CN112994538A - Fourier neural network based SRM torque ripple suppression control system and method - Google Patents

Fourier neural network based SRM torque ripple suppression control system and method Download PDF

Info

Publication number
CN112994538A
CN112994538A CN202110138493.5A CN202110138493A CN112994538A CN 112994538 A CN112994538 A CN 112994538A CN 202110138493 A CN202110138493 A CN 202110138493A CN 112994538 A CN112994538 A CN 112994538A
Authority
CN
China
Prior art keywords
torque
current
module
current moment
neural network
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
Application number
CN202110138493.5A
Other languages
Chinese (zh)
Other versions
CN112994538B (en
Inventor
党选举
原翰玫
潘登
李晓
伍锡如
张向文
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.)
Guilin University of Electronic Technology
Original Assignee
Guilin University of Electronic Technology
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 Guilin University of Electronic Technology filed Critical Guilin University of Electronic Technology
Priority to CN202110138493.5A priority Critical patent/CN112994538B/en
Publication of CN112994538A publication Critical patent/CN112994538A/en
Application granted granted Critical
Publication of CN112994538B publication Critical patent/CN112994538B/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
    • 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
    • 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
    • H02P6/06—Arrangements for speed regulation of a single motor wherein the motor speed is measured and compared with a given physical value so as to adjust the motor speed
    • 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
    • H02P6/34—Modelling or simulation for control purposes
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/64—Electric machine technologies in electromobility

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Control Of Electric Motors In General (AREA)
  • Control Of Motors That Do Not Use Commutators (AREA)

Abstract

The invention discloses a system and a method for controlling suppression of torque ripple based on a Fourier neural network SRM, wherein a Fourier neural network frequency spectrum detection module is set up by referring to the relation between torque and actual output torque, and frequency spectrums containing the content of each subharmonic in output torque signals of the Fourier neural network are quickly acquired on line; the dynamic characteristics of the torque are considered in a combined mode, a signal generator for compensating the reference torque is designed according to the torque harmonic information and the historical torque harmonic information acquired on line, the output of the signal generator compensates the reference torque output by the speed controller, the harmonic information corresponding to the output torque ripple caused by the reference torque which is not expected is eliminated, so that the more ideal reference torque is obtained, under the coordination of the torque distributor and the current inner loop control, the amplitude of each subharmonic of the output torque of the control system is greatly reduced, and the torque ripple is effectively inhibited.

