CN110224649B - Method for DTC prediction control based on support vector machine - Google Patents

Method for DTC prediction control based on support vector machine Download PDF

Info

Publication number
CN110224649B
CN110224649B CN201910595422.0A CN201910595422A CN110224649B CN 110224649 B CN110224649 B CN 110224649B CN 201910595422 A CN201910595422 A CN 201910595422A CN 110224649 B CN110224649 B CN 110224649B
Authority
CN
China
Prior art keywords
output
prediction model
dtc
vector machine
support vector
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.)
Active
Application number
CN201910595422.0A
Other languages
Chinese (zh)
Other versions
CN110224649A (en
Inventor
李耀华
周逸凡
秦玉贵
赵承辉
秦辉
苏锦仕
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changan University
Original Assignee
Changan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Changan University filed Critical Changan University
Priority to CN201910595422.0A priority Critical patent/CN110224649B/en
Publication of CN110224649A publication Critical patent/CN110224649A/en
Application granted granted Critical
Publication of CN110224649B publication Critical patent/CN110224649B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
    • H02P21/0003—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
    • H02P21/0017—Model reference adaptation, e.g. MRAS or MRAC, useful for control or parameter estimation
    • 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
    • H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
    • H02P21/24—Vector control not involving the use of rotor position or rotor speed sensors
    • H02P21/28—Stator flux based control
    • H02P21/30—Direct torque control [DTC] or field acceleration method [FAM]
    • 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
    • H02P27/00—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage
    • H02P27/04—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage
    • H02P27/06—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage using dc to ac converters or inverters
    • H02P27/08—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage using dc to ac converters or inverters with pulse width modulation
    • H02P27/12—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage using dc to ac converters or inverters with pulse width modulation pulsing by guiding the flux vector, current vector or voltage vector on a circle or a closed curve, e.g. for direct torque control
    • 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
    • H02P6/00—Arrangements for controlling synchronous motors or other dynamo-electric motors using electronic commutation dependent on the rotor position; Electronic commutators therefor
    • H02P6/10—Arrangements for controlling torque ripple, e.g. providing reduced torque ripple

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Control Of Ac Motors In General (AREA)

Abstract

The invention relates to a method for DTC predictive control based on a support vector machine, which comprises the steps of firstly selecting the existing motor operation parameters as input and output samples; respectively carrying out normalization processing on the selected input and output samples; training the normalized sample to obtain a Support Vector Machine (SVM) -based prediction model of the output voltage vector for the DTC; normalizing the current motor operation parameters, inputting the normalized current motor operation parameters into a prediction model, and outputting an applied basic voltage vector through the prediction model; and controlling the switching state of the inverter according to the basic voltage vector output by the prediction model, and completing the DTC prediction control based on the support vector machine. The invention can reduce the related calculation amount, reduce the torque ripple and ensure constant switching frequency.

