2014 International Symposium on Computer, Consumer and Control

An Improved Vector-Control System of PMSM Based on

Fuzzy Logic Controller
Risha Na*,1, Xudong Wang#,1
*
School of Electrical Engineering, Harbin University of Science & Technology
No.52 Xufu Road, Nangang District, Harbin, China
1
[email protected]

Abstract— This paper presents a closed loop vector control of proposed method is then implemented for a 45kW PMSM.
permanent magnet synchronous motor (PMSM) AC servo The performance of the proposed PMSM drive is then
system with improved current and speed controller. The effective compared with that of a conventional PI controller based
current and speed controller are designed based on integral PMSM drive in both simulation and experiment.
separation strategy and fuzzy-PI double mode control strategy. A
performance comparison of the proposed PMSM drive system II. MATHEMATICAL MODEL OF PMSM ANDPRINCIPLE
with conventional PI controller based drive system is provided in
OF VECTOR CONTROL OF SYSTEM
simulation and experiment. Comparative simulation and
experimental results demonstrate the better dynamic responses
A. MATHEMATICAL MODEL OF PMSM
in both current and speed for the proposed fuzzy control based
PMSM drive system. Before constructing the mathematical model, take the
following assumptions:
Keywords— Permanent magnet synchronous motor (PMSM)˗
˗ x Neglect saturation of iron core in motor.
vector control˗fuzzy logic control
x Regardless of the vertex current and hysteresis loss.
I. INTRODUCTION x Symmetrical three-phase sinusoidal waveform current.
Nowadays the permanent magnet synchronous motor
­ d
(PMSM) is very popular for variable-speed drive systems due ° u Rs id  < d  Z< q (1)
to its high torque–current ratio, large power–weight ratio, high ° d dt
®
°u d
efficiency, high power factor and robustness [1-2].Despite the Rs iq  < q  Z< d
many advantageous features of a PMSM, the precious speed °̄ q dt
control of the motor is difficult due to nonlinear coupling ­< d Ld id < f (2)
®
¯< q Lq iq
among its winding currents and the rotor speed [3-4]. The
most important characteristics of high performance drive
3
systems are fast and precise speed response, quick recovery of Te p (< d iq < q id ) (3)
speed from any load disturbances. 2
There are many studies on improved methods to handle the J dZ (4)
˜ Te  TL
control issues in motor driving systems, e.g., model-reference p dt
adaptive controller, sliding-mode controller, variable-structure Where u d , u q are stator voltage component on d-q axis; so
controller, etc. [5–8] The designs of these controllers depend as i d , i q for stator current component; and \ d , \ q for flux
on exact machine model and accurate model parameters.
However, it is very difficult of obtaining the exact parameters linkage component; R s is stator winding resistance; L d , L q are
of the PMSM. On the other hand, the designs of intelligent winding inductance component;  f is flux linkage which is
controllers, such as fuzzy logic, do not need exact generated by permanent magnet; T e is motor output torque; T L
mathematical model of the system. Simplicity and less is load torque; P is pair number of motor poles; J is rotational
intensive mathematical design requirements are the main inertia; Z is electric angular velocity.
features of intelligent controllers, which are suitable to deal B. SELECTION OF CONTROL STRATEGY
with nonlinearities and uncertainties of electric motors [9–13]
Among the various intelligent controllers, fuzzy logic For the designed salient pole permanent magnet
controller is the simplest one and which is better than others in synchronous motor, use maximum torque/current (MTPA)
terms of quick response time, insensitivity to parameter and strategy to control it. In vector control, for each T e , i d and i q
load variation. [14–16] In this paper, fuzzy-PI double mode there are many groups of selections, which are added up to
controller is utilized in speed control loop, and current compose the constant torque curve. The nearest point from the
controller is designed based on integral separation strategy. origin of coordinate in constant torque trajectory is the
The complete simulation model of the proposed PMSM minimum current vector that can generate that certain required
drive system is developed using Matlab/Simulink. The torque. Connecting these minimum current point that produce
different required Torques, the maximum torque/current

