This article has been succefully presented an adaptive
fuzzy sensorless speed control for PMSM drive based on
reduced-order. It demonstrated through co-simulation by
using Simulink and ModelSim. The system separate to
three parts, the vector control is used to decouple the
nonlinear characteristics of PMSM, the AFC is designed
to cope with the dynamic uncertainty effect, the reducedorder EKF is applied to estimate the rotor FA without
using sensor. The VHDL is used to describe the behavior
of whole system.The paper has been demonstrated to be
the best tracking rotor speed under different external load
condition.
5 trang |
Chia sẻ: huongthu9 | Lượt xem: 531 | Lượt tải: 0
Bạn đang xem nội dung tài liệu Mô phỏng bộ điều khiển mờ thích nghi và kalman mở rộng trong điều khiển tốc độ động cơ pmsm không sử dụng cảm biến, để tải tài liệu về máy bạn click vào nút DOWNLOAD ở trên
80 Tạp chí Khoa học Lạc Hồng Số 04
Journal of Science of Lac Hong University
Vol. 4 (12/2015), pp. 80-84
Tạp chí Khoa học Lạc Hồng
Số 4 (12/2015), trang 80-84
MÔ PHỎNG BỘ ĐIỀU KHIỂN MỜ THÍCH NGHI VÀ KALMAN MỞ
RỘNG TRONG ĐIỀU KHIỂN TỐC ĐỘ ĐỘNG CƠ PMSM KHÔNG SỬ
DỤNG CẢM BIẾN
Modelsim/Simulink co-simulation of adaptive fuzzy and EKF-based flux
angle and rotor speed estimation for PMSM
Nguyễn Vũ Quỳnh1, Hoàng Thị Nga2, Trần Trung Hướng3
1vuquynh@lhu.edu.vn, 2hoangthinga87@gmail.com, 3trunghuongphuyen@gmail.com
Khoa Cơ Điện – Điện Tử Trường Đại học Lạc Hồng
Đến tòa soạn: 25/12/2014; Chấp nhận đăng: 16/2/2015
Abstract. This article has presented an adaptive fuzzy controller for permanent magnet synchronous motor. The rotor speed
estimation based on reduced-order extended kalman filter (reduced-order EKF). The sensor less algorithm controls have
implemented by very high speed integrated circuit hardware description language (VHDL). The simulation work is
performed by MATLAB/Simulink and ModelSim co-simulation mode. The simulation results shown that the motor’s speed
has good dynamic performance and isn’t sensitive to the parameter variations.
Keywords: Adaptive fuzzy; Motor controller; VHDL; Extended kalman filter
Tóm tắt. Bài báo này trình bày thuật toán mờ thích nghi dùng để điều khiển tốc độ cho động cơ đồng bộ nam châm vĩnh cửu.
Tốc độ của rotor được ước lượng dựa trên bộ lọc kalman mở rộng giảm bậc. Toàn bộ thuật toán điều khiển của động cơ được
lập trình bằng ngôn ngữ mô tả phần cứng VHDL. Hệ thống mô phỏng được kết hợp giữa Matlab/Simulink và ModelSim. Kết
quả mô phỏng thể hiện, tốc độ động cơ tốc độ động cơ đáp ứng tốt với tốc độ đặt và không ảnh hưởng khi thông số của hệ
thống thay đổi.
Từ khóa: Mờ thích nghi; Điều khiển động cơ; VHDL; Bộ lọc kalman mở rộng
1. INTRODUCTION
The PMSM controller needs an optical encoder to
calculate the rotor speed. However, sensor presents some
disadvantages such as drive cost, machine size, reliability
and noise immunity; therefore, a sensorless control
without encoder for motor drive become a popular
research topic in literature [1-3]. The reduced-order EKF
is a good choice for estimation the rotor speed without
encoder. The EKF requires heavy on-line matrix
computing for a fix-pointed processor system. In
realization, a fix-pointed processor using digital signal
processor (DSP) or field programmable gate array
(FPGA) both can provide a solution in this issue.
