Sliding Mode Speed Control of a Permanent

Magnet Synchronous Motor using Twelve Sectors

Modified DTC
Aritra Pal Partha Roy Susanta Ray
Dept. of Electrical Engineering Dept. of Electrical Engineering Dept. of Electrical Engineering
Jadavpur University Jadavpur University Jadavpur University
Kolkata – 700032, India Kolkata – 700032, India Kolkata – 700032, India
[email protected] [email protected] [email protected]

Abstract— This paper aims to design and realize twelve sectors can be overcome using different control techniques such as
of modified DTC drive, especially for Permanent Magnet Model Predictive Control (MPC) [3], Sliding Mode Control
Synchronous Motor (PMSM). Conventional drives follow six (SMC) [4], etc.
sectors and experience fluctuations in output responses.
Twelve-sector DTC drive topology reduces the torque ripple, Previously, classical DTC was applied in different
speed variation, current variation, etc. The Sliding Mode applications, selecting the voltage vector only in one
Control analogy controls the feedback error, which follows direction for six sector operations. Then, it was modified
variable structure control. A comparison between SMC and and the sectors were considered on both sides of the voltage
PI-controlled twelve sectors modified DTC is studied. Along vectors for easier operation and better responses. Finally,
with this, six sectors and twelve sectors modified DTC are
compared in the vicinity of the PI and SMC approaches.
those sectors were split into two parts, and twelve sectors of
modified DTC were introduced. Design and implementation
Keywords— Direct Torque Control (DTC), Permanent of the twelve sectors of DTC with the fundamental
Magnet Synchronous Motor (PMSM), Proportional-Integral (PI) controller as a PI controller is a common practice [5]. This
control, Sliding Mode Control (SMC), Torque ripple. paper proposes a twelve-sector DTC with a Sliding Mode
I. INTRODUCTION Controller (SMC) to control PMSM.
In recent years, Permanent Magnet Synchronous Motors The selection of the sectors to be used to control the
(PMSM) have had a wide range of applications in practical switching patterns in DTC has two types of control, such as
fields, especially in industries. This is a high-speed, efficient switching time calculation [3] and voltage vector selection
motor that has comparatively easier control techniques. The simply following the Look-Up Table (LUT) [5]. The torque
control law follows Variable Frequency Drive (VFD) in and flux ripples produced in DTC cause noise, vibration,
both high-speed and low-speed applications. Speed Control and power loss [1]. This issue can be mitigated by
of PMSM is done through different methodologies both in employing Sliding Mode Control (SMC), which is well
scalar and vector controls. There are different suited to nonlinear uncertainty, disturbances, and parameter
methodologies for the controllers, and the controllers could variations for faster dynamic response and robust control
be connected to the machine in series or cascades to achieve operation [2], [4], and [6].
better performance [1]. In this paper, DTC has been proposed in twelve sectors
Amongst the conventional control methods of PMSM, of operation and the ripples are minimized by SMC.
Direct Torque Control (DTC) is the most popular one. The Eventually, the responses are compared between the
DTC follows the torque and flux of the stator, and it proposed and existing schemes for different parameters with
controls the speed by controlling either the torque and flux different case studies.
or both simultaneously. It can vary the switching frequency The remaining part of this paper is organized as follows:
of the inverter by varying the torque and flux. Direct Torque Section II presents the dynamic model of PMSM with its
Control is best suited to the vector control strategies of respective equations. Section III follows two control
PMSM, as in DTC the flux of the stator field is controlled methodologies, DTC and SMC, for controlling PMSM.
despite the flux of the rotor field in the FOC. Essentially, the Sector IV is about to compare the twelve-sector modified
control of the flux of the stator field is an easier one. SMC-DTC with the twelve-sector modified PI-DTC. The
Another advantage of DTC is that speed sensors are not corresponding simulation results in the MATLAB
required in DTC. The only disadvantage is that it creates environment are accomplished to study different parameters
remarkable torque and flux ripples [1], [2]. These ripples for different case studies, especially ripple percentages.
Finally, this paper ends with Section V, which has a brief The equation for the electromagnetic torque [2] is
conclusion and proposes future works. 3p
T |ψ ||ψ | sin δ!
II. DYNAMIC MODEL OF PMSM 2L
/
The dynamic model of PMSM in the d-q reference frame So, ∆T |ψ ||ψ | ∆δ! (10)
can be represented as [7], :4

Here, L = Stator inductance.

