Control of The Permanent Magnet Synchronous Motor Using Model Reference Dynamic Inversion
Control of The Permanent Magnet Synchronous Motor Using Model Reference Dynamic Inversion
Control of The Permanent Magnet Synchronous Motor Using Model Reference Dynamic Inversion
Abstract: Worldwide energy-saving emission has stimulated extensive application of permanent magnet synchronous
motor in industry. This work is a contribution to velocity control of the permanent magnet synchronous motor. The
model of the permanent magnet synchronous motor has multi-variable, highly nonlinear, strong coupling character
with external load; in order to control this complicated nonlinear model, the hierarchy model reference dynamic
inversion control method has been developed. Based on the band of the different variables of the system, the control
system can be divided into long period control loop and short period control loop. The model reference dynamic
inversion method is proposed as the kernel controller in different loops. Based on the desired closed loop reference
model, the open loop controller can be designed for the long period loop and short period loop respectively, and this
open loop controller can make the response of the system the same as that of the reference model. Comparing with the
traditional control method, this model reference dynamic inversion control method combine the virtues of the model
reference controller with that of the dynamic inversion controller, and this method is a fully nonlinear control method
in essence, which can avoid linearization error when using the approximate linearization controller and the problem of
the parameter tuning when utilizing the traditional PI hierarchy controller. Finally, the simulation experiments have
been set up and the simulation results have illustrated that this control method can control the permanent magnet
synchronous motor stably and successfully.
Key words: Permanent Magnet Synchronous Motor (PMSM), Dynamic Inversion (DI), Velocity Tracking Control,
model reference, hierarchy controller
techniques such as vector control [5] or direct torque open loop controller, which making the real reaction
control (DTC) [6] are standard for this type of drives. tracing that of the desired reference model. In order to
Based on the characteristics of the model of use not only the virtue of the directly nonlinear control
permanent magnet synchronous motor, many modern but also that of the model reference control, the model
control methods and intelligent control methods have reference dynamic inversion control method is utilized
been applied to the permanent magnet synchronous to control the PMSM.
motor, such as nonlinear control [7,8] and sliding mode This paper is arranged as follows: In section 2 the
control[9] have been developed to address the speed and model of the PM synchronous servo motor is introduced;
position control of PMSMs. The state feedback In section 3, the kernel of the system controller, model
linearization of nonlinear systems control theory has reference dynamic inversion, is described in detail; the
been introduced into the motor control in [10]. The PMSM controller based on the hierarchy model
adaptive control strategy has been used in the control of reference dynamic inversion control method is designed
permanent magnet synchronous motor in [11]. Besides in section 4; In section 5, the control effects and
these nonlinear control methods mentioned above, the performance of the hierarchy model reference dynamic
intelligent control methods, such as the artificial inversion controller on PMSM are evaluated under
intelligence expert system [12], fuzzy control [13], and different loads with simulink in Matlab; Finally,
neural networks [14, 15] have also been utilized in the conclusions are summarized in section 6.
motor drive system, and the great progress has been
achieved. Brock [16] used fuzzy logic controllers to
adjust the gain of a controller in a sliding mode controller
for speed and position control of PMSM. The control
2. Mathematical model of PMSM
strategies in which recurrent fuzzy neural networks A PMSM is composed of three phases stator windings
(FNN) were used to adjust the gain of the SMC for and permanent magnets mounted on the rotor surface
position control of a PMSM was used by Wai [17]. (surface mounted PMSM) or buried inside the rotor
The purpose of this paper is to address the (interior PMSM). The electrical equations of the PM
application of a relatively new control system design synchronous motor can be described in the rotor rotating
approach known as model reference dynamic inversion reference frame, written in the (dq) rotor flux reference
to design the speed servo controller for high frame [22].
performance PMSMs. Dynamic inversion is the The mathematic model of PMSM is based on the
nonlinear control method in essence and has been following assumptions:
successfully applied to large flexible complicated (1) Neglecting the saturation of armature;
system in documents. There has been a considerable (2) Neglecting the wastages of eddy and magnetic
amount of work in the application of dynamic inversion hysterics;
to control helicopters [18] and other aircraft [19]. (3) There is no rotor damp resistance.