Description

Method for DTC prediction control based on support vector machine
Technical Field
The invention belongs to the field of motor control, and particularly relates to a method for DTC predictive control based on a support vector machine.
Background
Direct Torque Control (DTC) controls electromagnetic torque by controlling the magnitude and load angle of the stator flux linkage. The direct torque control of the permanent magnet synchronous motor generally uses two hysteresis comparators to respectively control stator flux linkage and torque deviation, and because the hysteresis comparators avoid a large amount of calculation under a rotating coordinate system, the dynamic performance of the permanent magnet synchronous motor is improved, and the response is faster.
In a direct torque control system of a permanent magnet synchronous motor realized by a traditional switch meter, a voltage vector is continuously applied in a sampling period, and the increase and decrease of actual torque exceed expected requirements, so that overshoot pulsation is caused.
In order to solve the problems, prediction control is introduced, a cost function is introduced, a torque error and a stator flux linkage error are comprehensively considered and controlled, and a space vector modulation technology is adopted to reduce torque pulsation. However, some problems still exist in predictive control, firstly, the scientificity of the cost function is still to be researched, secondly, in order to enable the torque error and the flux linkage error to be on the same dimension, a weighting coefficient is usually introduced, and the selection of the weighting coefficient mostly depends on experience and lacks of powerful theoretical support. Meanwhile, with the predictive control type, the cost of calculating six basic voltage vectors at the same time is required, which is a large amount of calculation, contrary to the original intention of introducing direct torque control.
Disclosure of Invention
The invention aims to overcome the defects and provide a method for DTC predictive control based on a support vector machine, which can reduce the related calculation amount, reduce the torque ripple and ensure that the switching frequency is constant.
In order to achieve the above object, the present invention comprises the steps of:
the method comprises the following steps:
the method comprises the following steps: selecting the existing motor operation parameters as input and output samples;
step two: respectively carrying out normalization processing on the selected input and output samples;
step three: training the normalized sample to obtain a Support Vector Machine (SVM) -based prediction model of the output voltage vector for the DTC;
step four: normalizing the current operation parameters of the motor;
step five: inputting the operation parameters subjected to normalization processing in the fourth step into a prediction model, and outputting an applied basic voltage vector through the prediction model;
step six: and controlling the switching state of the inverter according to the basic voltage vector output by the prediction model, and completing the DTC prediction control based on the support vector machine.
Further, in the step one, the existing motor operation parameters are the motor operation parameters for selecting the voltage vector by using a cost function, and the cost function is shown as the following formula (1):
Figure BDA0002117498510000021
wherein
Figure BDA0002117498510000022
For the purpose of reference to the amplitude values,
Figure BDA0002117498510000023
for the purpose of reference to the torque,
Figure BDA0002117498510000024
is the actual amplitude, TeIs the actual torque.
Furthermore, the number of the cost functions is equal to the number of the basic voltage vectors, and the basic voltage vector corresponding to the minimum cost function is selected as the applied voltage vector; selecting stator flux linkage error psi under all working conditionsrOutput torque error TrThe load angle delta and the torque angle alpha are used as input for training a prediction model, and an applied voltage vector is selected as output.
Further, in the second step, the expression of normalizing the input and output samples is shown in the following formula (2):
Figure BDA0002117498510000025
wherein x isminIs the minimum value, x, of the input or output samplesmaxIs the maximum of the input samples or the output samples.
Further, in step three, a support vector machine is used for training.
Further, a radial basis function is adopted as a support vector machine kernel function, as shown in equation (3):
k(x，xi)＝exp(-γ‖x-xi‖2),γ＞0 (3)
where x is the input, xiTo support the vector, γ is a kernel parameter.
Further, dividing the input samples and the output samples selected in the step one into 3 groups, respectively making a primary verification set for the subset data of each group, taking the rest 2 groups of subset data as training sets to obtain 3 submodels, wherein each submodel respectively corresponds to a group of punishment parameters c and kernel function parameters g, and taking the average of the classification accuracy of the verification sets in the 3 submodels as the performance index of the classifier.
Further, if the average of the classification accuracy is greater than or equal to 90%, a group of parameters c and g with the highest classification accuracy is selected as parameters of the prediction model, and if the average of the classification accuracy is less than 90%, the training is carried out again.
Further, in step four, the stator flux linkage error ψ at time k is obtainedrError of output torque TrThe load angle δ and the torque angle α are followed by normalization.
Furthermore, the normalization processing mode in the fourth step is the same as that in the second step.
Compared with the prior art, the invention has the following beneficial technical effects:
the method obtains the operation data of the motor adopting the cost function, obtains a corresponding prediction model by off-line training by using a support vector machine method, and selects the voltage vector according to the current operation parameters of the motor by using the trained model, thereby realizing the rapid control of the motor. The method improves the performance of a direct torque control system of the permanent magnet synchronous motor, reduces related calculation amount, reduces torque pulsation and has constant switching frequency. According to the method, the prediction model which is well learned offline is introduced, so that the pulsation of the torque is reduced compared with direct torque control, and compared with prediction control, the introduction of a cost function is avoided, so that the huge calculation amount of the cost function is avoided, the purpose of simplifying the calculation amount is achieved, and the real-time performance of motor control is optimized.
Drawings
FIG. 1 is a graph of model training in accordance with the present invention;
FIG. 2 is a functional block diagram of the present invention;