V R i L ω L i (1)
V R i L ω L i ω ψ
Classical DTC was not sufficiently advantageous, so we
(2) should proceed with modified DTC [2] where the hysteresis
Here, V = D-axis voltage, V = Q-axis voltage, i = D-axis control is applied, depending on the values of error in torque
current, i = Q-axis current, R = Resistance of stator coil
and flux. The voltage vectors are selected according to the
per phase, ω = Electrical angular velocity of motor, L =
table for different sectors. Normally six sectors of DTC are
used. By introducing twelve sector DTC one can get a
D-axis inductance, L = Q-axis inductance, ψ = Rotor comparatively smoother operation and response with lesser
Flux. ripples. The LUT [5] needed for twelve sector operations is
The respective electromagnetic torque is as, given in Table I.

T p ψi L L i i
TABLE I. LOOK UP TABLE FOR TWELVE SECTORS MODIFIED
(3) DIRECT TORQUE CONTROL
Here, p = no. of pair of poles.
Hψ HTe S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12
The expression for the fluxes in the d-q axes is as,
ψ L i ψ 2 V2 V3 V3 V4 V4 V5 V5 V6 V6 V1 V1 V2
ψ ψ L i (4)
1 V2 V2 V3 V3 V4 V4 V5 V5 V6 V6 V1 V1
1
Where the respective stator flux is, -1 V1 V1 V2 V2 V3 V3 V4 V4 V5 V5 V6 V6

ψ ψ ! ψ " # V R I !dt (5) -2 V6 V1 V1 V2 V2 V3 V3 V4 V4 V5 V5 V6

Here, V = stator voltage, I = stator current. 2 V3 V4 V4 V5 V5 V6 V6 V1 V1 V2 V2 V3

,
And, the flux angle is, δ tan*+ ! (6) 1 V4 V4 V5 V5 V6 V6 V1 V1 V2 V2 V3 V3
, 0
The basic mechanical speed and rotation angle equations are -1 V7 V5 V0 V6 V7 V1 V0 V2 V7 V3 V0 V4
as [3], -2 V5 V6 V6 V1 V1 V2 V2 V3 V3 V4 V4 V5
- /
ω ω T T T012 ! (7)
. .
θ! ω
Where the hysteresis controllers of the flux and torque
(8)
are defined as: Torque Increase = 2, Torque Small Increase
Here, θ = Rotor Angle, B = Viscous Damping Coefficient, J = 1, Torque Decrease = -2, Torque Small Decrease = -1,
= Inertia Constant, T012 = Load Torque, T = Frictional Flux Increase = 1, Flux Decrease = 0. The input-output
Torque. relations of this stator flux and torque controller are shown
in Fig. 1(a) and 1(b) respectively.
III. CONTROL METHODOLOGIES
This section discusses simple overviews of the
topologies of Direct Torque Control (DTC) and Sliding
Mode Control (SMC).

A. Direct Torque Control Topology for PMSM

PMSM has several control mechanisms for both scalar
and vector control. DTC is one of the best control
techniques among vector control techniques. It could be a
sensorless operation, and it has a wide range of control
spans. The only drawback is the formation of torque and
current ripple, which can be minimized using other control Fig. 1. Hysteresis controllers of the (a) stator flux and (b) torque are
quantified
techniques such as PI, SMC, etc. along with fundamental
DTC. This method can control the speed by controlling both The responses obtained for the proposed method are
the torque and flux of the system. compared with state-of-the-art methods in Section IV, and
The equation for flux [8] is, the results are discussed. The results clearly show that
,4
torque ripples can be suppressed by using SMC instead of
V , this simply implies, ∆ψ V ∆t (9) PI for both six-sector and twelve-sector operations.
 B. Sliding Mode Control Topology for PMSM TABLE II. SYSTEM PARAMETERS

Sliding Mode Control (SMC) is a nonlinear control Parameters Symbol Values Unit
method that has remarkable performance in several fields of
control methodologies. This control technique is sufficiently No. of Pole Pairs p 2 -
robust and popular due to its comparatively accurate results
and comfortable tuning analogy. SMC has first-order as well Stator Resistance R 0.767 Ω

ψE
as higher-order approaches. The second-order approach is PM Flux (Peak) 0.1377 Wb
mostly used. In this paper, commonly known as the "Super-
twisting Algorithm", which is simply second-order SMC, is Q-axis Inductance L 4.607 mH
followed. SMC can be considered as a nonlinear variant of
the classical PI controller. It can also solve the chattering D-axis Inductance L 4.713 mH
issue while being used as a continuous function of time with
smooth approximation and good robustness properties as Rated Power P 0.4 kW