Dynamic inversion method has proven to be a very The relations of voltage, torque and flux of PMSM
effective technique for the control of the nonlinear are described as follows:
system. The model reference adaptive control has also
L 1
been successfully used in many systems [20, 21]; with r 0 0 uq
iq R iq L
= + u d (1)
L id
this control method, the response of the reference model
id 1
to the input signal is the same as the reaction of the 0 r
r R L L
controlled system with the controller acted upon; In
order to fulfill this function, the minus of the output of
where id and iq are the d and q axis stator currents,
the response of desired model reference and the output
of the reaction of the real system is utilized to adjust the R and L are the stator phase resistance and inductance
If id = 0 , the electromagnetic torque Te is proportional functions of the system state vector x .If g ( x ) is
state when x 0 , the control input can be given as The preceding derivation suggests that any given system
with appropriate properties can have the desired
u 0 = g 1 ( x ) f ( x ) (9)
dynamic response. The application of dynamic inversion
The second part is v , which can de designed as control method requires that the control matrix g be
invertible. The properties of the control law will depend
v(t ) = g 1 ( x) [ x ]des (10)
on the properties of the control matrix. Obtaining an
Then, the whole control input is then taken as [19] invertible control matrix with suitable properties
depends on careful choices of the models to be used and
u (t ) = v(t ) + u0 (t ) (11)
the states to be fed back.
Letting As for the high order system, such as the PMSM, the
hierarchy control method, with model reference
[ x ]des = w (12)
dynamic inversion as the kernel controller in each layer,
Combine with equation (8); using the Laplace can be utilized to fulfill such function. The approach is
transform, the equation can be derived below to apply dynamic inversion separately to the low
frequency and high frequency dynamics of the system to
sX ( s ) = W ( s ) (13)
be controlled. The similar approach was demonstrated
successfully in the controller for the X-38[24].
If the W ( s ) is taken as
The proposed structure is shown in Fig. 1. d is the
W( s ) = K ( s )[ X des ( s ) X ( s )] (14)
desired velocity given by the engineer and the idcmd and
Then combining with equation (10), the equation can
be given
the iqcmd are the desired currents of d-axis and q-axis
sX ( s ) = K ( s )[ X des ( s ) X ( s )] (15)
respectively. The structure is similar in form to that used
According to the equation (15), the transfer function in classical control system design process, such as PI
of the close loop is controller in the PMSM controller [25]. There is,
however, a notable difference that the control laws in
X ( s) K ( s)
D( s )= = (16) dynamic inversion blocks are decidedly nonlinear.
X des ( s ) sI + K ( s )
In both the cases of the long period and short period
Then if the transfer function of the desired close loop dynamics, the dynamics to be inverted will depend on
both mechanical dynamics and electrical model of the
reference dynamic model is chosen as D ( s ) , then the
PMSM.
open controller of the system could be selected as iqcmd
d uq
1
K ( s ) = sD( s )[ I D ( s )] (17)
iq
When designing the open loop controller, the r idcmd
properties and dynamics of the resulting closed loop
ud
system can be selected according to the control request
id
given by the application engineer.
Synchronous Motor
method of design the short period controller of q-axis In the first simulation, the desired velocity is
control voltage is the same as that used in d-axis. 700rad/s, while desired current of the d-axis is 0A and the
Discussion: This controller is the combination of the load is kept none all the time. Fig. 2 shows the velocity
hierarchy controller, the model reference controller and tracing process of the PMSM without load. The solid line
the dynamic inversion controller; it is superior to the (-) represents the desired velocity while the dot line (--) is
common used hierarchy PID controller in the PMSM the real-time simulation velocity. With the controller
control in that this method is nonlinear controller, and the designed in last section, it can be seen that real-time
controller and the parameters are achieved directly from velocity arrived at the desired velocity very fast and
the desired reference model. If the desired reference remain the desired velocity in no-load situation without
model is given, then controller will be designed any vibration. Fig. 3 and fig. 4 show the control current of
according to the reference model and the controller will the q-axis and d-axis separately. The d-axis current stays
make the system trace desired trajectory the same as the at zero all the time, while the q-axis current increases at
reference model does. the start process and keep static at almost 0.08A when the
velocity of the PMSM arrives at the desired velocity, this
current is no zero due to the resistance of the stator. In the
no-load simulation, the result proves the validity of the
controller.