978-1-4799-5277-9/14 $31.00 © 2014 IEEE 326

DOI 10.1109/IS3C.2014.92
Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.
 &RQVWHQW
trajectory of PMSM is formed, as shown in Fig.1. AB is the 073$7UDMHFWRU\
T0 T2 iq
WRUTXHFXUYH
MTPA trajectory of current vector MTPA control curve, by 9ROWDJHOLPLWHOOLSVH T1 &XUUHQWOLPLW
optimizing the location of current vector working point, A FLUFOH

improvement of utilization rate of unit current can also be

acquired. When using maximum torque/current control method, C

the current vector should satisfy: ¹0 ¹1 B

­ w Tem / is id
° 0
° wid (5)
®
° w Tem / is
0
° wiq
¯
p (6) FIG. 1. MTPA trajectory of current vector
Tem p[I f iq  ( Ld  Lq )iq id ] [e0iq  ( X d  X q )iq id ]
Z
III. DESIGN OF SPEED AND CURRENT CONTROLLER
i i 2
i 2
Where s q d
,put (5) into (6) the following
equation can be obtained: A. CURRENT REGULATOR DESIGN
I f  I f 2  4( Ld  Lq ) 2 iq 2 (7) Nature of vector control is the control of stator current.
i d
2( Ld  Lq ) Current regulator’s control goal is that current curve has no big
overshoot when starting, and good tracking performance of real
T *
em i (1  id * )
*
q
(8) current to current reference during steady state operation.
Again put (7) into (8) the following equation can be Current regulator with PI regulation uses only a set of PI
obtained: parameters from starting to steady-state operation, so it can not
­id * f1 Tem* guarantee a good control characteristic during the whole
° (9) process.
® *
°̄iq
*
f 2 Tem When the motor starts the presence of ki will decrease
system stability, increase overshoot, and even cause a big shock,
Applying this control strategy, the motor can accelerate or while the steady state, ki can promote the system regulation,
decelerate with maximum torque (corresponding to the A point and eliminate steady-state error. For this feature, the integral
torque) to track the torque from starting to the target, thereby
separation method is adopted, comparing deviations between
enhancing the tracking rate, while the method can reduce motor,
inverter and the whole system loss to minimum, and then reference current i dref , i qref and actual feedback current id, iq
improve motor efficiency. with the setting threshold value of A, B respectively. When
'i d > A ( 'i q > B), set ki = 0 in order to prevent overshoot,
C. PRINCIPLE OF PMSM VECTOR CONTROL SYSTEM
when 'i d <A ( 'i q <B), increase ki to speed up speed
The diagram shown in Fig.2 is vector control system block
diagram with MTPA control strategy, the control system adjustment. Its functional implementation model is shown as
adopts speed - current double closed loop control structure, Fig.3. M
three-phase current, i A, , i B and i C are measured by stator W
Park-1
P VDC
V V
current sensor, and through the coordinate transformation to nref
+
Iqref
+
Current vsqref DQ
Controller
S
Speed
convert them into DC component i d , i q in d-q coordinate and n - Cntroller Idref - vsdref V
IPM
Current
Controller ¢£
take them as current loop feedback. The deviation between + -
©
speed information and target which is regulated by speed PI to Speed
Iq DQ ¢£ Ia
obtain current reference of d-q axis i dref and i qref , and Calculation

Id Ic
through current regulator and coordinate transformation the ¢£ abc
© Park Clark
deviations between i d ˈ i dref and i qref , i q are calculated to
Position PMSM
Calculation
judge VD and VE in sectors. Using PWM signal produced by
space vector pulse width modulation˄SVPWM˅technology FIG.2. PMSM vector control system frame
to control inverter, thus the double closed-loop vector control
of permanent magnet synchronous motor is achieved.