Description

Fourier neural network based SRM torque ripple suppression control system and method
Technical Field
The invention relates to the technical field of switched reluctance motors, in particular to a system and a method for controlling SRM torque ripple suppression based on a Fourier neural network.
Background
In recent years, with the vigorous development of new energy automobiles, a high-performance Motor driving system becomes more important, and a novel Switched Reluctance Motor (SRM) has the advantages of no winding and magnetic materials of a rotor, high torque density, high temperature resistance, easiness in cooling, wide speed regulation range, high safety and the like, and is widely applied to the fields of mining production, aerospace, mechanical transmission and the like. However, the SRM has a large torque ripple caused by the nonlinear electromagnetic characteristics generated by the simple and unique double salient pole structure when the SRM operates at a low speed, and the popularization and application of the SRM in the new energy automobile industry are limited, while the traditional motor control mode is difficult to be applied to the SRM drive system, so that the research on the control system for inhibiting the SRM torque ripple has important significance.
In the aspect of researching an SRM control system from the perspective of signal analysis, research has been carried out to propose a method for suppressing torque ripple of a switched reluctance motor by injecting multiple harmonic currents into a conventional reference current, and experiments prove that the method is feasible. Studies have been made to provide a relationship between phase current harmonics and vibration, in which phase current waveforms are quasi-sinusoidal by optimizing the switching angle to suppress harmonics of the control current and thereby reduce SRM vibration. However, the above methods rely on prior knowledge, and it is difficult to achieve an ideal effect of suppressing torque ripple when parameters of a magnetic saturation region or a motor drift.
Disclosure of Invention
The invention aims to solve the problem that the conventional method for inhibiting the torque ripple of the switched reluctance motor cannot achieve an ideal effect, and provides a system and a method for controlling the SRM torque ripple inhibition based on a Fourier neural network.
In order to solve the problems, the invention is realized by the following technical scheme:
the SRM torque ripple suppression control method based on the Fourier neural network comprises the following steps:
step 1, training and learning the collected actual torque of the switched reluctance motor at the current moment k and the historical moments 0,1, … and k-1 based on a Fourier neural network to obtain the output torque of the current moment k and the historical moments 0,1, … and k-1, and extracting harmonic coefficient vectors of the output torque of the current moment k and the historical moments 0,1, … and k-1 by detecting the subharmonic information of the output torque of the current moment k and the historical moments 0,1, … and k-1 in real time on line;
step 2, calculating a reference compensation torque of the current moment k by using the harmonic coefficient vector of the output torque of the current moment k and the historical moments 0,1, …, k-1 obtained in the step 1;
step 3, obtaining the actual rotating speed of the current moment k after deriving the collected actual rotor position angle of the current moment k of the switched reluctance motor, and obtaining the reference torque of the current moment k after carrying out PI (proportional integral) speed regulation on the rotating speed deviation between the set rotating speed and the actual rotating speed of the current moment k;
step 4, correcting the reference torque of the current moment k obtained in the step 3 by using the reference compensation torque of the current moment k obtained in the step 2 and the collected actual torque of the switched reluctance motor at the current moment k to obtain the final reference torque of the current moment k;
step 5, dividing the final reference torque of the current moment k obtained in the step 4 into three phases according to the collected actual rotor position angle of the current moment k of the switched reluctance motor, and performing torque-to-current conversion on the three phases to obtain control currents of all phases of the current moment k;
step 6, tracking the actual current of each phase at the current moment k of the switched reluctance motor, which is acquired in the step 5, with the control current of each phase at the current moment k of the corresponding phase, so as to obtain the current tracked by each phase at the current moment k;
and 7, performing power conversion on the current obtained by the step 6 after tracking each phase at the current moment k, and controlling three phase lines of the switched reluctance motor to realize the control of the rotating speed and the torque of the switched reluctance motor at the current moment k.
In step 2, the reference compensation torque T at the current time kcComprises the following steps:
Figure BDA0002927732730000021
in the formula, alphaiIs the harmonic coefficient vector at the ith time, H is the excitation function vector, HTFor transposing the excitation function vector H, H ═ cos (ω)0kTs),…,cos(Nω0kTs),sin(ω0kTs),…,sin(Nω0kTs)]，ω0At fundamental angular frequency, TsFor a sampling period, N represents the harmonic order, and k is the current time.
In step 4, the final reference torque T at the current time k isref' is:
Tref′＝Tref-Tc-Te
in the formula, TrefReference torque, T, at the present moment kcFor the reference compensation torque at the present time k, TeIs the actual torque at the current time k.
The system for realizing the method and restraining the SRM torque ripple comprises a position detection module, a torque detection module, a phase current detection module, a derivation module, a rotating speed subtracter, a PI regulation module, a torque distribution module, a torque-current conversion module, a current hysteresis control module, a power conversion module, a first torque subtracter, a second torque subtracter, a Fourier neural network frequency spectrum detection module and a signal generator module.