5. Simulation Experiments 1000
Dv
In this section, the proposed approach above has been Rv
600
control scheme, using MATLAB/Simulink. The
parameters of the PMSM are given in Tab. I. The
400
simulation results are shown in Figs. 2-12.
The dynamic performance of the control system is 200
evaluated by three simulation experiments using the step
speed of 700 rad/sec as desired input and three kinds of 0
0 5 10 15 20 25 30 35 40 45 50
different torques as load. time (s)
Tab. 1 PMSM Parameters Fig.2 the process of the velocity tracing of the PMSM without
load
Machine parameters value
0.5
Resistance of the stator windings( ) 2.875
0.4
Number of pole pairs 4
q-ax is current (A )
Combined viscous friction of rotor and 0 Fig. 3 the changing process of the current of the q-axis of the
PMSM without load
load B(N.m.s/rad)
-14
x 10 1000
5
Dv
Rv
800
d-axis current (A)
V e lo c ity (ra d /s )
600
0
400
200
-5
0 5 10 15 20 25 30 35 40 45 50
time (s)
The second simulation is to study and test the validity Fig.6 the process of the velocity tracing of the PMSM with a
of the controller with abrupt plus step load. In the second plus step load
simulation experiment, the desired velocity is 700rad/s, 0.35
while desired current of the d-axis is 0A and a step load is
0.3
acted upon the axis of the PMSM at one point after the
PMSM has arrived at its desired velocity. Fig. 5 shows a 0.25
q -a x is c u rre n t (A )
equivalent current in a few seconds; the current of d-axis Fig.7 the changing process of the current of the q-axis of the
will remain zero all the time. This simulation experiment PMSM with a plus step load
have verified that this controller can control the PMSM x 10
-14
8
with abrupt plus load.
6
0.3
4
0.25
d-ax is c urre nt (A )
2
0.2
T o rq u e (N .m )
0.15 -2
-4
0.1
-6
0.05
-8
0 5 10 15 20 25 30 35 40 45 50
0 time (s)
0 5 10 15 20 25 30 35 40 45 50
time (s) Fig. 8 the changing process of the current of the d-axis of the
Fig.5 the plus step load acted upon the PMSM in simulation 2 PMSM with a plus step load
meet with the load which moves upward and downward Fig. 9 the load series acted upon the PMSM in simulation 3
all the time. Thus, this simulation is carried out to 1000
Dv
guarantee the motor can keep the desired velocity with Rv
800
abrupt changing load. In the third simulation experiment,
the desired velocity is 700rad/s, while desired current of
v e lo c ity (ra d /s )
600
the d-axis is 0A,the load will have some abrupt plus and
minus steps. In the simulation process, the load acted
400
upon the PMSM is in series which will increase from
0N.m to 0.2N.m at 10s, and then remain static for 15s 200
and then will increase to 0.3N.m and will remain at
0.3N.m for another 15s, and then will decrease to 0
0 5 10 15 20 25 30 35 40 45 50
0.2N.m and then remain 0.2N.m, which is shown in fig. time (s)
9.It is can be noticed in fig.10 that the real-time velocity Fig. 10 the process of the velocity tracing of the PMSM with
can trace the desired velocity very well. The solid line (-) a series of changing load
represents the desired velocity while the dot line (--) is 0.4
the real-time simulation velocity. In fig.10, when the 0.35
load is increasing, the velocity will drop a little and will
0.3
return soon, and when the load is decreasing, the
q -a x is c u rre n t(A )
velocity will increase a little and will return soon. This 0.25
coefficients of the out-loop controller or by changing the Fig. 11 the changing process of the current of the q-axis of the
desired close loop reference model. Fig.12 shows the PMSM with a series of changing load
change process of the current the d-axis. It can be seen x 10
-13
2
that the current of d-axis will keep zero all the time. The
third simulation experiment shows that with the up-step 1.5
0.5
0.35
0
0.3
-0.5
0.25
-1
Torque (N .m )
0.2
-1.5
0.15
-2
0.1 0 5 10 15 20 25 30 35 40 45 50
time (s)
0.05
Fig. 12 the changing process of the current of the d-axis of the
0
0 5 10 15 20 25 30 35 40 45 50
PMSM with a series of changing load
time (s)
From the simulations above, it is noticeable that with short period or inner loop. The control performance of
the controller designed in the last section, the PMSM can the hierarchy model reference dynamic inversion