FIG.3. Current regulator model

327

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.
 B. SPEED REGULATOR DESIGN
Because the vector control system dynamic performance
of permanent magnet synchronous motor (PMSM) with
traditional PI regulation as speed regulator is not ideal, and
simple fuzzy logic control system is difficult to eliminate the
steady-state error due to the lack of integrator either. Based on
the above problem, a dual-mode controller combining fuzzy
theory with PI regulation is proposed and designed. The
simulation model is shown in Fig.4. The controller takes the
relationship between speed deviation and preset threshold value
as the basis for comparison, fuzzy logic is used when the
(b) Change rate of speed error EC
deviation is greater than the setting value, while PI control is
used when the deviation is less than the setting value.
1) Determination of universe
The controller has two input variables and one output
variable, namely speed deviation e, changing rate of speed error
ec and the controller output variable u. Define them as E, EC
and U corresponding to their fuzzy variables respectively, and
express them as: E = EC = U = {NB, NM, NS, ZO, PS, PM, PB}
through fuzzy set, the domain of E and EC are [-6, 6]
respectively, the universe of U is [0, 6].
2) Fuzzy rules and fuzzy reasoning
Fuzzy rules are the core of the speed regulator design, (c) Fuzzy controller output U
fuzzy logic control system is expressed by a series of language FIG. 5. Membership function of fuzzy logic controller
description of expert knowledge, expert knowledge is often TABLE 1. FUZZY CONTROL RULES
composed by conditional statements like "if ... then". To EC
simplify the calculation, using triangular membership functions, U
NB NM NS ZO PS PM PB
and all the membership functions distribute symmetrically, in
the entire universe, the more approaching to the original point NB NB NB NM NM NM NS PS
NM NB NM NM NM NS 0 PS
the more dense near the origin of the membership function, so
NS NM NM NS NS 0 PS PM
that will help to improve the steady-state accuracy. E, EC and U E ZO NM NS NS 0 PS PS PM
of the membership function graph is shown in Fig.5. Fuzzy PS NM NS 0 PS PS PM PM
control rule table is listed in Table 1 PM NS 0 PS PM PM PM PB
PB NS PS PM PM PM PM PB
This paper adopts a simple reasoning compound algorithm-
Mamdani method and Fig.6 is output surface of fuzzy inference.
3) The choice of quantization factor and scale factor
Set speed error and changing rate of speed error factor as
Ke and Kec, the output scale factor as Ku, the selection of these
three parameters has a great influence on the performance of
PMSM control system.
When keeping Kec unchanged and increase Ke,
response speed of system becomes faster, and when Ke is too
large, system overshoot becomes bigger and transition time
FIG. 4. Fuzzy-PI double fuzzy controller model becomes longer, while keeping the same Kec, and a bigger Ke,
system overshoot reduces, the time to steady-state becomes
longer; the bigger scale factor Ku is chosen, the faster system
response becomes, but it will generate oscillation if it is too big,
so as the time to steady-state becomes longer when it is too
small.
From the analysis above, the quantitative factor and scale
factor affect the dynamic and static performance of the system
commonly, fuzzy logic control is same as conventional PI
control that there is a contradiction between dynamic and static
performance, because of fuzzy-PI double mode control which
proposed in this paper has PI control, therefore it can improve
(a) Speed error E
the system steady-state performance, the parameter selection
should focus on improving the system response speed and

328

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.
reduce the overshoot, through the analysis above and repeated (a) PI-control
200
simulation study, obtaining the best group of quantitative
150
factors and scale factors data.
100
4) Clarity

Id&Iq(A)
50
The Clarity of fuzzy logic controller designed in this paper
chooses centroid method, by using this method, the output is 0

much smoother, and the formula is shown as below. -50

n

¦ UF P (UF )
i 1
i i
(10)
-100
0 0.005 0.01 0.015
Time(s)
0.02 0.025 0.03

U n (b) Integral separation control

¦ P (UF ) i FIG. 8. Comparison of simulated current output on d-q axis
i 1 It is visible that when using PI regulation, there is a bigger
Where:  is the membership degree of the first i rule; fluctuation of current waveform on d-q axis when starting, and
UFi is the output center point of the first i rule. it takes a long time to reach steady-state, At steady period, there
is deviation between reference current and actual current, track
IV. SIMULATION AND EXPERIMENT RESULTS effect is not ideal; when using integral separation method, the
current on d-q axis tracks steadily, and can reach steady-state in
A. ANALYSIS OF SIMULATION RESULTS
a short time.
On the basis of using current controller who adopts
integral separation method, taking simulation experiments with
conventional PI speed controller and fuzzy-PI double mode
control speed controller respectively. Set the motor start-up
with 10N·m and increase it to 20N·m at t=0.015s, the simulation
results are shown below, Fig.9 shows torque, speed and phase
current waveforms who adopts conventional PI speed regulator,
Fig.10 shows torque , speed and phase current waveforms with
fuzzy-PI double mode speed regulator.
It’s visible that when using conventional PI speed
FIG. 6. Output surface of fuzzy interence controller, there is obvious torque, speed and three-phase
As shown in Fig.7, establish PMSM vector control current waveforms fluctuation when starting, and it takes longer
system model in MATLAB7.6/Simulink environment, and time to reach steady-state. It leads to a significant drop in speed
take the simulation experiment, the parameters in simulation waveform when increasing the load suddenly, and can not
system are same as that in experiment platform. Set the recover steady-state in a short period of time; When using
simulation parameters like this: R S  , L d =0.3mH, L q fuzzy-PI double mode speed controller, the motor torque, speed
=0.65mH,  f =0.076Wb, P =8, J =0.0008kg·m2. and three-phase current waveforms are smooth, and there are no
PWM period Ts = 0.0001. The simulation time is 0.03s, overshoots, and then it takes shorter time to reach steady-state,
the given speed n ref = 1000r/min, the motor load is set to so it can recover steady-state in a short time when increasing
20N·m, When adopting traditional PI method and integral the load suddenly.
separation method, the waveforms of current on d-q axis are
shown in Fig.8. B. EXPERIMENT RESULTS
Fig.11 is PMSM experiment platform, the experiment
adopts 45KW PMSM, voltage supply is 320V, rated torque is
200N·m. The motor operates at the speed of 1000rpm in the
whole experiments.
Set the speed of motor as 1000rpm, and the load as 70N·m,
In Fig.12, (a) shows current waveform on d-q axis when using
PI control method, and (b) shows current waveform on d-q axis
when using Integral separation control method, there is no
overshoots at starting, and current can track steady in the whole
FIG. 7. Simulation model for PMSM vector control system
operating process.
200 200
150
100
100
Te (N m)
Id&Iq(A)