The input ends of the position detection module, the torque detection module and the phase current detection module are connected with the switched reluctance motor. The output end of the position detection module is connected with one input end of the rotating speed subtracter through the derivation module, the other input end of the rotating speed subtracter inputs a set rotating speed, the output end of the rotating speed subtracter is connected with the input end of the PI adjusting module, and the output end of the PI adjusting module is connected with one input end of the first torque subtracter. The output end of the torque detection module is connected with the input end of the Fourier neural network spectrum detection module, the output end of the Fourier neural network spectrum detection module is connected with the input end of the signal generator module, and the output end of the signal generator module is connected with the other output end of the first torque subtracter. The output end of the first torque subtracter is connected with one input end of the second torque subtracter, and the output end of the torque detection module is connected with the other input end of the second torque subtracter. The output end of the second torque subtracter is connected with the input end of the torque distribution module, the output end of the torque distribution module is connected with the input end of the torque-current conversion module, and the output end of the torque-current conversion module is connected with one input end of the current hysteresis control module. The output end of the phase current detection module is connected with the other input end of the current hysteresis control module. The output end of the current hysteresis control module is connected with the three-phase line of the switched reluctance motor through the power conversion module.
According to the method, the relation between the reference torque and the actual output torque is utilized, a Fourier Neural Network (FNN) is built to detect each subharmonic information of the output torque on line in real time, relevant harmonic items influencing the output torque are extracted, the reference torque is subjected to feedforward compensation, corresponding harmonic information which is not expected in the reference torque and causes large torque pulsation is eliminated, a more ideal reference torque is obtained, each subharmonic content of the output torque is indirectly counteracted from the angle of the reference torque, and the purpose of inhibiting the torque pulsation is achieved.
Compared with the prior art, the invention has the following characteristics:
1. according to the periodic characteristics of the torque ripple, the Fourier neural network can be used for quickly acquiring the frequency spectrum containing the content of each subharmonic in the output torque signal on line.
2. And in consideration of the dynamic characteristics of the torque, designing a reference torque compensation signal generator according to the torque harmonic information and the historical torque harmonic information acquired online, wherein the output of the reference torque compensation signal generator is used for compensating the output reference torque of the speed controller, and the corresponding harmonic information which is undesirable and causes output torque pulsation in the reference torque is eliminated online.
3. The torque pulsation is restrained by online self-adaptive adjustment of a torque set value.
4. The system can form an embedded system, extracts torque ripple information based on current fluctuation, indirectly inhibits torque ripple through a torque set value compensation mode, has small calculation amount of a technical method, and is convenient for on-line control of the switched reluctance motor.
Drawings
Fig. 1 is a block diagram of a conventional TSF control.
FIG. 2 is a block diagram of a dual closed-loop speed and torque control.
Fig. 3 is a diagram of a torque fourier neural network architecture.
FIG. 4 is a structure diagram of a rotating speed and torque double closed-loop control system of torque Fourier neural network feedforward.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to specific examples.
1) Control structure of switched reluctance motor
1.1TSF basic control architecture
In the basic control structure of the TSF, an outer ring outputs a reference torque for a pi (proportional integral) regulator for closed-loop control of the rotation speed, the reference torque outputs a reference current through a torque distribution function and torque-current conversion, and a motor tracks the reference current to realize the control of the rotation speed, and a control block diagram is shown in fig. 1.
Under an ideal linear model, the torque T and the current i of each phasejThe relationship is as follows:
Figure BDA0002927732730000031
where i is the total current, θ is the motor rotation angle, and n is the number of phases. L isjAnd a j-th phase inductance.
In order to avoid generating negative torque, the switched reluctance motor generally completes the turn-on and turn-off processes in the inductance rise region, and the calculation formula of the torque-current converter is obtained by the formula (1):
Figure BDA0002927732730000041
torque T of j phase in the formulajAs a reference torque TrefReference torque, L, through the j-th phase of the torque distribution functionjIs the j phase inductance of the switched reluctance motor.
The control structure is simple, a torque detection module is not needed, only the rotating speed outer ring is needed, the stability of the torque is difficult to guarantee, and the strong nonlinear characteristic of the actual phase inductance is not considered in the formula (2), so that the problem of torque pulsation is prominent.
1.2 speed and torque double closed loop control
On the basis of direct torque-current control, torque negative feedback is added to form a double closed-loop control structure of outer loop rotating speed and inner loop torque. The output of the PI (proportional integral) proportional integral regulator minus the torque feedback value is the reference torque. Under the torque-current conversion obtained by the linear model, torque negative feedback is introduced, at the moment, the reference torque contains information of output torque feedback, the control current obtained through the torque-current conversion is compensated, the torque ripple is further inhibited, and the control structure is shown in fig. 2.
2) Torque Fourier neural network design
The SRM torque waveform is periodic and satisfies the dirichlet convergence theorem and thus can be represented by a fourier series:
Figure BDA0002927732730000042
wherein ω is0Fundamental angular frequency, fundamental angular frequency ω of torque signal for three-phase 6/4-pole switched reluctance motor used0Calculated from the following formula:
Figure BDA0002927732730000043
wherein n is the motor rotation speed (rpm), (3) wherein a0Is a constant term ofnAnd bnIs cosine term and positiveThe chord coefficient can be written as a cosine term by combining the two terms, and the expression is as follows:
Figure BDA0002927732730000044
the coefficient a is knownnAnd bnReflects the amplitude and phase information of a certain harmonic wave, and adopts the writing form of the formula (3) for the convenience of explanation. The actual torque signal is a finite frequency band periodic signal, and the discretized Fourier series is expressed as follows:
Figure BDA0002927732730000045
where k is 0,1, … N, N ω0For the highest angular frequency, T, after discretization of the torque signalsIs the sampling period. f. ofsFor the sampling frequency, it is selected to satisfy the shannon sampling theorem,