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1 and fig. 2, the method for DTC prediction control based on the support vector machine of the present invention includes the following specific steps:
the method comprises the following steps: based on the parameters of the motor operation for which a cost function has been used for voltage vector selection, a suitable sampling frequency is determined from which suitable input and output samples for DTC predictive control based on the support vector machine are selected.
The motor for selecting the voltage vector by adopting the cost function has the cost function expression shown in (1).
Figure BDA0002117498510000041
Wherein
Figure BDA0002117498510000042
For the purpose of reference to the amplitude values,
Figure BDA0002117498510000043
for the purpose of reference to the torque,
Figure BDA0002117498510000044
is the actual amplitude, TeIs the actual torque.
When the number of the alternative basic voltage vectors is 6, 6 cost functions are obtained, and the basic voltage vector corresponding to the minimum cost function is selected from the 6 cost functions, namely the applied voltage vector. Recording all parameters of the motor in the running process, wherein the sampling period is ten times of the sampling period of the motor, and selecting the stator flux linkage error psi under all working conditionsrOutput torque error TrThe load angle delta and the torque angle alpha are used as input for training a prediction model, and an applied voltage vector is selected as output.
Step two: and D, normalizing the input and output samples determined in the step one.
In order to make the input and output in the same magnitude respectively, normalization processing needs to be performed on the input and output samples respectively, that is, x before normalization is reduced to a y value according to the following formula, and the expression is shown in formula (2).
Figure BDA0002117498510000045
Wherein x isminIs the minimum value in the sample, xmaxIs the maximum value in the sample; i.e. when normalizing the input samples, xminIs the minimum value, x, in the input samplemaxIs the maximum value in the input samples; when normalizing the output samples, xminIs the minimum value, x, in the output samplemaxIs the maximum value in the output samples.
Step three: and training the normalized sample by using a support vector machine to obtain a support vector machine-based prediction model of the output voltage vector for the DTC.
The normalized samples were trained using a support vector machine. According to the method idea of the support vector machine, a radial basis function shown in the formula (3) is selected as a kernel function of the support vector machine.
k(x,xi)＝exp(-γ‖x-xi‖2),γ>0 (3)
Where x is the input, xiTo support the vector, γ is a kernel parameter.
Randomly scrambling the input samples and the output samples selected in the step one to form a new sequence again, and in order to ensure the accuracy of an offline training support vector machine-based DTC prediction model, a sampling cross validation method includes dividing the samples selected in the step one into 3 groups, making each group a validation set respectively, using the rest 2 groups of subset data as a training set, and thus obtaining 3 sub-models, wherein each sub-model respectively corresponds to a group of penalty parameters c and kernel function parameters g, and the average of classification accuracy of the validation sets in the 3 sub-models is used as a performance index of a classifier. If the average of the classification accuracy is larger than or equal to 90%, selecting a group of parameters c and g with the highest classification accuracy of 3 sub-models as the parameters of the prediction model, and if the average of the classification accuracy is smaller than 90%, retraining. The classification accuracy of the submodels and the average calculation of the classification accuracy are shown in formula (4).
Figure BDA0002117498510000051
Wherein etaiIs the classification accuracy of the ith sub-model, ni1Selecting the number of correct voltage vectors, n, for the ith sub-modeli2Selecting the total number of voltage vectors, η, in the ith submodelaveIs the average of the classification accuracy.
Step four: and carrying out normalization processing on the current operation parameters of the motor.
By obtaining the stator flux linkage error psi at time krError of output torque TrAfter the load angle δ and the torque angle α, normalization processing is performed, as in equation (2).
Step five: and inputting the normalized parameters in the fourth step into a prediction model for the DTC based on the support vector machine. And predicting the basic voltage vector applied by the motor by using a trained prediction model for the DTC based on the support vector machine.
The processed data is input into a trained support vector machine, and the applied basic voltage vector is output through a prediction model for a DTC based on the support vector machine. This process replaces the use of a cost function to select the base voltage vector.
Step six: and controlling the switching state of the inverter according to a basic voltage vector output by a prediction model for the DTC based on the support vector machine, thereby realizing the operation of the motor.
According to the direct torque control principle, a basic voltage vector output based on a prediction model for a DTC of a support vector machine determines a switching state of an inverter, thereby controlling the operation of a motor.
Example 1
Firstly, collecting the operation parameters of the motor adopting the cost function to select the voltage vector, including the stator flux linkage error psirError of output torque TrLoad angle delta, torque angle alpha and selected voltage vector, adoptThe sampling frequency is ten times of the motor adopted frequency, and the sampled data is subjected to normalization processing, as shown in formula (2). And (3) performing off-line training by using the support vector machine with the formula (3) as a kernel function, obtaining a model with the highest accuracy by adopting a cross validation method, wherein the calculation method of the accuracy is shown as the formula (4), and using the model with the highest accuracy as a prediction model for the DTC based on the support vector machine.
Secondly, when the motor runs, the stator flux linkage error psi at a certain momentrError of output torque TrThe load angle delta and the torque angle alpha are used as input and are input into a prediction model through normalization processing, and a basic voltage vector selected by the motor at the moment is obtained. And determining the switching state of the inverter according to the basic voltage vector, thereby controlling the motor. The relationship between the selected basic voltage vector and the states of the inverter switches is shown in table one. For example, when the voltage vector is selected to be a 0 voltage vector (i.e., u) based on the support vector machine's prediction model for DTC1) When the inverter state is SA＝1，SB＝0，SC＝0。
Table-table of relation between basic voltage vector and state of inverter switch
Figure BDA0002117498510000061