50
0
0
-50 -100
-100
-150
-200
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
Time(s) Tim e (s )

329

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.
 (a) Torque response waveform 250
1000 200

Spe e d(rpm)
150
500
100
0

Id&Iq(A)
50
0 0.005 0.01 0.015 0.02 0.025 0.03
Tim e (s ) 0
(b) Speed response waveform
200 -50

100 -100
Ia bc(A)

0 -1500 100 200 300 400 500 600 700

Time(ms)
-100
(a) PI control
250
-200
0 0.005 0.01 0.015 0.02 0.025 0.03
Tim e (s ) 200

(c) Phase current response waveforms 150

FIG. 9. Simulated response when using PI control
100

Id&Iq(A)
50
150
0
100
Te (N m)

-50
50
-100
0
0 0.005 0.01 0.015 0.02 0.025 0.03 -150
Tim e ( s ) 0 100 200 300 400 500 600 700
Time(ms)
(a) Torque response waveform (b) Integral separation control
1000 FIG. 12. Comparison of current output on d-q axis
Speed(rpm)

1100
500

0 0.005 0.01 0.015 0.02 0.025 0.03 1050

Tim e ( s )

(b) Speed response waveform

Speed(rpm)

200
1000
100
Ia bc(A)

-100 950
-200
0 0.005 0.01 0.015 0.02 0.025 0.03
Tim e ( s )

(c) Phase current response waveforms 900

0 10 20 30 40 50
FIG. 10. Simulated response when using Fuzzy-PI double mode control Time(s)
(a) PI control
1100

