1994-20 10 China academic journal electronic publishing house. Copyright ki.net.
Automation Instrument Volume 2 3 1 Phase 2 20 10
Supported by Shanghai Key Discipline Construction Fund (No.:B504).
Date of receipt of revised draft: August 26th, 2009.
The first author Xiong Xiang, male, born in 1984, is now a control science and control engineer in East China University of Science and Technology.
Master degree in engineering; Mainly engaged in advanced control and adaptive control research.
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS 2
Xiong bingjun
(School of Information Science and Engineering, East China University of Science and Technology, Shanghai 200237)
According to the basic principle and method of vector control, the torque, magnetic field and magnetic force control system is constructed by using Matlab /Simulink module in the rotating coordinate system based on rotor magnetic field orientation.
Simulation model of vector control system of chain closed-loop AC asynchronous motor. On this basis, the model reference adaptive method is used to estimate the speed of the speed sensorless vector control system.
In order to solve the problem that the reference model in the conventional speed identifier is easily affected by the initial value of integration and drift, the traditional MRAS method is improved and simulated.
Really. The simulation results show that the design is feasible and the calculated speed can track the measured speed well.
Keywords: Matlab /Simulink MRAS Vector Control Variable Frequency Speed Regulation System Neural Network Sensorless
China Library Classification Number: TM343 Document Identification Number: A.
Abstract: According to the basic principle and method of vector control, the simulation model of vector control is established by using Matlab /Simulink module.
Based on the rotor flux oriented rotating coordinate system, a control system for providing torque and flux for AC asynchronous motor is established.
On this basis, the speed estimation of the vector control system with the No.2 speed sensor is studied by using the model reference adaptive method. In ...
In addition, in order to solve the problem that the reference model in the conventional speed recognizer is easily affected by the initial value of integration and drift,
The traditional MRAS is improved and modeled and simulated. The simulation results verify the feasibility of the design.
The calculated rotational speed can track the measured rotational speed well.
Keywords: Matlab/Simulink model reference adaptive system vector control variable frequency speed regulation system neural network speed sensor
Introduction to 0
With the development of power electronics technology, the control technology of AC asynchronous motor
Technology has changed from scalar control to vector control. In the vector control system
Generally speaking, closed-loop control of speed is essential. For the sake of reality
At present, speed sensors are usually used for speed closed-loop control and magnetic field orientation.
Perform speed detection. The speed sensor is easy to install and maintain.
Affected by the environment, the simplicity and cheapness of asynchronous motors are seriously affected.
Sex and reliability. Therefore, the vector control system without speed sensor becomes
It is the main research content of AC speed regulation.
At present, people have proposed various speed identification methods to replace speed.
Degree sensors, such as dynamic estimation method, model reference adaptive method, expansion
Kalman filter method, neural network method, etc. In which model reference adaptation
This method has the characteristics of good stability and less calculation [1].
Based on the vector control theory of rotor magnetic field orientation, the rotor magnetic field orientation under static seat is studied.
A speed method based on model reference adaptive theory is proposed in the standard system.
The algorithm is deduced, and the system is realized by Matlab /Simulink software.
Simulation.
Vector control of 1 AC asynchronous motor
According to the parameter vector used for orientation, the vector control can
According to the rotor field orientation and stator field orientation, it is divided into vector control.
At present, the vector control method of rotor magnetic field orientation is widely used.
A control method of high performance AC motor based on [2].
When the two-phase synchronous rotating coordinate system is oriented according to the rotor flux linkage, there should be
ψrd =ψr, ψrq = 0, that is:
Te = np
optical microscope
Lawrencium (Lw)
isq
isd =
1 + Tr p
optical microscope
ψr
ψr =
optical microscope
1 + Tr p
IP service equipment; Teaching system design
λ =
optical microscope
Trψr
isq ( 1)
Where: Lm =
three
2
M is the coaxial equivalent winding between stator and rotor in d2q coordinate system.
Mutual inductance between groups; Lr =Lrl +Lm is equivalent two-phase rotor winding in d2q coordinate system.
Group self-cognition; λ is the rotational angular velocity of the d2q coordinate system relative to the rotor;
P is the derivative operator, that is, p = d/dt；; S stands for stator; R stands for rotor; d
Represents the d axis; Q stands for q axis; M represents mutual inductance between coaxial stator and rotor;
Np is a polar logarithm; Tr =Lr /Rr is the rotor time constant.
5 1
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS Xiong Xiang et al
1994-20 10 China academic journal electronic publishing house. Copyright ki.net.
Process automation instrument vol131no122010.
2 simulation model of variable frequency speed regulation system
Figure 1 shows the speed sensorless vector control system of AC asynchronous motor.