Figure BDA0002927732730000051
according to the Fourier series expansion (6) and the actual torque fT(k) For training the samples, fnn(k) For the output of the fourier neural network, the constant term, the cosine terms and the sine term in equation (6) are neuron activation functions, the coefficients thereof are neural network weights, and the fourier neural network is constructed as shown in fig. 3.
Continuously correcting the weight coefficient a by using the deviation between the actual torque signal and the output signal of the Fourier neural network0,a1,…,an,b1,…,bnThe stable amplitude is the amplitude of each harmonic of the torque signal, as can be seen from FIG. 3, if the highest angular frequency needs to be detected to be N ω0Then 2N +1 neurons are needed.
Defining an error function:
e(k)＝fT(k)-fnn(k) (8)
performance index function:
Figure BDA0002927732730000052
training a neural network weight by using an improved algorithm of a gradient descent method:
Figure BDA0002927732730000053
Figure BDA0002927732730000054
Figure BDA0002927732730000055
wherein eta is learning rate, eta is more than 0 and less than 1, and alpha is more than 0 and less than 1 and is momentum factor. k is the current time of the day and,
Figure BDA0002927732730000056
k next time, current time, previous time to current time, and a previous time to previous time, respectivelynA value;
Figure BDA0002927732730000057
respectively the next moment of the current k, the current moment, the previous moment of the current moment and the previous moment a of the previous moment0The value is obtained.
3) Harmonic suppression strategy for torque signals
Ideally, the ripple-free torque waveform signal is a straight line, and only a constant term and no cosine term are obtained by performing fourier series decomposition on the straight line. The torque ripple exists in the actual torque waveform, and the smaller the amplitude of the cosine term of the fourier series is, the closer the torque waveform is to a straight line, the smaller the torque ripple is. In the SRM control system, a torque-current conversion module converts a reference torque into a reference current, if the output current of a torque-current conversion model is more accurate, the waveform of the reference torque is more similar to that of the output torque, the reference torque and the output torque have more consistent amplitude-frequency characteristics, the harmonic corresponding to the output torque can be changed by changing the harmonic of the reference torque, and the amplitude of the harmonic of the output torque waveform is reduced in such a way, so that the aim of inhibiting torque pulsation is fulfilled. The method comprises the steps of detecting partial harmonic content of output torque by using a torque Fourier neural network, eliminating components of torque pulsation in reference torque output by a speed controller in a feed-forward control mode, changing frequency spectrum of a reference torque waveform, indirectly reducing amplitude of each subharmonic of the output torque, and finally achieving torque pulsation suppression.
Based on a basic SRM control structure, the invention relates to a switched reluctance motor torque ripple suppression control system based on a Fourier neural network, a corresponding control block diagram is shown in FIG. 4, and the control strategy is arranged above a dotted line.
The torque fourier neural network expression, according to equation (6), is trained on-line on the actual torque signal,
Figure BDA0002927732730000061
in the formula TfnnFor the output of the Fourier neural network, after the training is stabilized, the coefficient a of the cosine term of equation (13)1,…,a30,b1,…,b30For the coefficients of the first N-30 th harmonic of the output torque signal, the coefficient vector is written:
α＝[a1,…a30,b1,…b30] (14)
order to
H＝[cos(ω0kTs),…,cos(30ω0kTs),sin(ω0kTs),…,sin(30ω0kTs)] (15)
Formula (13) can be written as
Tfnn＝a0+αHΤ (16)
Will t0The coefficient vector obtained at the moment is stored in the memory and recorded as alpha0The signal generator outputs a reference torque compensation signal at the momentNumber:
Figure BDA0002927732730000062
as shown in FIG. 4, Tc 0After the reference torque signal is compensated, because the torque-current conversion under the linear model of the switched reluctance motor is not accurate enough, the reference torque is still not optimal at the moment, the output torque still has torque pulsation, the Fourier neural network training is carried out again by taking the output torque signal as a training sample, and t1The coefficient vector alpha of the torque signal is obtained at a time1At t0Time of day compensation signal Tc 0The reference torque compensation signal is updated on the basis, the processing of which is aimed at taking into account the dynamic characteristics of the torque, i.e. the characteristics related to the historical state, to a sufficient extent.
Figure BDA0002927732730000063
Signal Tc 1After the feed forward compensation, the amplitude of the harmonics in the reference torque causing the torque ripple is further reduced. By analogy, at the current tkAt the moment, a torque signal coefficient vector alpha is obtainedkTo further reduce torque ripple, at current tkA time t preceding the timek-1Compensation signal obtained at a time
Figure BDA0002927732730000064
And calculating a reference torque compensation signal at the moment on the basis of:
Figure BDA0002927732730000065
taking into account the compensation signal TcNot only the torque harmonic coefficient vector at the current time but also the harmonic coefficient vector at the historical time, let alpha0,α1,…,αk-1Represents t0,t1,…tk-1Each calendarCoefficient vector of fourier neural network output at history time:
Figure BDA0002927732730000066
the reference torque compensation signal T is summarized by the equations (18) and (19)CGeneral expression of (a):
Figure BDA0002927732730000067
harmonic coefficient vector alpha at the current moment along with reduction of torque pulsation in control processkGradually becomes smaller, and T is known from the formula (21)CAnd tends to be stable. T isCTo the reference torque T in a feed-forward mannerrefAnd (4) compensating, and realizing the suppression control of the SRM output torque ripple by a reference torque compensation method.
Based on the above analysis, the method for controlling the suppression of the torque ripple based on the SRM comprises the following steps:
step 1, training and learning the collected actual torque of the switched reluctance motor at the current moment k and the historical moments 0,1, … and k-1 based on a Fourier neural network to obtain the output torque of the current moment k and the historical moments 0,1, … and k-1, and extracting harmonic coefficient vectors of the output torque of the current moment k and the historical moments 0,1, … and k-1 by detecting the subharmonic information of the output torque of the current moment k and the historical moments 0,1, … and k-1 in real time on line.