1050
Speed(rpm)

1000

FIG. 11. PMSM experiment platform 950

900
0 10 20 30 40 50
Time(s)
(b) Fuzzy-PI double fuzzy control

330

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.
 FIG. 13. Comparison of speed response with load variation [1] S. Morimoto, M. Sanada, and Y. Takeda, “Wide speed
Fig.13 is speed waveform when the motor start-up with operation of interior permanent magnet synchronous
40N·m load at speed of 1000rpm, then increase to 70N·m, (a) motors with high performance current regulator,” IEEE
Trans. Ind. Appl., vol. 30, no. 4, pp. 920–926, Jul./Aug.
shows the speed waveform when using PI control method; (b) 1994.
shows the speed waveform when using fuzzy-PI double mode [2] N. Bianchi and T. M. Jahns, “Design, analysis, and
control method, It’s visible that the last one can recover steady control of interior PM synchronous machines,” in
in a shorter time, and has a better dynamic characteristic. In Tutorial Course Notes IEEE IAS Annu. Meeting, Seattle,
Fig.14, (a) and (b) are phase-current waveforms of the speed WA, Oct. 3–7, 2004, pp. 1.1–1.10.
when using PI and fuzzy-PI double mode control strategy in [3] S. Morimoto, M. Sanada, and Y. Takeda, “Effects and
compensation of magnetic saturation in flux-weakening
speed controller, It’s visible that when using fuzzy- PI double controlled permanent magnet synchronous motor drives,”
mode control strategy the phase-current waveform can operate IEEE Trans. Ind. Appl., vol. 30, no. 6, pp. 1632–1637,
steadier, and has no fluctuation. Nov./Dec. 1994.
[4] C. Mademlis and V. G. Agelidis, “On considering
magnetic saturation with maximum torque to current
control in interior permanent magnet synchronous motor
drives,” IEEE Trans. Energy Convers., vol. 16, no. 3, pp.
246–252, Sep. 2001.
[5] M. M. I. Chy and M. N. Uddin, “Development and
implementation of a new adaptive intelligent speed
controller for IPMSM drive,” IEEE Trans. Ind. Appl., vol.
45, no. 3, pp. 1106–1115, May/Jun. 2009.
[6] M. Tursini, F. Parasiliti, and D. Zhang, “Real-time gain
tuning of PI controllers for high-performance PMSM
drives,” IEEE Trans. Ind. Appl., vol. 38, no. 4, pp. 1018–
1026, Jul./Aug. 2002.
[7] D. F. Chen, T. H. Liu, and C. K. Hung, “Nonlinear
adaptive backstepping controller design for a matrix-
converter based PMSM control system,” in Conf. Rec.
˄a˅PI contro IEEE-IAS Annu. Meeting, 2003, vol. 1, pp. 673–678.
[8] S. Bolognani, L. Peretti, and M. Zigliotto, “Design and
implementation of model predictive control for electrical
motor drives,” IEEE Trans. Ind. Electron., vol. 56, no. 6,
pp. 1925–1936, Jun. 2009.
[9] Y. Yi, D. M. Vilathgamuwa, and M. A. Rahman,
“Implementation of an artificial-neural-network based
real-time adaptive controller for an interior permanent
magnet motor drive,” IEEE Trans. Ind. Appl., vol. 39, no.
1, pp. 96–104, Jan./Feb. 2003.
[10] M. A. Rahman, M. N. Uddin, and M. A. Abido, “An
artificial neural network for online tuning of a genetic
based PI controller for interior permanent magnet
synchronous motor drive,” Can. J. Elect. Comput. Eng.,
vol. 31, no. 3, pp. 159–165, Jul. 2006.
[11] Y. Chen, B. Yang, X. Gu, and S. Xing, “Novel fuzzy
˄b˅Fuzzy-PI double mode control control strategy of IPMSM drive system with voltage
FIG. 14. Comparison of three-phase current response when starting booster,” in Proc. 6th World Congr. Intell. Control
Autom., Jun. 21–23, 2006, vol. 2, pp. 8084–8087.
V. CONCLUSION [12] M. N. Uddin and M. A. Rahman, “High speed control of
In this paper, an improved current and speed controller based IPMSM drives using improved fuzzy logic algorithms,”
IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 190–199,
closed loop vector control of PMSM drive system has been proposed. Feb. 2007.
The dq-axis armature current was controlled optimally based on
[13] C. T. Lin and C. S. G. Lee, Neural Fuzzy Systems, A
integral separation strategy. The fuzzy-PI double mode controller Neuro-Fuzzy Synergism to Intelligent Systems.
was used in speed loop, and it helped to improve the system Englewood Cliffs, NJ: Prentice-Hall, 1996, pp. 533–607.
performance. A performance comparison of the fuzzy control based [14] K. Malekian and J. M. Monfared, “A genetic based fuzzy
PMSM drive system and conventional PI controller based drive logic controller for IPMSM drive over wide speed range,”
system was also provided in the simulation. Simulation results in Proc. IEMDC, May 3–5, 2007, vol. 1, pp. 847–853.
demonstrated the higher efficiency and better dynamic response of [15] Y. Chen, B. Yang, X. Gu, and S. Xing, “Novel fuzzy
the proposed method as compared with traditional method over a control strategy of IPMSM drive system with voltage
wide speed range. The complete drive system was also booster,” in Proc. 6th World Congr. Intell. Control
experimentally implemented and the experimental results also Autom., Jun. 21–23, 2006, vol. 2, pp. 8084–8087.
showed high efficiency and high dynamic performance of the [16] M. N. Uddin and M. A. Rahman, “High speed control of
proposed fuzzy control based PMSM drive system. IPMSM drives using improved fuzzy logic algorithms,”
IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 190–199,
REFERENCES Feb. 2007.

331

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on November 05,2024 at 06:27:13 UTC from IEEE Xplore. Restrictions apply.