Especially, FPGA is better for the implementation of the
digital system than DSP [3].
In this article, a co-simulation is designed for
sensorless speed control for PMSM drive (Fig.1). The
reduced-order EKF is used to estimates the rotor flux
angle (FA) and velocity. The vector control is applied for
PMSM drive with Clark, Park, invert Clark and invert
Park transformation. After using vector control, the
PMSM will be decoupled and controlling a PMSM is like
controlling a DC motor. The adaptive fuzzy controller
(AFC) is applied for controlling the velocity of PMSM.
The AFC herein uses singleton fuzzifier, triangular
membership function, product-inference rule and central
average defuzzifier method [4-6].
2. VELOCITY CONTROLLER DESIGN
The structure of AFC was shown in Fig.2. It includes
Fuzzification (FI), inference mechanism, knowledge base,
adjust mechanism, defuzzification (DFI), and PI
controller.
The input linguist values of fuzzy controller:
(1)
(2)
Membership function was symmetrical triangle and
controlling rules is:
If e was Am and de was Bn then uf was Cm,n (3)
Crisp value in the output of fuzzy controller:
(4)
In which cm,n dn,m is adjusting parameters for fuzzy
controller.
In this article, the adaptive feature of system was added
for changing the knowledge base of fuzzy controller.
With this adjusting mechanism, the controller can control
the motor with a balance speed in case load changed. The
adjust mechanism was added to fuzzy controller, so the
controller becomes AFC. The input of adjust mechanism
were error between estimated rotor speed ( ) and
designed speed ( ). Using the method of gradient
descent determines the adaptive control law for the
system:[5]
Definition of instantaneous value function:
Modelsim/Simulink Co-Simulation of Adaptive Fuzzy and Ekf-Based Flux Angle and Rotor Speed Estimation for PMSM
81 Tạp chí Khoa học Lạc Hồng Số 04
(5)
Parameters cm,n was determined by variation of the
instantaneous value function
,
,
( 1)
( 1)
( )
m n
m n
J k
c k
c k
¶ +
D + = -
¶ (6)
In which: α = 0÷1 shows the adaptive rate of the system.
Bilinear transform of electromagnetic torque with
considering the mechanical load:
(7)
In which Ts is sampling time, z-1 is a stage of delay time
Figure 1. The controller with AFC, vector control and reduced-order EKF
Relationship between current and output of velogcity
controller was described by this equation (PI controller):
(8)
In which Kp, Ki are the gain of PI controller; uf is
output value of AFC.
From equation (7) and (8), we obtained the relationship
between motor speed and output functions of the AFC:
(9)
In which: ,
Variation of the instantaneous value function J(k+1)
from equation (5) was:
(10)
Therefore, the parameters of the AFC could be adjusted
through the function
(11)
Figure 2. The structure of AFC
3. EKF-BASED FA AND ROTOR SPEED
ESTIMATION FOR PMSM [7]
The equation of PMSM on the axis:
(12)
EMF is defined as
Where , is voltage on fixed
coordinate; is current on fixed coordinate; is
angular position at magnet flux; s is differential operator.
(13)First, redefine the system
Nguyễn Vũ Quỳnh, Hoàng Thị Nga, Trần Trung Hướng
82 Tạp chí Khoa học Lạc Hồng Số 04
and (14)
and assume that the rotor angular speed is constant at
each sampling period. Then from (13)~(14), the state
equation of PMSM stochastic model can be straightly
obtained by
and (15)
The Jacobian matrices can be expressed as:
(16)
The simplified exponential matrix can be:
(17)
Where , , and
. Further, due to the PMSM stochastic model in
(14) has not input signals and the states of
z
and (
and ) cannot be directly observed, the EKF algorithm
cannot be applied. Considering (11)~(12), and can
be indirectly calculated by the discrete model:
(18)
However, it is not a causal system because
and cannot be measured at sampling instant time
n. To solve this problem, we assume that the current is
constant at each sampling period, and the (17) can be
further simplified as
(19)
The initial values of , and need to be chosen.
