International Journal of Power Electronics and Drive Systems (IJPEDS)

Vol. 12, No. 4, December 2021, pp. 2070~2082

ISSN: 2088-8694, DOI: 10.11591/ijpeds.v12.i4.pp2070-2082  2070

Artificial intelligent techniques based on direct torque control of

induction machines

Mona Magdy, Salama Abu-Zaid, Mahmoud A. Elwany

Department of Electrical Engineering, Al-Azhar University, Cairo, Egypt

Article Info ABSTRACT

Article history: The direct torque control (DTC) system, which is based on induction
machine drive is a developed and simple control method. It allows high
Received Jan 15, 2021 dynamic performance with very simple hysteresis control scheme; However,
Revised Sep 9, 2021 its disadvantages are high current, torque, and flux ripple. In this paper, DTC
Accepted Sep 23, 2021 of induction machine drive has been improved by using the applications of
artificial intelligence (AI) approaches to reduce the current, torque, and flux
ripples and also get better performance of the machines. At the conclusion of
Keywords: this study, the outcomes of traditional DTC and artificial intelligent methods
are compared. By the program MATLAB/Simulink, the modeling and
Artificial neural networks simulation results of the DTC system for induction machine (IM) have been
Direct torque control proposed.
Fuzzy logic control
Induction motor drive
Vector control This is an open access article under the CC BY-SA license.

Corresponding Author:
Mona Magdy
Departement of Electrical Engineering
Al-Azhar University
Tanta, Al-Gharbiyah, Egypt
Email: [email protected]

1. INTRODUCTION
Electrical engines are one of the most pertinent components in electricity industry [1]. Induction
Motors are generally utilized in industries due to low cost, less maintenance and hardness, so it plays a
significant job in the world. There are two fundamental sorts of induction motors which are the squirrel-cage
and the wounded rotor. Both of them are used, However Squirrel-cage is considered more dependable and
less expensive to make [2].
Speed control is a significant issue because it is constantly attempted to reach to the attractive
reaction in short time. Changing pole numbers and voltage control. Some of these techniques are presently
out of utilization. Nowadays, vector control or field-oriented control (FOC) is used to solve the issue of
controlling the speed of motor by changing the voltage and the frequency [3].
The most famous induction motor drive control strategy is the FOC or (vector control). FOC
empowers induction motor control similarly to separately excited direct current (DC) motors [4]. Vector
control is more specified than scalar control since the controlling of torque and flux is decoupled, so it
utilized in high-performance drives [5]. The principle advantages of vector control are high dynamic
performance, lower torque ripple, and switching frequency. Vector control have disadvantages that are
needed for two co-ordinate transformations, current controllers, and rise machine parameter instability [6].
To play out the vector control method, it's important to break down the stator current phase into two
symmetrical parts in a synchronously rotating reference edge and this requires two transformations, Clarke
and Park transformation.

Journal homepage: http://ijpeds.iaescore.com

Int J Pow Elec & Dri Syst ISSN: 2088-8694  2071

Vector control have two techniques for induction motors, one of them is the direct field oriented and
the other is the indirect field Oriented [7]. More than 10 years back, DTC was acquainted with give a quick
and great unique torque performance and it can be viewed as just an option in contrast to such the FOC
method. The design of DTC considered is a lot less complex than the system of FOC because of none of the
frame transformer. It additionally needn't a pulse width modulator or a position encoder [8]-[10].
The artificial intelligence (AI) techniques are equipped for learning the ideal mapping between the
framework's inputs and outputs signals without the need for understanding the framework's definite
numerical model. It can be considered as rearranged numerical models of mind-like systems and they work
as parallel circulated registering systems [11], [12]. AI techniques, like expert-system (ES), fuzzy logic (FL),
artificial neural network (ANN or NNW), and genetic algorithms (GA) have as of late been used broadly
responsible for induction motor drives. AI appears to have more noteworthy effect on power gadgets,
electronics of the engine, and motor drives region which is obvious through the productions in the writing.
AI-based induction machine drives having gotten more noteworthy consideration, Since the century 1990 s.
Apart from the existing control techniques, intelligent control strategies, for example, FL control, GA, NN
control, and ES demonstrated to be unrivaled. AI controller may be the better controller for induction
machines [13]-[18].