Block diagram. The system consists of motor, inverter, flux observer and speed identifier.
It is a closed-loop vector with speed and flux linkage of current inner loop.
Control system.
Figure 1 Block diagram of speed sensorless vector control system
Figure 1 Block diagram of vector control system with No.2 speed sensor
Simulation model of variable frequency speed regulation system based on vector control
The realization steps are as follows: firstly, the asynchronous motor stator in the three-phase coordinate system is electrically connected.
The streams 1a, 1b and 1c undergo three-phase/two-phase (Clarke) conversion, and then pass through two-phase/
Two-phase rotation (Park) transformation is used to obtain power in synchronous rotation coordinate system d2q.
Flow Id, Iq, and then imitate the control method of DC motor to get DC.
Finally, the control quantity of the motor is realized through the corresponding coordinate inverse transformation.
Control of asynchronous motor. Its essence is equivalent to AC motor.
DC motor, respectively used for speed control, magnetic field (φr
Control) two components for independent control. By controlling the rotor flux linkage,
Decompose the stator current to get two components, torque and magnetic field, and then sit down.
Scale transformation to achieve orthogonal or decoupling control [3].
2. 1 speed identification based on MRAS
2. 1. 1 basic model reference adaptive system
Vector control system for realizing rotor flux orientation, flux linkage view
Measurement is very important. In speed sensorless control, it is usually used.
Stator voltage and stator current electricity based on two-phase stationary α2β coordinate system
Rotor flux is estimated by pressure model [4-5]. According to the two-phase static coordinates
According to the basic equation of asynchronous motor, voltage and current can be obtained.
Two forms of rotor flux estimation model.
The voltage model is calculated as follows:
ψrα =
Lawrencium (Lw)
optical microscope
[ ∫( usα - Rs isα ) dt - σLs isα ]
ψrβ =
Lawrencium (Lw)
optical microscope
[ ∫( usβ - Rs isβ ) dt - σLs isβ ] ( 2)
After calculating the voltage model value, the basic model reference is adaptive.
The current model of the system is calculated as follows:
pψrα =
optical microscope
Tr
isα -
ψrα
Tr
- ωrψrβ
pψrβ =
optical microscope
Tr
isβ -
ψrβ
Tr
- ωrψrα ( 3)
Where ψrα and ψrβ are the α axis and β axis in the two-phase stationary α2β coordinate system, respectively.
Rotor flux linkage of shaft; Isα and isβ are the sum of α axes in a two-phase stationary α2β coordinate system.
Stator current of beta axis; Usα and usβ are the α axis in the two-phase stationary α2β coordinate system.
And the stator voltage of the beta axis; σ is leakage inductance.
The difference between the reference model and the adjustable model output (rotor flux linkage) is fixed.
Meaning:
e =ψr - ψ3
r ( 4)
Based on popov's hyperstability theory, an adaptive method for rotor estimation is derived.
The convergence speed is [6]:
ωr = kp +
Kiribati
S
English (5)
Where: kp and ki are the proportional systems in the PI regulator with adaptive structure respectively.
Numbers and integral constants.
The specific steps of speed identification based on MRAS are: selecting voltage
This model is a reference model, and the current model is an ideal model, so the model is built.
Referring to the adaptive system, choose the appropriate adaptive law and make the adjustable mode.
The speed of the model is close to the real motor speed. The structural block diagram of this method is as follows
As shown in figure 2.
Fifty two
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS Xiong Xiang et al
1994-20 10 China academic journal electronic publishing house. Copyright ki.net.
Automation Instrument Volume 2 3 1 Phase 2 20 10
Fig. 2 Block diagram of model reference adaptive system
Fig. 2 block diagram of ras
The adaptive mechanism adopts PI regulator, that is, the proportional integral is selected as
The law of adaptability. In the model reference adaptive system, the reference model should
This is ideal, that is, Formula (2) should always reflect the actual situation of the motor.
State. In this equation, the stator resistance Rs is a variable parameter Rs.
If it is not accurate, it will have a great impact on the low-frequency integration results. In addition, low
Using filter instead of pure integration link can effectively overcome the integrator's
Some defects, such as error accumulation or DC drift; But around the frequency
Or below the cut-off frequency, the amplitude and phase deviation will be serious.
Affect the accuracy of flux estimation.
2. 1.2 improved model reference adaptive system
The advantage of the model reference adaptive structure is that the output of the model is not needed.
It is the actual rotor flux linkage, as long as it is an auxiliary variable related to it.
Therefore, a new auxiliary variable can be used as the output of the model to construct its.
His MRAS speed identification method.
By improving Figure 2, we can get the corresponding principle block diagram, such as
As shown in figure 3.