Claims (5)

1. A method for DTC prediction control based on a support vector machine, characterized by: the method comprises the following steps:
the method comprises the following steps: selecting the existing motor operation parameters as input and output samples, wherein the existing motor operation parameters are motor operation parameters for voltage vector selection by adopting a cost function, and the cost function is shown as the following formula (1):
Figure FDA0002951145850000011
wherein
Figure FDA0002951145850000012
For the purpose of reference to the amplitude values,
Figure FDA0002951145850000013
for the purpose of reference to the torque,
Figure FDA0002951145850000014
is the actual amplitude, TeIs the actual torque;
step two: the expressions for normalizing the selected input and output samples are shown in the following formula (2):
Figure FDA0002951145850000015
wherein x isminIs the minimum value, x, of the input or output samplesmaxIs the maximum value in the input sample or the output sample;
step three: training the normalized sample by using a support vector machine to obtain a support vector machine-based prediction model of an output voltage vector for a DTC, and adopting a radial basis function as a kernel function of the support vector machine, wherein the kernel function is shown as a formula (3):
k(x,xi)＝exp(-γ||x-xi||2) γ > 0 (3) wherein x is the input, xiFor support vectors, γ is the kernel parameter;
step four: normalizing the current motor operation parameters by obtaining the stator flux linkage error psi at the moment krError of output torque TrAfter the load angle delta and the torque angle alpha are processed, normalization processing is carried out;
step five: normalizing the stator flux linkage error psi in the fourth steprError of output torque TrInputting the load angle delta and the torque angle alpha into a prediction model, and outputting an applied voltage vector through the prediction model;
step six: and controlling the switching state of the inverter according to the voltage vector output by the prediction model, and completing the DTC prediction control based on the support vector machine.
2. The method of claim 1, wherein the method comprises the following steps: in the first step, the number of the cost functions is equal to the number of the voltage vectors, and the voltage vector corresponding to the minimum cost function is selected as the applied voltage vector; selecting stator flux linkage error psi under all working conditionsrOutput torque error TrThe load angle delta and the torque angle alpha are used as input for training a prediction model, and an applied voltage vector is selected as output.
3. The method of claim 1, wherein the method comprises the following steps: in the third step, the input samples and the output samples selected in the first step are divided into 3 groups, the subset data of each group are respectively made into a primary verification set, the rest 2 groups of subset data are used as training sets, 3 submodels are obtained, each submodel respectively corresponds to one group of punishment parameters and nuclear parameters, and the average of the classification accuracy of the verification sets in the 3 submodels is used as the performance index of the classifier.
4. A method for DTC prediction control based on support vector machine according to claim 3, characterized in that: and if the average of the classification accuracy is greater than or equal to 90%, selecting a group of punishment parameters and kernel parameters with the highest classification accuracy as parameters of the prediction model, and if the average of the classification accuracy is less than 90%, retraining.
5. The method of claim 1, wherein the method comprises the following steps: the normalization processing mode in the fourth step is the same as that in the second step.
CN201910595422.0A 2019-07-03 2019-07-03 Method for DTC prediction control based on support vector machine Active CN110224649B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910595422.0A CN110224649B (en) 2019-07-03 2019-07-03 Method for DTC prediction control based on support vector machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910595422.0A CN110224649B (en) 2019-07-03 2019-07-03 Method for DTC prediction control based on support vector machine