Through the recursive calculation, the state value of
is estimated at each sampling
period, then the rotor angular speed and rotor position can
be respectively derived by
and (20)
Finally, a summary for estimating the rotor position
and rotor speed based on reduced-order EKF is shown by
the following design procedures:
Step A: Set the values for Qd, R and P0
Step B: Calculate the zα(n), zβ(n) from (19).
Step C: Estimate the temporary state variables
(21)
Step D: Obtain the temporary covariance matrix
(22)
Step E: Calculate the Kalman gain from
(23)
Step F: Tune the present state variables
(24)
With ;
Where kij is the element of Kalman gain kn.
Step G: Update the present covariance matrix .
(25)
Step H: Calculate the rotor angular speed and rotor flux
position from (20); back to Step B.
4. SIMULATION STRUCTURE AND
SIMULATION RESULTS
The Fig.3 shown the simulation structure of system, the
work-1 to work-3 respectively performs the function of
AFC, vector control and reduced-order EKF. All works in
ModelSim are coded by VHDL and the FPGA resource
usages shown in table 1. The PMSM and inverter are
executed by SimPowerSystem blockset. The ModelSim
executes the co-simulation using VHDL code. The
PMSM parameters are shown in table 1.
The desired speed is set 500rpm, 1000rpm, 1500rpm,
2000rpm, 2500rpm, 2000rpm, 1500rpm, 2000rpm and
1500rpm for testing. The results of the actual rotor FA,
the estimated rotor AF under different motor speed are
shown in Fig.4. The error of the estimated rotor FA and
the actual rotor FA are among 0.5% at high speed (Fig.4a)
and 1% at low speed (Fig.4b).
Modelsim/Simulink Co-Simulation of Adaptive Fuzzy and Ekf-Based Flux Angle and Rotor Speed Estimation for PMSM
83 Tạp chí Khoa học Lạc Hồng Số 04
Figure 3. The co-simulation system of sensorless controller build in Matlab/Simulink and Modelsim
The Fig.4 shows that the reduced-order EKF apparently
gives accuracy, especially in high speed condition. In the
next step, the estimated rotor FA is feed-backed to the
current loop and the estimated rotor speed is feed-backed
to the speed loop as Fig.1.
Figure 4. Comparision between actural rotor FA and estimated
rotor FA at 500 rpm (a) and 2500 rpm (b)
Figure 5. With AFC and reduced-order EKF, the rotor speed
can track speed command very well (a), the id current
approached to zero, the motor torque was controlled by iq
current (b).
In Fig.5a and Fig.6, they showed that the rotor speed
using AFC can track the designed speed very well. The
rotor speed is running without overshoot and
sluggishness. The id current in Fig.5b approached to zero,
it shown that the vector control algorithm was correct.
From Figs.4~5 demonstrates that the AFC sensorless
control can give good speed tracking.
Figure 6. With AFC and reduced-order EKF, the desired speed
is sine shape, the rotor speed still track desired speed very
well.
Nguyễn Vũ Quỳnh, Hoàng Thị Nga, Trần Trung Hướng
84 Tạp chí Khoa học Lạc Hồng Số 04
5. CONCLUSION
This article has been succefully presented an adaptive
fuzzy sensorless speed control for PMSM drive based on
reduced-order. It demonstrated through co-simulation by
using Simulink and ModelSim. The system separate to
three parts, the vector control is used to decouple the
nonlinear characteristics of PMSM, the AFC is designed
to cope with the dynamic uncertainty effect, the reduced-
order EKF is applied to estimate the rotor FA without
using sensor. The VHDL is used to describe the behavior
of whole system.The paper has been demonstrated to be
the best tracking rotor speed under different external load
condition.
Table 1. FPGA resources and PMSM’s parameter
FPGA resources
Item Les RAM bits
Work-1 2,887 0
Work-2 1,828 98,304
Work-3 4,226 75,264
Table 2. PMSM’s parameter
PMSM’s parameter
Item Value
R 1.3W
Ld, Lq 6.3mH
P 4
Jm 0.000108 kg*m2
F 0.0013 N*m*s
6. REFERENCES
[1] S. Bolognani, O. Roberto, and Z. Mauro, “Sensorless full-
digital PMSM drive with EKF estimation of speed and rotor
position”, IEEE Transactions on Industrial Electronic,
vol.46, no.1, pp.184-191, Feb 1999.