2. PRINCIPLE OF DIRECT TORQUE CONTROL

Figure 1 illustrates a diagram of conventional DTC of induction motor drive, where (φs*, Te*) are
the reference values of flux and torque, (φs, Te) are estimated the values of stator flux and torque. “θs is the
stator flux angle, ω is the real speed value, and ω* is the command speed value”. Method of DTC can be
performed where two hysteresis controllers can be used (one for torque and another for flux), estimators of
torque and flux, and a switching table can be used to determine the suitable voltage vector [19]. The inverter-
fed machine's torque is controlled by hysteresis control of torque size and stator flux, which chooses one of
six active voltage vectors and two zero voltage inverters [20]. Switching table for selecting voltage vectors
using for choosing rapid torque reaction, low inverter switching frequency, and low harmonic losses without
the complicated field orientation by restricting flux and torque errors with the optimum range inside the
corresponding flux and torque hysteresis regions [19]. Figure 2 shows the diagram of PI controller used to
calculate the reference torque due to the difference between the motor's instantaneous speed and the
reference.
DTC concept depends on choosing the optimum voltage vector to spin the flux and generate the
needed torque. In a voltage feed three phases, the switching commands from every inverter leg are
complementary. So, by each leg a logic condition Ci (i=a, b, c) may be created, with1indicating that the
upper switch is commanded to be closed and 0 indicating that the lower one is commanded to be closed [21] .

Figure 1. Graph of the DTC technique

Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2072  ISSN: 2088-8694

Figure 2. A speed PI controller diagram

Voltage vectors can be expressed by the following equation:

𝑗2𝜋 𝑗4𝜋
2
𝑉𝑠 = √ 𝑈0 [𝐶1 + 𝐶2 𝑒 3 + 𝐶3 𝑒 3 ] (1)
3

Per (1), there are eight switching regions: one of them is two zero voltage vectors and the other is six
non-zero voltage vectors, as shown in Figure 3. Figure 3 shows the torque is explained by a quadrature axis
and the stator flux is explained by a direct axis. As can be seen, choosing a suitable voltage vector results in
preserve the torque and stator flux inside acceptable bounds.
Takahashi recommended the Table 1 for switching of conventional DTC. Feedback torque and flux
are determined of the voltages and currents of the machine terminal and the block for signal calculation
computes the number of the sector where the flux vector is now located. There are six effective voltage
vectors every spanning. The voltage vector table takes the flux value, torque value, and section number and
uses a look-up table to provide suitable control for the inverter, as shown in Table 1 [21].

Figure 3. The dq level is divided into six

Table 1. Conventional DTC switching table sections angular

Sector
1 2 3 4 5 6
HTe Hψ
1 1 V1 V6 V5 V4 V3 V2
0 V7 V0 V7 V0 V7 V0
-1 V5 V4 V3 V2 V1 V6
1 -1 V2 V1 V6 V5 V4 V3
0 V7 V0 V7 V0 V7 V0
-1 V4 V3 V2 V1 V6 V5

3. DIRECT OF TORQUE CONTROL USING FUZZY LOGIC

3.1. Principles of fuzzy torque control
A fuzzy logic is presented in this section to decrease torque ripple while improve motor
performance. The fuzzy controller, which replaces the switching table in the conventional DTC, achieves this
goal. A diagram of the FL- DTC method of the induction motor is show in Figure 4. Generally, fuzzification,

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2073

fuzzy reasoning, and defuzzification are the three major components of a fuzzy controller. A membership
function is used to do the fuzzification. Fuzzy controller's effectiveness is based on both the membership
function shape and the fuzzy reasoning rules. A functional diagram using controller of fuzzy logic is shown
in Figure 5.

Figure 4. DTC diagram using controller of fuzzy logic

The flux error of stator, electromagnetic torque error, and the stator flux angle are the three input variables,
while the stator voltage vector is output. The input and output variables are called fuzzy varibles.

3.2. Fuzzy control rules

Figure 5 shown how decisions are made based on the inputs and the rule basisq. The rules for the
control approach used for the system to be controlled are entered into a data base by a Fuzzy Logic Control
designer. The fuzzy logic DTC rules are applied to the twelve sectors DTC.

Figure 5. A functional diagram using controller of fuzzy logic

4. ARTIFICIAL NEURAL NETWORKS

4.1. Artificial neural networks principles
Artificial neural networks (ANNs) are machines that mimic the human brain's learning and
generalization abilities. Despite not understanding the system's exact mathematical model, the ANNs can

Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2074  ISSN: 2088-8694

discover the optimal mapping between the system's outputs and input signals [22]. A neuron's mathematical
model is given by:

y = ϕ (∑𝑁
𝑖=1 Wi ∗ Xi + b) (2)

Where (x1, x2 ... xN) are the neuron input signals, the corresponding weights are (w1, w2, ... wN), and the
bias parameter is b. A tangent sigmoid function is Φ, and y is the neuron's output signal. The neurons that
match to the computational nodes are the essential components of an ANN. Every node uses constant weights
to multiply its input signals, summarizes the result and maps the summition to a nonlinear function; then the
result is transferred to its output and integrated an activation function is as illustrated in Figure 6. The ANN is
illustrated in Figure 7, it may be trained to perform a particular role by changing the values of the
interconnections (weights) between elements [23]. Usually, neural networks are modified or educated so that
a given input contributes to a desired output [24].

