910431278.7A CN110048657B (en) | 2019-05-22 | 2019-05-22 | Decoupling control method for neural network inverse system of bearingless asynchronous motor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910431278.7A CN110048657B (en) | 2019-05-22 | 2019-05-22 | Decoupling control method for neural network inverse system of bearingless asynchronous motor |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110048657A true CN110048657A (en) | 2019-07-23 |
CN110048657B CN110048657B (en) | 2020-10-16 |
Family
ID=67283314
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910431278.7A Expired - Fee Related CN110048657B (en) | 2019-05-22 | 2019-05-22 | Decoupling control method for neural network inverse system of bearingless asynchronous motor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110048657B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114489166A (en) * | 2020-10-27 | 2022-05-13 | 通用汽车环球科技运作有限责任公司 | Method, system and apparatus for torque control using a root of a pseudo-ANN |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0775398A (en) * | 1993-03-15 | 1995-03-17 | Hosei Ho | Vector controller for induction motor |
US7346462B2 (en) * | 2006-03-29 | 2008-03-18 | General Electric Company | System, method, and article of manufacture for determining parameter values associated with an electrical grid |
CN105553365A (en) * | 2016-02-01 | 2016-05-04 | 四川长虹电器股份有限公司 | Automatic identification control method for parameters of permanent magnet brushless motor |
JP6210936B2 (en) * | 2014-05-30 | 2017-10-11 | ミネベアミツミ株式会社 | Self-excited resonance type power factor correction circuit and light source driving device |
CN109217766A (en) * | 2018-09-26 | 2019-01-15 | 河南科技大学 | The independent reversed decoupling control system of induction-type bearingless motor |
CN109217761A (en) * | 2018-09-26 | 2019-01-15 | 河南科技大学 | The inverse kinematics of current-control type induction-type bearingless motor decouple System with Sliding Mode Controller |
-
2019
- 2019-05-22 CN CN201910431278.7A patent/CN110048657B/en not_active Expired - Fee Related
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0775398A (en) * | 1993-03-15 | 1995-03-17 | Hosei Ho | Vector controller for induction motor |
US7346462B2 (en) * | 2006-03-29 | 2008-03-18 | General Electric Company | System, method, and article of manufacture for determining parameter values associated with an electrical grid |
JP6210936B2 (en) * | 2014-05-30 | 2017-10-11 | ミネベアミツミ株式会社 | Self-excited resonance type power factor correction circuit and light source driving device |
CN105553365A (en) * | 2016-02-01 | 2016-05-04 | 四川长虹电器股份有限公司 | Automatic identification control method for parameters of permanent magnet brushless motor |
CN109217766A (en) * | 2018-09-26 | 2019-01-15 | 河南科技大学 | The independent reversed decoupling control system of induction-type bearingless motor |
CN109217761A (en) * | 2018-09-26 | 2019-01-15 | 河南科技大学 | The inverse kinematics of current-control type induction-type bearingless motor decouple System with Sliding Mode Controller |
Non-Patent Citations (4)
Title |
---|
WEIMING SUN 等: "Speed sensorless of a bearingless induction motor based on super-twisting-model reference adaptive system", 《2017 20TH INTERNATIONAL CONFERENCE ON ELECTRICAL MACHINES AND SYSTEMS (ICEMS)》 * |
WENSHAO BU 等: "Stator Flux Orientation Inverse System Decoupling Control Strategy of Bearingless Induction Motor Considering Stator Current Dynamics", 《IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING》 * |
何方舟 等: "无轴承异步电机稳定性能优化控制研究", 《计算机仿真》 * |
杨泽斌,汪明涛,孙晓东: "基于转矩绕组无功功率MRAS的无轴承异步电机无速度传感器矢量控制系统", 《四川大学学报(工程科学版)》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114489166A (en) * | 2020-10-27 | 2022-05-13 | 通用汽车环球科技运作有限责任公司 | Method, system and apparatus for torque control using a root of a pseudo-ANN |
Also Published As
Publication number | Publication date |
---|---|
CN110048657B (en) | 2020-10-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110429881B (en) | Active-disturbance-rejection control method of permanent magnet synchronous motor | |
CN100433537C (en) | Method for controlling bearing-less AC asynchronous motor neural network inverse decoupling controller | |
Poza et al. | Vector control design and experimental evaluation for the brushless doubly fed machine | |
CN107276473B (en) | Permanent-magnet synchronous motor with five degrees of freedom without bearing Fuzzy Neural Network Decoupling controller | |
CN109217766A (en) | The independent reversed decoupling control system of induction-type bearingless motor | |
CN110784140A (en) | Control strategy for five-phase asynchronous motor according to stator magnetic field orientation | |
Bu et al. | Sliding mode variable structure control strategy of bearingless induction motor based on inverse system decoupling | |
CN102361429A (en) | Bearing-free asynchronous motor control method based on neural network inverse system theory | |
CN101814892A (en) | Support vector machine inverse system composite controller based on bearingless synchronous reluctance motor | |
CN103501148A (en) | Method for controlling operation of non-radial displacement sensor of bearingless permanent magnetic synchronous motor | |
CN110138299A (en) | Induction-type bearingless motor reversed decoupling control system based on rotor resistance on-line identification | |
CN102097986A (en) | Construction method for neural network generalized inverse decoupling controller of bearing-free synchronous reluctance motor | |
CN102136822B (en) | Five-DOF (freedom of degree) bearingless synchronous reluctance motor decoupling controller and construction method thereof | |
CN102790577A (en) | Constructing method for suspended subsystem controller of bearingless permanent magnet synchronous motor | |
CN102130647B (en) | Speed-sensor-less construction method for detecting rotating speed of bearing-less asynchronous motor | |
CN202043069U (en) | Decoupling controller of five-degree-freedom bearingless synchronous reluctance motor | |
CN110048657A (en) | Induction-type bearingless motor Neural Network Inverse System decoupling control method | |
CN104767452A (en) | Self-adaptative inverse decoupling control method based on non-linear filters for bearing-free asynchronous motor | |
CN201928221U (en) | Neural net generalized inverse decoupling controller for bearingless synchronous reluctance motor | |
CN102790578B (en) | Manufacturing method of bearingless asynchronous motor neural network generalized inverse decoupling controller | |
Bu et al. | A control method of bearingless induction motor based on neural network | |
CN115001335A (en) | Bearing-free flux switching motor rotor suspension control method based on neural network | |
Ji et al. | Design of neural network PID controller based on brushless DC motor | |
CN102790576A (en) | Construction method for bearing-less permanent-magnet synchronous motor decoupling controller | |
Sahraoui et al. | Performance enhancement of sensorless speed control of DSIM using MRAS and EKF optimized by genetic algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201016 |