Fig. 3 block diagram of improved model reference adaptive system
Fig. 3 block diagram of improved MRAS
The stator voltage vector equation of the reference model can be written in the following form.
Namely:
optical microscope
Lawrencium (Lw)
×
dψrα
Trembling insanity (abbreviation for Delirium Tremens)
= usα-Rs is α-σLs ×
disα
Trembling insanity (abbreviation for Delirium Tremens)
optical microscope
Lawrencium (Lw)
×
dψrβ
Trembling insanity (abbreviation for Delirium Tremens)
= usβ-Rs is β-σLs ×
disβ
Trembling insanity (abbreviation for Delirium Tremens)
(6)
Where Ls =Lsl+Lm is equivalent two-phase stator winding in d2q coordinate system.
Group self-cognition.
In vector control based on rotor magnetic field orientation, the equivalent electric field is calculated by the equivalent electric field.
As can be seen from the road, εr =
optical microscope
Lawrencium (Lw)
dψr
Trembling insanity (abbreviation for Delirium Tremens)
Rotor flux vector induction power generation
Pressure, so the formula (6) can be converted into:
εr
α =
optical microscope
Lawrencium (Lw)
×
dψrα
Trembling insanity (abbreviation for Delirium Tremens)
= usα-Rs is α-σLs ×
disα
Trembling insanity (abbreviation for Delirium Tremens)
εr
β =
optical microscope
Lawrencium (Lw)
×
dψrβ
Trembling insanity (abbreviation for Delirium Tremens)
= usβ-Rs is β-σLs ×
disβ
Trembling insanity (abbreviation for Delirium Tremens)
(7)
2.2 speed control module
In the actual system, due to the changes of system state and parameters, etc.
In the process, there will be uncertainty of state and parameters, which is difficult for the system to achieve.
Achieve the best control effect. Based on the above problems, this paper adopts RBF neural network.
The parameters of PID controller are adjusted online by neural network. according to
The PID control system of RBF neural network is shown in Figure 4.
Fig. 4 P ID control system based on RBF neural network
Fig. 4 P ID control system based on RBF neural network
The control error of the system is:
e ( k) = r( k) - y ( k) (8)
The input of PID is:
x ( 1) = e ( k) - e ( k - 1)
x ( 2) = e ( k)
x(3)= e(k)-2e(k- 1)+e(k-2)(9)
The specific expression of the control algorithm using incremental PID is:
u(k)= u(k- 1)+KP[r(k)-y(k)]+ki[e(k)]+
kd [ e ( k) - 2e ( k - 1) + e ( k - 2) ]
Du = kp [ r( k) - y ( k) ] + ki [ e ( k) ] +
KD[e(k)-2e(k- 1)+e(k-2)]( 10)
The adjustment performance index function of neural network is:
J ( k) =
1
2
[ r( k) - y ( k) ]2 ( 1 1)
Through the gradient descent method, we can get [7]:
δKP =-η
9J
9kp
= - η
9J
9y
×
9y
9Du
×
9Du
9kp
=ηe ( k)
9y
9Du
x ( 1)
δki =-η
9J
9ki
= - η
9J
9y
×
9y
9Du
×
9Du
9ki
=ηe ( k)
9y
9Du
x (2)
δKD =-η
9J
9kd
= - η
9J
9y
×
9y
9Du
×
9Du
9kd
=ηe ( k)
9y
9Du
x (3)
( 12)
Where η is the learning rate. The output of the controlled object becomes the control input.
The Jacobian matrix information algorithm for sensitivity information is as follows:
9y
9Du
≈
9yL ( k)
9Du
=∑
m
j = 1
ωj hj
Chinese, Japanese and Korean characters
b2
j
( 13)
Where: hj is the output of the j th hidden layer point; Cji is the middle of Gaussian transformation function.
Cardiac position parameters; Bj is the width parameter of Gaussian function of the j th hidden node.
The structure of neural network is 3-6- 1, that is, the input layer has three nodes.
Point, the hidden layer has 6 nodes, the output layer has 1 nodes, and the learning rate is
53
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS Xiong Xiang et al
1994-20 10 China academic journal electronic publishing house. Copyright ki.net.
Process automation instrument vol131no122010.
0.25，a = 0。 05, β = 0.0 1, and the initial value of PID = [0. 03, 0.00 1, 0. 1],
The initial weight = [3,4, 1], and the sampling period is 0. 00 1. Because RBF God
If the network PID controller is simplified, it can not be directly described by transfer function.
It is impossible to simply use Simulink to simulate it. In this article,
RBF neural network PID controller adopts S2 function in Matlab.
Now [8].