Doi:10.1109/41.744410.
[2] M. C. Huang, A. J. Moses, & F. Anayi, “The comparison of
sensorless estimation techniques for PMSM between
extended Kalman filter and flux-linkage observer”,
in Twenty-First Annual IEEE Applied Power Electronics
Conference and Exposition, 2006. APEC'06.
[3] L. Idkhajine, E. Monmasson and A. Maalouf, “Extended
kalman filter for ac drive sensorless speed controller, FPGA-
based solution or DSP-based solution”, Proceedings of IEEE
International on Industrial Electronics (ISIE), pp. 2759-
2764, 2010. Doi: 10.1109/ISIE.2010.5636605.
[4] Ying-Shieh Kung, Nguyen Vu Quynh, Chung-Chun Huang
and Liang-Chiao Huang, “Simulink/ModelSim co-
simulation of sensorless PMSM speed controller”, IEEE
Symposium on Industrial Electronics and Applications, pp.
24-29, 2011. Doi: 10.1109/ISIEA.2011.6108709.
[5] Ying-Shieh Kung and Ming-Hung Tsai, “FPGA-Based speed
control IC for PMSM drive with adaptive fuzzy control”,
IEEE transactions on power electronics, vol.22, no.6,
pp.2476,2486, Nov. 2007. Doi:
10.1109/TPEL.2007.909185.
[6] Y. S. Kung, C. C. Huang, & M. H. Tsai, “ FPGA realization
of an adaptive fuzzy controller for PMLSM
drive”, Industrial Electronics, IEEE Transactions on, vol.56,
no.8, 2923-2932, 2009.
[7] Ying-Shieh Kung, Nguyen Vu Quynh, Nguyen Trung Hieu,
and Jin-Mu Lin, “FPGA realization of sensorless PMSM
speed controller based on extended kalman filter”,
Mathematical Problems in Engineering, vol. 2013, article ID
919318, 13 pages, 2013. Doi:10.1155/2013/919318.
Nguyễn Vũ Quỳnh
Sinh năm 1979. Anh nhận bằng thạc sỹ về thiết bị, mạng và nhà máy điện của trường Đại học Sư phạm Kỹ
thuật Tp. Hồ Chí Minh năm 2005 và bằng Tiến sỹ về Kỹ thuật điện của trường Southern Taiwan University of
Science and Technology, Đài Loan 2013. Hiện anh là giảng viên khoa Cơ Điện – Điện Tử - Đại học Lạc
Hồng. Hướng nghiên cứu chính là thiết kế và thực hiện các hệ thống đo lường, điều khiển, các hệ thống
nhúng, bộ điều khiển thông minh và FPGA.
Hoàng Thị Nga
Sinh năm 1987. Chị nhận bằng kỹ sư Điện – Điện tử của trường Đại học Lạc Hồng năm 2010. Hiện chị là giảng
viên khoa Cơ Điện – Điện Tử - Đại học Lạc Hồng. Hướng nghiên cứu chính là thiết kế hệ thống điều khiển, các
hệ thống nhúng, bộ điều khiển thông minh và FPGA.
Trần Trung Hướng
Sinh 1984. Anh nhận bằng kỹ sư Điện –Điện tử của Trường Đại học Lạc Hồng năm 2009. Hiện là nhân viên công
ty Aureole Fine Chemical Products ( AFCP) khu công nghiệp Long Bình, Biên Hòa, Đồng Nai. Hướng nghiên
cứu chính in 3D, thuật toán điều khiển.
TIỂU SỬ TÁC GIẢ
Các file đính kèm theo tài liệu này:
- mo_phong_bo_dieu_khien_mo_thich_nghi_va_kalman_mo_rong_trong.pdf