Figure 6. Representation of the artificial neuron

Figure 7. Structure of neural network

4.2. ANN structure for DTC

In direct torque control, the artificial neural network may be employed in two methods:
- To achive the desired response, replace the standard PI speed controller. A feed forward network with
three hidden and one output layer, as well as one input, is provided. The transfer functions for the buried

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2075

layers are 'log-sigmoid,' 'tan-sigmoid,' and 'pure line.' MATLAB simulink is used to train the Levenberg-
Marquardt Back-propagation algorithm with 20 neurons in every hidden layer.
- The switching table selector component and the two hysteresis controllers are replaced. The ANN inputs,
the difference between the estimated electromagnetic torque, the reference torque, and the position of the
flux stator vector specified by the associated sector number, all contribute to the inaccuracy between the
estimated flux value and its reference value. Three neurons make up the ANN output layer. Every neuron
represents one of the three vector pairs that will be used to drive the induction motor [25], [26]. Figure 8
shows the design of DTC using ANN techniques as a switching state selector. The rest of the system is
identical to DTC's conventional structure, as seen in Figure 1.

Figure 8. Design of DTC using ANN techniques

5. SIMULATION MODEL AND RESULTS

5.1. Model and parameters
Figure 9 shows the DTC simulink model for induction motor drive. The model was created using the
blocks from the MATLAB program. Table 2 shows the parameters of the induction motor. After running
models, the comparison between ripple values shows in Table 3.

Figure 9. DTC model for induction motor drive in Simulink

Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2076  ISSN: 2088-8694

Table 2. Parameters for I.M

Variable Symbol Value (unit)
Rated voltage P 220 V
Rated power V 2238 VA
Rated frequancy F 50 HZ
Pole pairs P ‫ــــ‬
Resistance of stator Rs 0.435 Ω
Inductance of stator Ls 2 mH
Resistance of rotor Rr 0.816 Ω
Inductance of rotor Lr 2 mH
Mutual inductance Lm 69.31 mH

Table 3. Comparison between ripple values for DTC methods

Tourque Flux Current
Switching Table for CDTC 0.45 0.03 0.47
Fuzy Logic Control 0.25 0.01 0.41
ANN Replace Conventional PI 0.35 0.029 0.4
ANN Replace Switching Table 0.4 0.028 0.4

5.2. Electromagnetic torque results

5.2.1. Rotor speed results
Figures 10-13 show the results of rotor speed for DTC. Figure 10 using classical, Figure 11 by using
fuzzy logic. ANN shows in Figure 12 by replacing PI and in Figure 13 by replacing the switching table.

Figure 10. Rotor speed for the classical DTC

Figure 11. Rotor speed for DTC using fuzzy logic

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2077

Figure 12. Rotor speed for DTC using ANN by replacing conventional PI

Figure 13. Rotor Speed for DTC using and replace PI ANN by replacing switching table

5.2.2. Electromagnetic torque results

Figures 14-17 show the results of electromagnetic torque for DTC. Figure 14 using classical,
Figure 15 by using fuzzy logic. ANN shows in Figure 16 by replacing PI and in figure 17 by replacing the
switching table.

Figure 14. Electromagnetic Torque for the classical DTC

Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2078  ISSN: 2088-8694

Figure 15. Electromagnetic Torque for DTC using fuzzy logic

Figure 16. Electromagnetic Torque for DTC using ANN by replacing conventional PI

Figure 17. Electromagnetic Torque for DTC using ANN by replacing switching table

5.2.3. Stator current results

Figures 18-21 show the results of stator current for DTC. Figure 18 using classical, Figure 19 by
using fuzzy logic. ANN shows in Figure 20 by replacing PI and in Figure 21 by replacing the switching table.

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2079

Figure 18. Stator current for the classical DTC

Figure 19. Stator current for the DTC using fuzzy logic

Figure 20. Stator Current for DTC using ANN by replacing conventional PI

Figure 21. Stator Current for DTC using ANN by replacing switching table
Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2080  ISSN: 2088-8694

5.2.4. Stator flux circle results

Figures 22-25 show the results of stator flux circle for DTC. Figure 22 using classical, Figure 23 by
using fuzzy logic. ANN shows in Figure 24 by replacing PI and in Figure 25 by replacing the switching table.