Publications (2)

Publication Number Publication Date
CN110224649A CN110224649A (en) 2019-09-10
CN110224649B true CN110224649B (en) 2021-04-06

Family

ID=67815805

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910595422.0A Active CN110224649B (en) 2019-07-03 2019-07-03 Method for DTC prediction control based on support vector machine

Country Status (1)

Country Link
CN (1) CN110224649B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111555684B (en) * 2020-04-03 2021-10-29 浙江工业大学 Permanent magnet synchronous motor multi-step model prediction torque control method with variable switching points

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102075136B (en) * 2011-01-10 2013-04-17 江苏大学 Soft measurement method for magnetic flux linkage of bearingless permanent magnet synchronous motor
CN103501148A (en) * 2013-09-24 2014-01-08 江苏大学 Method for controlling operation of non-radial displacement sensor of bearingless permanent magnetic synchronous motor
CN103633903B (en) * 2013-12-05 2016-01-20 江苏大学 A kind of switch magnetoresistance motor rotor position self-sensing method
CN103746627B (en) * 2013-12-31 2014-11-26 华北电力大学（保定） Direct torque control method of PMSM (Permanent Magnet Synchronous Motor)
CN106059428B (en) * 2016-07-07 2019-07-12 东南大学 The model predictive control method of three-phase Four-switch converter driving permanent magnet synchronous motor
CN107425762B (en) * 2017-07-04 2019-10-22 大连海事大学 A kind of three-phase switch reluctance machine prediction moment controlling system and method
CN108900128B (en) * 2018-09-06 2021-09-28 吉林大学 Direct torque control method of permanent magnet synchronous motor based on model predictive control
CN109687789B (en) * 2019-01-02 2020-11-03 长安大学 Method for vector prediction direct torque control based on inverter basic voltage

Also Published As

Publication number Publication date
CN110224649A (en) 2019-09-10

Similar Documents

Publication Publication Date Title
Du et al. Research on active disturbance rejection control with parameter autotune mechanism for induction motors based on adaptive particle swarm optimization algorithm with dynamic inertia weight
US20130035773A1 (en) Method for Generating Trajectories for Motor Controlled Actuators
CN110705174B (en) Five-degree-of-freedom magnetic suspension motorized spindle rotor displacement self-detection system and method
CN109687789B (en) Method for vector prediction direct torque control based on inverter basic voltage
CN111064408B (en) Asynchronous motor model prediction torque control method without weight
CN110224649B (en) Method for DTC prediction control based on support vector machine
CN114915225A (en) Permanent magnet synchronous motor parameter identification method based on optimized BP neural network
Douiri et al. Genetic algorithms based fuzzy speed controllers for indirect field oriented control of induction motor drive
CN112803861A (en) Zero-vector-free algorithm for predictive control of three-vector model of permanent magnet synchronous motor
CN109245640A (en) A kind of asynchronous motor forecast Control Algorithm and device
CN110707973A (en) Method and system for rapidly searching efficiency optimal point of permanent magnet synchronous motor
CN112564557B (en) Control method, device and equipment of permanent magnet synchronous motor and storage medium
CN113283063B (en) Friction force identification method, module, device and computer readable medium
Chaouch et al. Optimized torque control via backstepping using genetic algorithm of induction motor
CN116404927A (en) PMSM current harmonic dynamic suppression system of hydrogen fuel cell automobile
CN115932594A (en) Multi-innovation least square online parameter identification method for power battery
CN115936303A (en) Transient voltage safety analysis method based on machine learning model
Bilal et al. A Comparative Study of Machine Learning Algorithms for Controlling Torque of Permanent Magnet Synchronous Motors through a Closed Loop System
Pant et al. Optimal tuning of pi speed controller using nature inspired heuristics
CN113158134A (en) Method and device for constructing non-invasive load identification model and storage medium
CN111082711A (en) Brushless direct current motor backstepping method control method and system
Tian et al. Weighting factors design in model predictive direct torque control based on cascaded neural network
CN113098351B (en) Surface-mounted permanent magnet synchronous motor model prediction torque control method
CN116305642B (en) Method and device for analyzing tolerance sensitivity of permanent magnet synchronous motor and computer readable storage medium
CN116404925A (en) Induction motor model prediction speed control method for improving super local model

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