Step 2, calculating a reference compensation torque of the current moment k by using the harmonic coefficient vector of the output torque of the current moment k and the historical moments 0,1, …, k-1 obtained in the step 1; at this time, the reference compensation torque T at the present time kcComprises the following steps:
Figure BDA0002927732730000071
in the formula, alphaiAs harmonic coefficients at the ith timeVector, H is the vector of the excitation function, H ═ cos (ω)0kTs),…,cos(Nω0kTs),sin(ω0kTs),…,sin(Nω0kTs)]，ω0At fundamental angular frequency, TsFor the sampling period, N represents the harmonic order, HTIs a transpose of the excitation function vector H.
And 3, deriving the acquired actual rotor position angle of the switched reluctance motor at the current moment k to obtain the actual rotating speed at the current moment k, and carrying out PI speed regulation on the rotating speed deviation between the set rotating speed and the actual rotating speed at the current moment k to obtain the reference torque at the current moment k.
Step 4, correcting the reference torque of the current moment k obtained in the step 3 by using the reference compensation torque of the current moment k obtained in the step 2 and the collected actual torque of the switched reluctance motor at the current moment k to obtain the final reference torque of the current moment k; at this time, the final reference torque T at the current time kref' is:
Tref′＝Tref-Tc-Te
in the formula, TrefReference torque, T, at the present moment kcFor the reference compensation torque at the present time k, TeIs the actual torque at the current time k.
And 5, dividing the final reference torque of the current moment k obtained in the step 4 into three phases according to the collected actual rotor position angle of the current moment k of the switched reluctance motor, and performing torque-to-current conversion on the three phases to obtain the control current of each phase of the current moment k.
And 6, tracking the acquired actual current of each phase at the current moment k of the switched reluctance motor to the control current of each phase at the current moment k of the corresponding phase obtained in the step 5 to obtain the tracked current of each phase at the current moment k.
And 7, performing power conversion on the current obtained by the step 6 after tracking each phase at the current moment k, and controlling three phase lines of the switched reluctance motor to realize the control of the rotating speed and the torque of the switched reluctance motor at the current moment k.
Referring to fig. 4, the fourier neural network SRM-based torque ripple suppression control system for implementing the above method includes a position detection module, a torque detection module, a phase current detection module, a derivation module, a rotational speed subtractor, a PI adjustment module, a torque distribution module, a torque-current conversion module, a current hysteresis control module, a power conversion module, a first torque subtractor, a second torque subtractor, a fourier neural network spectrum detection module, and a signal generator module. The Fourier neural network spectrum detection module comprises a Fourier neural network module and a spectrum detection module.
The input ends of the position detection module, the torque detection module and the phase current detection module are connected with the switched reluctance motor. The output end of the position detection module is connected with one input end of the rotating speed subtracter through the derivation module, the other input end of the rotating speed subtracter inputs a set rotating speed, the output end of the rotating speed subtracter is connected with the input end of the PI adjusting module, and the output end of the PI adjusting module is connected with one input end of the first torque subtracter. The output end of the torque detection module is connected with the input end of the Fourier neural network spectrum detection module, the output end of the Fourier neural network spectrum detection module is connected with the input end of the signal generator module, and the output end of the signal generator module is connected with the other output end of the first torque subtracter. The output end of the first torque subtracter is connected with one input end of the second torque subtracter, and the output end of the torque detection module is connected with the other input end of the second torque subtracter. The output end of the second torque subtracter is connected with the input end of the torque distribution module, the output end of the torque distribution module is connected with the input end of the torque-current conversion module, and the output end of the torque-current conversion module is connected with one input end of the current hysteresis control module. The output end of the phase current detection module is connected with the other input end of the current hysteresis control module. The output end of the current hysteresis control module is connected with the three-phase line of the switched reluctance motor through the power conversion module.
From the angle of signal analysis, the periodic characteristics of the torque ripple are found, a Fourier neural network is built, and the frequency spectrum containing the content of each subharmonic in the output torque signal is quickly acquired on line; the dynamic characteristics of the torque are considered in a combined mode, a reference torque compensation signal generator is designed according to the torque harmonic information and the historical torque harmonic information which are acquired on line, the output of the reference torque compensation signal generator compensates the reference torque output by the speed controller, the harmonic information corresponding to the output torque ripple which is caused in the reference torque and is not expected is removed, so that a more ideal reference torque is obtained, under the coordination of a torque distributor and current inner loop control, the amplitude of each subharmonic of the output torque of a control system is greatly reduced, and the torque ripple is effectively inhibited. The system can form an embedded system, extracts torque ripple information based on torque fluctuation, indirectly inhibits torque ripple through a feedforward compensation mode of a torque reference value, has small calculation amount of a technical method, and is easy to control the switched reluctance motor on line.
It should be noted that, although the above-mentioned embodiments of the present invention are illustrative, the present invention is not limited thereto, and thus the present invention is not limited to the above-mentioned embodiments. Other embodiments, which can be made by those skilled in the art in light of the teachings of the present invention, are considered to be within the scope of the present invention without departing from its principles.