keep desired velocity all the time, only when the load has controller appeared to be superior in all respects to the
an abrupt change, the real-time velocity will have a little classical design for its high nonlinear character brought
overshoot and will return the desired velocity very soon. by dynamic inversion controller and it has the virtues of
The control current of the q-axis will change along with the model reference controller. Three simulation
the load. The control current of the d-axis will accurately experiments have been done to test the controller with
at the desired value zero. The simulations above have different load. The results show that the controller can
guaranteed that the controller proposed above can control control the velocity at desired value in different
the PMSM and can be used quickly in the reality. For occasions. These simulation results guaranty the right
decoupling control performance of the PMSM is much and validity of the controller designed. From the
less sensitive to the parameter variations [3,4], when the descriptions above, it can be seen that hierarchy model
accurate parameters of the PMSM are given and the reference dynamic inversion can be a viable approach for
desired close loop reference model which is decided by the control system design for PMSM.This compound
the desired control performance is achieved, the method can be a preferable control method for the
controller designed above can be used in controlling the complicated multi-input and multi-output (MIMO)
PMSM. system.
Acknowledgements
6. Conclusions The authors would like to thank The Ministry of Science
In order to meet with the worldwide demands for and Technology of the Peoples Republic of China for
energy-saving emission, in many areas, permanent her financial support in project: 2007BAF19B04
magnet synchronous motor is gradually replacing the
traditional three-phase asynchronous motor. This motor
is a complex multi-variable, nonlinear, strong coupling of References
the multiple-input multiple-output system. It is difficult [1]. Slemon, G. R. Electrical machines for
to design the controller. Therefore, the study of the variable-frequency drives. Proceeding of the IEEE [J],
control strategy is currently a major problem in the 1994, 82(8):1123-1139.
industry. [2].Cetin E., Oguz U.A hybrid controller for the speed
According to the model of the PMSM, the hierarchy control of a permanent magnet synchronous motor
model reference dynamic inversion speed controller for drive[J],Control Engineering Practice, 2008,16
PMSM has been designed. Dynamic inversion is the (3):260-270
nonlinear control method which is based directly on the [3]. Sen, P. C. Electric motor drives and control-past,
nonlinear model, which can keep the high fidelity present, and future[J], IEEE Transactions on Industrial
nonlinear information of the nonlinear dynamics and Electronics, 1990,37(1),562575.
electrical equation. The model reference method will [4]. Tzann-Shin Lee, Chih-Hong Lin, Faa-Jeng Lin. An
make the system have the desired response. Compared to adaptive H controller design for permanent magnet
classical approaches, this compound method provided a synchronous motor drives[J], Control Engineering
much more systematic approach for control law design. Practice,2005,13 (4): 425-439
Based on the model of the PMSM, the two time scale [5]. L. Tang, L. Zhong, M.F. Rahman, Y. Hu, A novel
approach was utilized in the control structure. The direct torque control for interior permanent-magnet
velocity was controlled in the long period or outer loop synchronous machine drive with low ripple in torque
controller and the current controller was utilized in the
and flux a speed-sensorless approach[J], IEEE Neural Network[C], IEEE International Symposium on
Transaction Industry Application, 2003,(39):17481756. Industrial Electronics, 2007:30-35
[6]. J. Rodrguez, J. Pontt, C. Silva, etc. Predictive [16]. Brock S., Deskur, J. ,Zawirski, K. Robust speed
Current Control of a Voltage Source Inverter[J], IEEE and position control of PMSM[C]. In Proceedings of the
Transaction Industry Electronics,2007,54:495503 IEEE International Symposium on Industrial Electronics.