2.3 Torque control module and flux control module
Both torque controller and flux controller adopt PI control algorithm, which can
Get:
intelligence quotient
three
= kp ( T
three
e - Te ) + ki ∫( T
three
e - Te ) dt ( 14)
I
three
d = kp ( phir
three
- phir) + ki ∫( phir
three
- phir) dt ( 15)
Where kp and ki are the proportional gain coefficient and the integral gain coefficient, respectively.
2.4 Simulation results and analysis
Using the above simulation model, the vector control variable frequency speed regulation system is improved.
Simulate no-load and constant-speed loading operation. When the load torque value is 0,
The obtained simulation diagram of stator current, speed and torque of asynchronous motor is shown in Figure 5.
As shown in the figure.
Fig. 5 Simulation diagram of stator current, speed and torque
Figure 5 Current, speed,
Torque of stator
The relevant parameters of the selected asynchronous motor are as follows: The rated data is
4 1 kw, 380 volts, 4 poles, 50 Hz, moment of inertia J = 1. 662km2, Rs =
0.087ω、Rr =0。 228ω、Ls =Lr =0。 8mH、Lm =34。 7mH .
Rotor flux reference with inverter current DC bus voltage of 780V
The value is 0. 96Gb specifies that all state variables are in powerful initialization.
The condition is 0, or the initial condition of asynchronous motor is [1, 0, 0, 0,
0, 0, 0, 0], so that the motor can be started in a stopped state. In order to increase
The simulation speed is fast, and ode23 t simulation algorithm is adopted.
In the starting stage of the motor, the flux linkage and electromagnetic torque are in the rising stage.
In the initial no-load state, the electromagnetic torque finally drops to zero. At t =
0 s, 1 s, the given quantity jumps instantly from 60 rad/s due to the rotating speed.
80 radians/second, and at the time of start-up, the rotor speed has stabilized at 0. 5 s。
Therefore, at steady state, the stator current starts and turns.
Torque current and electromagnetic torque are available when starting and given speed commands change.
Overshoot, torque current and electromagnetic torque under automatic adjustment of the system
The number began to decrease slowly and tended to be stable. As can be seen from the simulation, in
Under the control method adopted by the control system, the system has good
Static and dynamic performance, good sine of stator current; And estimate
Good speed steady-state accuracy, which can accurately track the change of motor speed;
The mechanical angular velocity of the motor can quickly track the given mechanical angular velocity.
Variable, the motor has good starting performance. Actual speed and recognition
The comparison of velocity simulation diagrams is shown in Figure 6.
Fig. 6 Comparison of Simulation Diagram of Actual Speed and Recognized Speed
Fig. 6 Comparison between actual speed and identified speed
3 Conclusion
The simulation test uses the motor stator voltage and current which are easy to measure.
Flow, combined with vector control and MRAS principle, real-time identification of motor speed.
Through theoretical analysis and simulation research, the model reference adaptive method is adopted.
Estimating the rotor speed of AC asynchronous motor has the advantages of small calculation and fast convergence.
The simulation results verify the feasibility and effectiveness of the system.
refer to
[ 1]cirricionem，PucciM。 A MRAS 2 based on velocity estimation method
Od with linear neurons for high performance induction motor drive
And its experiments [C]IEEE E2 International Conference
Motors and Drivers, IEMDC'03, 2003 (s 1): 6 17-623.
[2] Chen. Electric drive automatic control system [M]. Version 2. Beijing: Machinery Industry Press.
Society, 2004.
[3] Wang Zhongli, Duan Huida, Gao Yufeng. Matlab application technology-electrical engineering and self.
Application in animation specialty [M]. Beijing: Tsinghua University Publishing House, 2007.
Wang Qinglong, Zhang Chongwai, Zhang Hang. Speed sensorless vector control system for ac motor.
Structural model reference adaptive speed identification [J]. Journal of China Electrical Engineering, 2007,
27 ( 15) : 70 - 74.
Dai Yuxing, Wang Yaonan, Chen Jida. Model reference adaptive speed transmission based on DSP.
Sensor vector control [J]. Information and Control, 2003,32 (6): 507-511.
Liu xingtang Apply adaptive control [M]. Xi 'an: Northwestern Polytechnical University Press,
2003.
[7] Zhang Mingguang, Li Wenhui and Liu Manqiang. Adaptive PID control
Strategy based on RBF neural network identification [C] 17 IEEE
2005 International Conference on Neural Networks and the Brain,
ICNN & amp； B, 2005: 1854-1857.
Liu Jinkun. Advanced PID control and its Matlab simulation [M]. Beijing: Electronic industry.
Press, 2003.
54
Research on variable frequency speed regulation system of AC asynchronous motor based on MRAS Xiong Xiang et al