Claims (4)

1. The SRM torque ripple suppression control method based on the Fourier neural network is characterized by comprising the following steps of:
step 1, training and learning the collected actual torque of the switched reluctance motor at the current moment k and the historical moments 0,1, … and k-1 based on a Fourier neural network to obtain the output torque of the current moment k and the historical moments 0,1, … and k-1, and extracting harmonic coefficient vectors of the output torque of the current moment k and the historical moments 0,1, … and k-1 by detecting the subharmonic information of the output torque of the current moment k and the historical moments 0,1, … and k-1 in real time on line;
step 2, calculating a reference compensation torque of the current moment k by using the harmonic coefficient vector of the output torque of the current moment k and the historical moments 0,1, …, k-1 obtained in the step 1;
step 3, deriving the acquired actual rotor position angle of the switched reluctance motor at the current moment k to obtain the actual rotating speed of the current moment k, and carrying out PI speed regulation on the rotating speed deviation between the set rotating speed and the actual rotating speed of the current moment k to obtain the reference torque of the current moment k;
step 4, correcting the reference torque of the current moment k obtained in the step 3 by using the reference compensation torque of the current moment k obtained in the step 2 and the collected actual torque of the switched reluctance motor at the current moment k to obtain the final reference torque of the current moment k;
step 5, dividing the final reference torque of the current moment k obtained in the step 4 into three phases according to the collected actual rotor position angle of the current moment k of the switched reluctance motor, and performing torque-to-current conversion on the three phases to obtain control currents of all phases of the current moment k;
step 6, tracking the actual current of each phase at the current moment k of the switched reluctance motor, which is acquired in the step 5, with the control current of each phase at the current moment k of the corresponding phase, so as to obtain the current tracked by each phase at the current moment k;
and 7, performing power conversion on the current obtained by the step 6 after tracking each phase at the current moment k, and controlling three phase lines of the switched reluctance motor to realize the control of the rotating speed and the torque of the switched reluctance motor at the current moment k.
2. The SRM torque ripple suppression control method based on the Fourier neural network as claimed in claim 1, wherein in the step 2, the reference compensation torque T at the current time kcComprises the following steps:
Figure FDA0002927732720000011
in the formula, alphaiIs the harmonic coefficient vector at the ith time, H is the excitation function vector, HTFor transposing the excitation function vector H, H ═ cos (ω)0kTs),…,cos(Nω0kTs),sin(ω0kTs),…,sin(Nω0kTs)]，ω0At fundamental angular frequency, TsFor the sampling period, N represents the harmonicThe wave number k is the current time.
3. The SRM torque ripple suppression control method based on the Fourier neural network as claimed in claim 1, wherein in the step 4, the final reference torque T at the current time k isref' is:
Tref′＝Tref-Tc-Te
in the formula, TrefReference torque, T, at the present moment kcFor the reference compensation torque at the present time k, TeIs the actual torque at the current time k.
4. The SRM torque ripple suppression control system based on the Fourier neural network for realizing the method of claim 1 comprises a position detection module, a torque detection module, a phase current detection module, a derivation module, a rotating speed subtracter, a PI regulation module, a torque distribution module, a torque-current conversion module, a current hysteresis control module and a power conversion module; the method is characterized in that: the system also further comprises a first torque subtracter, a second torque subtracter, a Fourier neural network spectrum detection module and a signal generator module;
the input ends of the position detection module, the torque detection module and the phase current detection module are connected with the switched reluctance motor;
the output end of the position detection module is connected with one input end of a rotating speed subtracter through a derivation module, the other input end of the rotating speed subtracter inputs a set rotating speed, the output end of the rotating speed subtracter is connected with the input end of a PI (proportion integration) regulation module, and the output end of the PI regulation module is connected with one input end of a first torque subtracter;
the output end of the torque detection module is connected with the input end of the Fourier neural network spectrum detection module, the output end of the Fourier neural network spectrum detection module is connected with the input end of the signal generator module, and the output end of the signal generator module is connected with the other output end of the first torque subtracter;
the output end of the first torque subtracter is connected with one input end of the second torque subtracter, and the output end of the torque detection module is connected with the other input end of the second torque subtracter;
the output end of the second torque subtracter is connected with the input end of the torque distribution module, the output end of the torque distribution module is connected with the input end of the torque-current conversion module, and the output end of the torque-current conversion module is connected with one input end of the current hysteresis control module;
the output end of the phase current detection module is connected with the other input end of the current hysteresis control module;
the output end of the current hysteresis control module is connected with the three-phase line of the switched reluctance motor through the power conversion module.
CN202110138493.5A 2021-02-01 2021-02-01 Fourier neural network based SRM torque ripple suppression control system and method Active CN112994538B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110138493.5A CN112994538B (en) 2021-02-01 2021-02-01 Fourier neural network based SRM torque ripple suppression control system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110138493.5A CN112994538B (en) 2021-02-01 2021-02-01 Fourier neural network based SRM torque ripple suppression control system and method

Publications (2)

Publication Number Publication Date
CN112994538A true CN112994538A (en) 2021-06-18
CN112994538B CN112994538B (en) 2022-09-13

Family

ID=76346036

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110138493.5A Active CN112994538B (en) 2021-02-01 2021-02-01 Fourier neural network based SRM torque ripple suppression control system and method

Country Status (1)

Country Link
CN (1) CN112994538B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114679103A (en) * 2022-05-30 2022-06-28 广东工业大学 Sliding mode current compensation control system and method for switched reluctance motor
CN115225009A (en) * 2022-08-05 2022-10-21 苏州金钥匙自动化设备有限公司 Control system for reducing torque pulsation of switched reluctance motor
CN116961494A (en) * 2023-09-19 2023-10-27 江西五十铃汽车有限公司 Driving motor active anti-shake control method and system

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020041168A1 (en) * 2000-08-18 2002-04-11 Switched Reluctance Drives Limited Apparatus and method for controlling an electric machine
US6922036B1 (en) * 2000-11-30 2005-07-26 The Texas A&M University System Method and apparatus for reducing noise and vibration in switched reluctance motor drives
JP2010220475A (en) * 2010-07-06 2010-09-30 Daikin Ind Ltd Switched reluctance motor control method and apparatus therefor
CN102088265A (en) * 2011-03-08 2011-06-08 东南大学 Method for restraining torque ripple of permanent magnet motor based on direct torque control
CN102710214A (en) * 2012-06-11 2012-10-03 桂林电子科技大学 Nonlinear interference control method and control system for permanent magnet linear synchronous motor
CN104506092A (en) * 2014-12-11 2015-04-08 东南大学 Switched reluctance motor current hysteresis control method based on inductance Fourier decomposition
CN106357192A (en) * 2016-09-05 2017-01-25 桂林电子科技大学 Method and system for lowering torque pulsation of switched reluctance motor by current adaptive control
CN107241033A (en) * 2017-08-01 2017-10-10 桂林电子科技大学 Switched reluctance machines method for suppressing torque ripple and system based on Current Position
CN109327178A (en) * 2018-10-29 2019-02-12 江苏大学 A kind of switched reluctance machines Multi-step predictive control device building method
CN109742999A (en) * 2019-01-17 2019-05-10 桂林电子科技大学 A kind of the SRM method for controlling torque and system of dynamic neural network adaptive inversion
CN110022109A (en) * 2019-04-17 2019-07-16 桂林电子科技大学 Torque-current neural network model SRM torque pulsation control method and system
CN110572108A (en) * 2019-09-12 2019-12-13 桂林电子科技大学 Method and system for nonlinear compensation and control of inductance model of switched reluctance motor
CN111293952A (en) * 2020-02-27 2020-06-16 桂林电子科技大学 Switched reluctance motor torque control system and method based on inductance rate model
CN112117947A (en) * 2020-09-30 2020-12-22 桂林电子科技大学 SRM torque ripple suppression control system and method based on current injection method

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020041168A1 (en) * 2000-08-18 2002-04-11 Switched Reluctance Drives Limited Apparatus and method for controlling an electric machine
US6922036B1 (en) * 2000-11-30 2005-07-26 The Texas A&M University System Method and apparatus for reducing noise and vibration in switched reluctance motor drives
JP2010220475A (en) * 2010-07-06 2010-09-30 Daikin Ind Ltd Switched reluctance motor control method and apparatus therefor
CN102088265A (en) * 2011-03-08 2011-06-08 东南大学 Method for restraining torque ripple of permanent magnet motor based on direct torque control
CN102710214A (en) * 2012-06-11 2012-10-03 桂林电子科技大学 Nonlinear interference control method and control system for permanent magnet linear synchronous motor
CN104506092A (en) * 2014-12-11 2015-04-08 东南大学 Switched reluctance motor current hysteresis control method based on inductance Fourier decomposition
CN106357192A (en) * 2016-09-05 2017-01-25 桂林电子科技大学 Method and system for lowering torque pulsation of switched reluctance motor by current adaptive control
CN107241033A (en) * 2017-08-01 2017-10-10 桂林电子科技大学 Switched reluctance machines method for suppressing torque ripple and system based on Current Position
CN109327178A (en) * 2018-10-29 2019-02-12 江苏大学 A kind of switched reluctance machines Multi-step predictive control device building method
CN109742999A (en) * 2019-01-17 2019-05-10 桂林电子科技大学 A kind of the SRM method for controlling torque and system of dynamic neural network adaptive inversion
CN110022109A (en) * 2019-04-17 2019-07-16 桂林电子科技大学 Torque-current neural network model SRM torque pulsation control method and system
CN110572108A (en) * 2019-09-12 2019-12-13 桂林电子科技大学 Method and system for nonlinear compensation and control of inductance model of switched reluctance motor
CN111293952A (en) * 2020-02-27 2020-06-16 桂林电子科技大学 Switched reluctance motor torque control system and method based on inductance rate model
CN112117947A (en) * 2020-09-30 2020-12-22 桂林电子科技大学 SRM torque ripple suppression control system and method based on current injection method

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114679103A (en) * 2022-05-30 2022-06-28 广东工业大学 Sliding mode current compensation control system and method for switched reluctance motor
CN114679103B (en) * 2022-05-30 2022-12-06 广东工业大学 Sliding mode current compensation control system and method for switched reluctance motor
CN115225009A (en) * 2022-08-05 2022-10-21 苏州金钥匙自动化设备有限公司 Control system for reducing torque pulsation of switched reluctance motor
CN115225009B (en) * 2022-08-05 2023-09-29 苏州金钥匙自动化设备有限公司 Control system for reducing torque pulsation of switch reluctance motor
CN116961494A (en) * 2023-09-19 2023-10-27 江西五十铃汽车有限公司 Driving motor active anti-shake control method and system
CN116961494B (en) * 2023-09-19 2024-01-26 江西五十铃汽车有限公司 Driving motor active anti-shake control method and system

Also Published As

Publication number Publication date
CN112994538B (en) 2022-09-13

Similar Documents

Publication Publication Date Title
CN112994538B (en) Fourier neural network based SRM torque ripple suppression control system and method
Xu et al. Improved non-singular fast terminal sliding mode control with disturbance observer for PMSM drives
Liu et al. Disturbance-observer-based complementary sliding-mode speed control for PMSM drives: A super-twisting sliding-mode observer-based approach
CN110572108B (en) Method and system for nonlinear compensation and control of inductance model of switched reluctance motor
WO2022252289A1 (en) Mtpa control method using d-q axis inductance parameter identification of fuzzy-logical controlled permanent-magnet synchronous electric motor
Yang et al. Fast integral terminal sliding mode control with a novel disturbance observer based on iterative learning for speed control of PMSM
CN112117947B (en) SRM torque ripple suppression control system and method based on current injection method
Xia et al. Speed adaptive sliding mode control with an extended state observer for permanent magnet synchronous motor
US11522480B2 (en) SPMSM sensorless composite control method with dual sliding-mode observers
CN113193809A (en) Permanent magnet synchronous motor control method for improving second-order linear active disturbance rejection
Laghridat et al. A novel adaptive active disturbance rejection control strategy to improve the stability and robustness for a wind turbine using a doubly fed induction generator
Wei et al. Autoregressive moving average model-free predictive current control for PMSM drives
CN113346810B (en) Speed and current double closed-loop fuzzy control PMSM sensorless control method
Barambones et al. A robust position control for induction motors using a load torque observer
Yu et al. Energy-shaping and L2 gain disturbance attenuation control of induction motor
Yu et al. Model‐free predictive control of non‐isolated two‐stage AC–DC–DC converter based on linear extended state observer
Liu et al. Model predictive control of permanent magnet synchronous motor based on parameter identification and dead time compensation
Reghioui et al. Enhancement of space vector modulation based-direct torque control using fuzzy PI controller for doubly star induction motor
Li et al. Research on PMSM model predictive control for ship electric propulsion
Yao et al. Robust model predictive current control for six-phase PMSMS with virtual voltage vectors
Wang et al. Continuous-Control-Set Model-Free Predictive Control Using Time-Series Subspace for PMSM Drives
Mehrotra et al. Speed estimation of induction motor using artificial neural networks
Wei et al. Research on BLDC Control System Based on Model Prediction and Integral Separation PID Algorithms
Zhang et al. Dynamic performance optimization of pmsm based on the improved exponential reaching law and proportional resonance strategy
Simoes et al. Permanent magnet synchronous motor current harmonic compensation method using model predictive control

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