N
well. The proportional and integral controller gain is also
Rated Speed 1500 RPM
present in this controller like the classical PI controller.
Here, the gain values are tuned in a “trial and error” method Rated Current I 4.66 A
and the best values are adjusted to get the best response. The
value of the error signal is twisted due to the tuning and this Rated Torque T012 2.5 Nm
is why it is called the “Super-twisting Algorithm” [9], [10].
Inertia Constant J 6.876 ×10-3 Kg-m2
According to the second-order sliding mode control
principle the sliding variable of the error of flux magnitude, Rated Stator Flux ψ 0.3 Wb
S< ω∗ ω !. Based on a super-twisting algorithm, the
flux controller can be designed as, Viscous Damping B 2.7504 ×10-3 Nms

T∗ K / |S< | sgn S< ! T V

Input Voltage of
(11) G 100 V
Inverter

S< ! K sgn S< ! (12) Supply Frequency f 50 Hz

Here, K / = Proportional Gain, K = Integral Gain. As far as the speed ripple is concerned, it is less in
twelve-sector operations compared to six-sector while using
And the final expression becomes, a PI controller. It is shown in Fig. 3. In Fig. 4, the same is
T∗ K / |S< | sgn S< ! # K sgn S< ! dt (13) compared using an SMC.

Here, r 0.5 has been used in this system for good tuning
and control operation [4], [6].
IV. RESULTS AND DISCUSSIONS
This paper proposes direct toque control in twelve
operational sectors, with ripples attempted to be minimized
using SMC. The block diagram of the prosed scheme is
depicted in Fig. 2.
A complete model of a three-phase PMSM with an
inverter has been developed in the MATLAB® SIMULINK
environment. A 400W real-life motor with parameters given
in Table II is taken for simulation. PI Controller and SMC
for six and twelve sectors are employed as controllers.

Fig. 2. Block Diagram of Sliding Mode Control based Direct Torque Fig. 3. Simulation Result of Speed Response of PI-DTC (For both six
Control of Permanent Magnet Synchronous Motor Drive System and twelve sectors)
 Fig. 6. Simulation Result of Torque Response of SMC-DTC (For both six
and twelve sectors)

Similar responses for stator flux are shown in Fig. 7 and Fig.
8 respectively.
Fig. 4. Simulation Result of Speed Response of SMC-DTC (For both six
and twelve sectors)

The advantages of DTC exploited in those electric drives

were not only the speed but the torque to be controlled. For
this application, DTC is a very good option as it can provide
high dynamic performance with convenient costs. However,
the classical DTC has some drawbacks, and one of these is
the significant torque and current ripple generated in steady-
state operation.
Torque ripple is an important parameter, especially in
the case of direct torque control. It is also compared
between six-sector and twelve-sector operations, as well as
PI and SMC. The results are shown in Fig. 5 and Fig. 6
respectively.
Fig. 7. Simulation Result of Stator Flux Response of PI-DTC (For both six
and twelve sectors)

Fig. 5. Simulation Result of Torque Response of PI-DTC (For both six and
twelve sectors) Fig. 8. Simulation Result of Stator Flux Response of SMC-DTC (For both
six and twelve sectors)
 TABLE III. COMPARISONS OF RIPPLE PERCENTAGES OF THE V. CONCLUSION
RESPONSE OF THE PARAMETERS
This paper proposes twelve sectors SMC-DTC of PMSM to
With PI Controller With SMC
Ripple control the speed to obtain a ripple-free response. And the
Percentage Six Twelve Six Twelve comparison between the twelve sectors SMC-DTC of
(%) Sectors Sectors Sectors Sectors PMSM and six sectors SMC-DTC of PMSM is
accomplished. The important formulas, tables, and diagrams
Speed 2.07 0.78 0.53 0.48
are provided for each section, such as PMSM, DTC, SMC,
71.75 40.59 64.41 38.81 etc. The simulation results obtained from MATLAB
Torque
SIMULINK are shown, and the comparison of the ripple in
Stator Flux 9.1 5.25 9.01 5.6 the response for each parameter is mentioned. Where ripple-
free operation is required, this method can be a better choice
than the existing methods. This work can be validated in a
The speed ripple for the case of six sectors PI-DTC is hardware setup. This research work could be implemented
2.07% and for the twelve sectors operation, the ripple where torque ripples matter such as in Medical Instruments,
(0.78%) is less. Also, for six sectors, SMC-DTC, it is less, Belt Conveyors, Rolling Mills etc. Thus, the researchers get
i.e., 0.53%. But, for the proposed scheme of twelve sectors interested to work in this field. Also, it has several scopes of
SMC-DTC, the speed ripple is 0.48%, which is the least applications in industry and other fields like Aircraft,
among all the cases. Though our purpose is to reduce torque Robotics, Washing Machines, Electric Vehicles etc.
ripple, this is also a good achievement for the proposed
scheme as the speed response is the best among all the cases REFERENCES
mentioned here. [1] Y. Wang, L. Geng, W. Hao, and W. Xiao, "Control Method for
Optimal Dynamic Performance of DTC-Based PMSM Drives," in
Twelve sectors of SMC-DTC, i.e., the proposed scheme, IEEE Transactions on Energy Conversion, vol. 33, no. 3, pp. 1285-
can produce the least ripple for the torque ripple, 38.81%. 1296, Sept. 2018, doi:10.1109/TEC.2018.2794527.
For the other three cases, that means, for six sectors PI-DTC [2] L. Meng and X. Yang, "Comparative analysis of direct torque control
(71.75%), twelve sectors PI-DTC (40.59%), and six sectors and DTC based on sliding mode control for PMSM drive," 29th
Chinese Control and Decision Conference (CCDC), 2017, pp. 736-741,
SMC-DTC (64.41%), it is more. This means that if we use doi: 10.1109/CCDC.2017.7978529.
twelve sector operation in the case of SMC, the torque [3] M. Preindl and S. Bolognani, "Model Predictive Direct Torque Control
ripple will be the least so our purpose is solved. with Finite Control Set for PMSM Drive Systems, Part 2: Field
Weakening Operation," in IEEE Transactions on Industrial
If the flux ripple percentages are observed, we can see Informatics, vol. 9, no. 2, pp. 648-657, May 2013,
that when we proceed from six sectors operation to twelve doi:10.1109/TII.2012.2220353.
[4] Arie Levant, “Sliding order and sliding accuracy in sliding mode
sectors operation in the case of PI-DTC, flux ripples are control”, in International Journal of Control, 58(6), 1993, 1247-1263).
going to be decreased, i.e., from 9.1% to 9.01%. But for [5] A. Al-Quteimat, A. Roccaforte and U. Schäfer, "Performance
SMC DTC, it is the opposite, i.e., 5.25% to 5.6%. But this improvement of direct torque control for doubly fed induction
will not affect the system significantly. Except for the generator with 12 sector methodology," 2016 IEEE International
Conference on Renewable Energy Research and Applications
twelve sectors of PI-DTC, the proposed scheme's flux ripple (ICRERA), 2016, pp. 242-246, doi: 10.1109/ICRERA.2016.7884545.
is less than the six sectors of operation for both PI and SMC. [6] G. Bartolini, A. Ferrara, A. Levant, E. Usai, “On second order sliding
Still, the ripple is less in the proposed scheme than in the mode controller” in Variable structure systems, sliding mode and
first case, so it is permitted in the proposed scheme where nonlinear control, Springer Lecture Notes in Control and Information
Sciences, Volume 247/1999.
the main purpose of reducing torque ripple has been served [7] L. Xu, G. Chen and Q. Li, "Ultra-Local Model-Free Predictive Current
properly. Control Based on Nonlinear Disturbance Compensation for Permanent
Magnet Synchronous Motor," in IEEE Access, vol. 8, pp. 127690-
Here, the responses are given and compared for all four 127699, 2020, doi: 10.1109/ACCESS.2020.3008158.
cases. The ripple percentages are provided in the tables for [8] P. C. Krause, O. Wasynczuk, S. D. Sudhoff, Analysis of electric
each parameter. Here, speed and torque ripples are the machinery, in IEEE Press, 1995.
smallest for the proposed scheme. But the flux ripple is [9] A. Chalanga, S. Kamal, L. M. Fridman, B. Bandyopadhyay, and J. A.
Moreno, "Implementation of Super-Twisting Control: Super-Twisting
comparatively less using the proposed methodology, but not and Higher Order Sliding-Mode Observer-Based Approaches," in
least. Though it can be reduced using the proposed scheme IEEE Transactions on Industrial Electronics, vol. 63, no. 6, pp. 3677-
compared to the first case. So, this paper wisely justifies that 3685, June 2016, doi: 10.1109/TIE.2016.2523913.
the twelve sectors SMC-DTC method is far superior to the [10] Y. Yan, S. Yu, and X. Yu, "Quantized super-twisting algorithm based
sliding mode control", Automatica, vol. 105, pp. 43-48, Jul. 2019.
other methods mentioned here to control PMSM.