Figure 22. Stator flux circle for the classical DTC Figure 23. Stator flux circle for DTC using fuzzy
logic

Figure 24. Stator flux circle for the DTC using ANN Figure 25. Stator flux circle for DTC using ANN by
by replacing conventional PI replacing switching table

6. CONCLUSION
Due to an importance method of the DTC for the IM drives its great performance and ease, so
several strategies have been created aims to access the best results for IM. In this paper, to improve the DTC
of IM, it's using the applications of AI techniques to reduce the ripples of current, torque, and flux. After
testing the system with all three controllers (conventional DTC, FLC, and ANN) by using software
MATLAB/Simulink.The simulation show the following results; 1) Fast dynamic response for different
mentioned methods. 2) High torque ripple for conventional DTC. 3) ANN can be applied on DTC by: a) the
switching table selector block, as well as the two hysteresis controllers, must be replaced. b) the conventional
PI controller is being replaced. ANN results in the two methods are close to each other, it's considered beter
than the conventional DTC. 4) It's noted that the FLC gives better responses from than the conventional DTC
and the artificial neural networks. All results can be noted for all the controllers as shown in Table 3.

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2081

ACKNOWLEDGEMENTS
In the the beginning, I'd to thanks my God for all blessings giving to me. Second, Thanks for my
teacheres they are the best mentors to me. Dr Salama Abu-Zaid he is a fantastic personality and he is guide
me in this paper. And too, I'd like to thanks my parents, brothers, husband, my son, and my friends for
helping me.

REFERENCES
[1] H. Arabaci, and O. Bilgin, “Squirrel Cage of Induction Motors Simulation via Simulink,” vol. 2, no. 3, p. 324-327,
June 2012
[2] A. M. Ali, S. Z. Abo, and A. H. Ahmed, “Modified Single Phase Induction Motor Fed from The Solar Energy,”
Journal of Engineering and Applied Sciences, vol. 13, no. 16, pp. 6904-6909, October 2018, doi:
10.36478/jeasci.2018.6904.6909.
[3] K. Gottapu, Y. V. Prashanth, P. Mahesh, Y. Sumith, and P. S. Kiran, “Simulation of DTC IM Based on PI &
Artificial Neural Network Technique,” International Journal of Advanced Research in Electrical, Electronics and
Instrumentation Engineering, vol. 2, no. 7, pp. 3340-3348, July 2013.
[4] Ch. Chengaiah and I. P. Siva, ‘‘Performance of Induction Motor Drive by Indirect Vector Controlled Method Using
Pi and Fuzzy Controllers,’’ International Journal of Science, Environment and Technology, vol. 2, no. 3, pp. 457-
469, 2013.
[5] K. Parker, “Topic 13: Vector Control of AC Induction,” ECE Electric Drives, April 2004.
[6] V. Kumar, and S. Rao, “Modified Direct Torque Control of Three-Phase Induction Motor Drives with Low Ripple
in Flux and Torque,” Leonardo Journal of Sciences, no. 18, pp. 27-44, June 2011
[7] B. K. Bose, “Power Electronics and Motor Drives,” Elsevier, 2010.
[8] I. Takahashi, and T. Noguchi, “A New Quick-Response and High-Efficiency Control Strategy of an Induction
Motor,” in IEEE Transactions on Industry Applications, vol. IA-22, no. 5, pp. 820-827, September 1986, doi:
10.1109/TIA.1986.4504799.
[9] P. Tiitinen, and M. Surandra, “The next generation motor control method, DTC direct torque control,” Proceedings
of International Conference on Power Electronics, Drives and Energy Systems for Industrial Growth, vol. 1, pp.
37-43, 1996, doi: 10.1109/PEDES.1996.537279.
[10] J.-Q. Yang, and J. Huang, “Direct torque control system for induction motors with fuzzy speed PI regulator,” 2005
International Conference on Machine Learning and Cybernetics, vol. 2, pp. 778-783, 2005, doi:
10.1109/ICMLC.2005.1527049.
[11] M. K. Eldin, and S. Abo-Zaid, “Resistance Estimation for Sensor less Permanent Magnet Synchronous Motor
Drives,” International Journal of Engineering and Innovative Technology (IJEIT), vol. 5, no. 6, pp. 8-12,
December 2015, doi: 10.17605/OSF.IO/D9VJG.
[12] R. Fuller, and R. Fullér, ‘‘Introduction to Neuro-Fuzzy Systems,’’ Springer Science & Business Media, 2000, doi:
10.1 007/978-3-7908-1852-9.
[13] K. Shi, T. F. Chan, and Y. K. Wong, “Modelling of the three-phase induction motor using SIMULINK,”
Conference: Electric Machines and Drives Conference Record, 1997. IEEE International, vol. 36, pp. 163-172,
June 1997, doi: 10.1109/IEMDC.1997.604326.
[14] T. C. Fun and K. Shi, Applied Intelligent Control of Induction Motor Drives, John Wiley & Sons, January 2011.
[15] P. Krause, O. Wasynczuk, S. D. Sudhoff, and S. Pekarek, “Analysis of Electric Machinery and Drive Systems,”
John Wiley & Sons, vol 75, 2013.
[16] N. Mohan, Advanced Electric Drives, John Wiley & Sons, 2014
[17] P. M. Menghal, and A. J. Laxmi, “Adaptive Neuro Fuzzy based dynamic simulation of induction motor drives,” 2013
IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), 2013, pp. 1-8, doi: 10.1109/FUZZ-IEEE.2013.6622452.
[18] P. M. Menghal, A. J. Laxmi, and N. Mukhesh, “Dynamic simulation of induction motor drive using artificial
intelligent controller,” Proceedings of the 2014 International Conference on Control, Instrumentation, Energy and
Communication (CIEC), 2014, pp. 301-305, doi: 10.1109/CIEC.2014.6959098.
[19] R. S. Soni, and S S. Dharmal, “Direct Torque Control of Three Phase Induction Motor Using Fuzzy Logic,”
International Journal of Engineering Trends and Technology (IJETT), vol. 6, no. 3, pp. 161-167, Dec. 2013.
[20] M. Koshy, and M. A Lekshmi, “Direct Torque Control Schemes for Induction Motor,” Jan. 2011.
[21] R. Toufouti, S. Meziane, and H. Benalla, “Direct Torque Control Strategy of Induction Motors,” Acta
Electrotechnica et Informatica, vol. 7, no. 1, pp. 1-7, July 2007.
[22] V. M. V. Rao, and A. Anand Kumar, “Artificial Neural Network and Adaptive Neuro Fuzzy Control of Direct
Torque Control of Induction Motor for Speed and Torque Ripple Control,” 2018 2nd International Conference on
Trends in Electronics and Informatics (ICOEI), 2018, pp. 1416-1422, doi: 10.1109/ICOEI.2018.8553871.
[23] A. Moualdia, D. Boudana, O. Bouchhida, and P. Wira, “Direct torque control based multi-level inverter and
artificial neural networks of wind energy conversion system,” 2016 8th International Conference on Modelling,
Identification and Control (ICMIC), 2016, pp. 49-54, doi: 10.1109/ICMIC.2016.7804199.
[24] M. H. Beale, M. T. Hagan, and H. B. Demuth, ‘‘Neural Network Toolbox™User’s Guide,’’ The MathWorks Inc,
vol. 103, 1992.

Artificial intelligent techniques based on direct torque control of induction machines (Mona Magdy Ragab)
2082  ISSN: 2088-8694

[25] M. Hamouda, S. Abo-Zaid, and Abd E. Kotb, “Sensor less DTC Using Artificial Intelligent Technique as Switching
Vector Selector,” vol. 7, no. 6, pp. 220-228, June 2016, doi: 10.13140/RG.2.2.22013.87526.
[26] A. Massoum, A. Meroufel, and Y. Djeriri, “Artificial Neural Network Based Direct Torque Control of Doubly Fed
Induction Generator,” vol. 14, no. 2, p. 8, June 2014.

BIOGRAPHIES OF AUTHORS

Mona Magdy Ragab was born in 6 june 1992 in Tanta (Egypt). She graduated from the
Faculty of Engineering, Al Azhar University, Cairo, Egypt, Department of at Electrical
Power and Machine, in 2015.

Salama Abu-Zaid, He is associated Professor, Head of Department of Electrical Power

and Machine, Faculty of Engineering (Girls), Al Azhar University, Cairo, Egypt.

Mahmoud A. Elwany, He is aProfessor, Department of Electrical Machine, Faculty of

Engineering (Girls), Al Azhar University, Cairo, Egypt. He was Former Dean of Girls
Engineering from 2015 to 2019.

Int J Pow Elec & Dri Syst, Vol. 12, No. 4, December 2021 : 2070 – 2082