[7]. Baik, I. C., Kim, K. H. and Youn, M. J. Robust 2003, 2: 667672, Bled-Slovenija.
nonlinear speed control of PM synchronous motor using [17]. Wai, R. J. Total sliding-mode controller for PM
boundary layer integral sliding mode control technique. synchronous servo motor drive using recurrent fuzzy
IEEE Transactions on Control Systems Technology [J], neural network[J]. IEEE Transactions on Industrial
2000, 8(1): 47-54. Electronics, 48(5), 926944.
[8]. Solsona, J., Valla, M. I., Muravchik, C. Nonlinear [18].Du, J.F. Kondak, K. Zhang, Y.O.etc.Modelling and
control of a permanent magnet synchronous motor with control of a small-scale unmanned helicopter[J],
disturbance torque estimation [J], IEEE Transactions on Proceedings of the Institution of Mechanical Engineers,
Energy Conversion, 2000, 15(2):163-168. Part I: Journal of Systems and Control Engineering, 2008,
[9]. Paponpen, K., Konghirun, M., An improved sliding 222(6): 481-492
mode observer for speed sensorless vector control drive [19]. Alfred C. Watts. Control of a High Performance
of PMSM[C].CES/IEEE 5th International Power Maneuvering Reentry Vehicle Using Dynamic
Electronics and Motion Control Conference, Shanghai, Inversion[C],AIAA Guidance, Navigation, and Control
China,2007,2,972-976 Conference and Exhibit, 2005, San Francisco,
[10]. Hemici, B.; Nezli, L.; Tadjine, M.; Boucherit, M.S. California,USA:1-16
Robust PID/backstepping control design for permanent [20]. Uchiyama,N.,Mori, A. Model reference control for
magnet synchronous motor drive[J].Control and human-operated robots and application to a robotic
Intelligent Systems,2006,34(3):194-204 manipulator[J], Proceedings of the Institution of
[11]. Lee Tzann-Shin, Lin Chih-Hong, Lin Faa-Jeng. An Mechanical Engineers, Part I: Journal of Systems and
Control Engineering, 2009, 223(8):1163-1170
adaptive H controller design for permanent magnet
[21]. Ohtake, H.,Tanaka, K., Wang, H. Fuzzy
synchronous motor drives [J]. Control Engineering model-based servo and model following control for
Practice, 2005, 13(4):425-439. nonlinear systems[J], IEEE Transactions on Systems,
[12]. Elmas Cetin, Ustun Oguz,Sayan Hasan H. A Man and Cybernetics, Part B: Cybernetics,
neuro-fuzzy controller for speed control of a permanent 2009,39(6):1634-1639
magnet synchronous motor drive [J]. Expert Systems [22]. Florent Morel, Xuefang Lin-Shi, Jean-Marie Retif
with Applications, 2008, 34(1):657-664. etc.A predictive current control applied to a permanent
[13].Cheng Kuang-Yao, Tzou Ying-Yu. Fuzzy magnet synchronous machine, comparison with a
optimization techniques applied to the design of a digital classical direct torque control [J],Electric Power Systems
PMSM servo drive [J].IEEE Transactions on Power Research.2008, 78(8):1437-1447
Electronics, 2004, 19(4):1085-1099. [23].Guchuan Zhu, Louis-A.Dessaint, Ouassima
[14]. Rahman, M. A., Hoque, M. A. On-line adaptive Akhrif,etc.Speed Tracking Control of a
artificial neural network based vector control of Permanent-Magnet Synchronous Motor with State and
permanent magnet synchronous motors [J], IEEE Load Torque Observer[J],IEEE Transactions on
Transactions on Energy Conversion, 1998, Industrial Electronics, 2000, 47(2):346-355
13(4):311-318. [24].Wacker, R., Enns, D., Bugajski, D.etc.X-38
[15]. Rajesh Kumar, R. A. Gupta and Ajay Kr. B. application of dynamic inversion flight
Identification and Control of PMSM Using Artificial control[J],Advances in the Astronautical Sciences, 2001,
107:233-250
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: