Published by : International Journal of Engineering Research & Technology (IJERT)

http://www.ijert.org ISSN: 2278-0181

Vol. 7 Issue 04, April-2018

DC Motor Speed Control Using Machine Learning

Algorithm
Jeen Ann Abraham Sanjeev Shrivastava
Department of Electronics and Communication. Department of Electronics and Communication.
RKDF College of Engineering RKDF College of Engineering
Bhopal, India. Bhopal, India.

Abstract—The control and automation has been accelerated to The paper is structured as follows. Background and
its next generation with the introduction of machine learning methodology is explained in Section II and Section III followed
algorithms and advanced computing capability of modern by Tuning of PID controller in Section IV. The Section V
processors, in control systems for various applications. Many discusses about neural network predictive controller and
researches are undergoing in the field of controller design and simulation model followed by the comparison of results
tuning using the deep learning algorithms. This paper addresses a obtained from both the controllers in Section VI. Finally,
comparison study on effectiveness and performance between a
Section VII summarizes the paper and outlines future work.
proportional integral derivative controller and a neural network
controller which was designed using the existing tools and libraries
from Matlab/Simulink package. Initially, the proportional II. BACKGROUND
integral derivative controller was tuned for a DC motor model The purpose of this research is to improve the tolerance of
with a balance between robustness and faster response. Later, the controller output for changes in system parameters with respect
neural network controller was trained for the same model to time. The benefits of increasing the tolerance is that the
followed by optimization of the controller. Then both controllers controller can mitigate the disturbances in plant process to
were experimented using the same model of DC motor by certain extend. For example, in case of motor operated valve
simulation in Matlab/Simulink. Followed to this, another used in polymer application, the valve friction increases if it’s
experimentation was done by changing the values for moment of not operating for a certain period due to deposit of particles
inertia of rotor followed by simulating the controller response between seating area. This results in change in moment of
without re-tuning the controller for newer model. Result analysis inertia of rotor and the controller will not be able to give proper
was done by studying the response for both controller from the output because the tuned controller parameters correspond to
plotted graph and root mean square error calculation. Simulation the system without any friction. The PID controllers may
experiments and comparison between the controllers demonstrate
produce overshoots as the performance of system deteriorates
that the controller with machine learning algorithms is better in
control systems for a longer period.
due to change in various system parameters. Self-tuning control
strategies well suits for such system with time varying
Keywords— Machine Learning; Controller; Tuning; Neural characteristics. Deep learning controllers find its applications in
Network; Simulation controlling complex nonlinear processes because of the
approximation properties to predict precise output from the
I. INTRODUCTION
sample inputs. DC motors are widely used in various control
ANN (Artificial Neural Networks) are found in most of the applications as final control element. These applications vary
modern applications since last decade. Even though it was from robot arm control used in automation industry to feed
identified in 1943 by Warren S. Mcculloch and Walter Pitts [1], pump driver in rocket engines. In literature [7] the authors
the inception of deep learning algorithms and advanced proposed a deep learning controller for speed control of DC
computing capability makes suitable environment for the ANN motor which was designed by learning the PID controller. In
to find its place in modern world applications such as pattern this paper, a study was conducted for improving the
recognition system, speech recognition system, diagnosing performance and effectiveness of neural network controller to
disease in medical applications and stock market prediction. that proposed in [7] using MATLAB/ Simulink.
Nowadays, the application of neural networks starts to evade the
control and automation field with the help of AI (Artificial III. METHODOLOGY
Intelligence) and machine learning algorithms. An intelligent With graphical user interface, Neural network tool box and
control of wind turbine proposed by Mahmoud et al. in [2] several libraries, simulation models for deep learning
explains the use of AI in wind turbine control applications. applications can be developed using MATLAB/Simulink
Behind the mask, the machine learning maps the inputs to software package. The case study for control application was
outputs by exploring the correlation between both, followed by done by modelling a DC motor with voltage as input and speed
approximation. Despite several advanced techniques and as output followed by simulating the same using different
algorithms, the Proportional-Integral-Derivative (PID) control controllers in MATLAB/Simulink. The dynamic equations for
still has its solid space in control systems used in majority of a DC motor is as follows:
industrial applications because of its efficiency and
performance [3] [4]. However, the performance for nonlinear
systems varies and depends on the tuning of controllers.

IJERTV7IS040033 www.ijert.org 33
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
 Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 7 Issue 04, April-2018

The values for the DC motor parameters are shown in

Figure1. These values are kept same as discussed in literature
[7] for a comparison study on performance between both
controllers. The equation for continuous time PID controller
with first order derivative filter is as follows:

Where, 𝐽 is the moment of inertia of the rotor, 𝐾𝑡 is motor

torque constant, 𝑖 is the armature current, 𝑏 is motor viscous
friction constant, 𝐿 is electric inductance, 𝑅 is electric
resistance, and 𝐾𝑒 is electromotive force constant [7]. The
model is then simulated with a PID controller and Neural Where, Kp is the proportional gain, Ti and Td are the integral
Network Predictive Controller (NNPC). Before simulation the and derivative times and N is the first-order derivative filter
PID controller was tuned and NNPC was trained for the divisor. The PID controller is tuned for the DC motor and results
designed DC motor. Levenberg-Marquardt backpropagation shows that an increase in robustness degrades the performance
algorithm is faster and have good performance that makes it and vice versa. The system was tuned for better performance,
well suitable for training neural networks compared to resilient but the controller output starts oscillating for a transient period
back propagation algorithms [5][6]. Finally, the study on as shown in Figure 2. In Figure 3 the system was tuned with
tolerance of controller response for change in friction of motor increased robustness, but the result was a sluggish response. As
was done by increasing the moment of inertia and simulating a result, the plant was re-tuned to achieve a trade-off between
the model without re-tuning the controllers. The data analysis robustness and performance.
was performed by calculation of Root Mean Square Error
(RMSE) between set speed and motor output speed by both
controllers.

IV. PID CONTROLLER AND TUNING

PID controllers has only three parameters, but it is not easy
to find accurate values to achieve the system stability. There is
a general trade-off between the robustness and performance of
the control systems [8-9]. Although PID controller can be
implemented both parallel and series forms, the parallel forms
can achieve better performance for oscillatory processes [10].
Initially the DC motor is simulated with PID controller for
experimenting the performance versus robustness.

Fig. 2. PID controller output with rapid response.

Fig. 1. DC motor parameters.

Fig. 3. PID controller output for better stability.

IJERTV7IS040033 www.ijert.org 34
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
 Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 7 Issue 04, April-2018

Fig. 4. Training data.

Fig. 5. Validation data

V. NEURAL NETWORK PREDICTIVE CONTROLLER ym is the predicted output of the neural network and ρ is the
The algorithm for NNPC is based on minimizing the cost control weighing factor. The performance optimization is done
function over a finite period. The cost function is minimized to by reducing the size of the interval of uncertainty using golden
generate a control input that meets the constraints to track the section search method discussed in literature [12]. The plant
reference trajectory by the plant within certain tolerance [11]. model must be developed to predict the future outputs using
The equation for cost function is as per equation 4. neural networks. The neural network plant model has a total of
14 hidden layers with 2 delayed plant inputs and outputs each at
a sampling rate of 0.2. Finally, the controller is trained using
Levenberg-Marquardt backpropagation algorithm with an
iteration of 2000 times. These predictions are then used by the
optimization algorithm to predict controller output. Figure 4 and
Figure 5 shows the interpretation of data obtained from training
and validation of the Neural network plant model. The control
of DC motor with voltage V as input and angular speed 𝑤 as
output based on PID controller and neural network predictive
Where N1 and N2 are the minimum and maximum costing controller in MATLAB environment is shown in Figure 5.
horizon, Nu is the control horizon, yr is the reference trajectory,

Fig. 6. Simulink model for feedback control of DC motor based on neural network controller.

IJERTV7IS040033 www.ijert.org 35
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
 Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 7 Issue 04, April-2018

Fig. 7. Response from PID controller

Fig. 8. Response from neural network predictive controller

Fig. 9. Magnified view of output from both controllers

VI. SIMULATION RESULTS done because minor change may happen in motor parameters
A random reference block which generates normally eventually.
distributed random numbers serves as the setpoint variable for Figure 7 and Figure 8 shows the DC motor output for the PID
simulation. The model plots setpoint and output from the DC controller and the neural network predictive controller.
motor with reference to time for each controller. The scope
helps to compare the variation between the two controllers
output. The simulation was done in MATLAB/ Simulink
environment for a period of 100 seconds. Performance study
was conducted by comparing both the controller output for
overshoot and time to achieve system stability. Also, the RMSE
is calculated on output from the DC motor for each controller
and shown in Table 1. Another experiment was done by
changing the moment of inertia for the rotor by increasing and
decreasing the value by 10 percent after training the controller
to study the response of DC motor for each controller. This was

IJERTV7IS040033 www.ijert.org 36
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
 Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 7 Issue 04, April-2018

Fig. 10. Output from both controllers for increase in moment of inertia of rotor.

Further magnified view for the response of DC motor for both REFERENCES
controllers is shown in Figure 9. It was observed that the PID [1] Warren S. Mcculloch and Walter Pitts , “A logical calculus of the
controller output overshoots for each step change in input while ideas immanent in nervous activity,” Bulletin of Mathematical
the overshoot is negligible for the output from NNPC. On the Biophysics, Vol. 5, 1943, pp. 115-133.
other hand, accuracy for PID controller is higher than that of [2] Tawfek K. Mahmoud, Zhao Yang Dong and Jin Ma, “A developed
neural network controller. When the moment of inertia of rotor integrated scheme based approach for wind turbine intelligent
reduced by ten percent there was no large effect on controller control,” IEEE transactions on sustainable energy, Vol. 8, No. 3,
output, but the same for PID controller starts oscillating for ten July 2017.
percent increase in the moment of inertia and as a result system [3] G. K. McMillan and D. M. Considine, “Process/ industrial
becomes unstable. The NNPC shows the same output with instruments and controls handbook,” McGraw-Hill, New York,
minor delay. Figure 16 shows the response of both controllers 1999.
with ten percent increase in moment of inertia. [4] Ang K.H., Chong G., Li Y., PID Control System Analysis,
Design, and Technology. IEEE Trans. Control Syst. Technol.
TABLE I. Root Mean Square Error for both controllers. 2005, 13, pp. 559–576.
Method PID Controller Neural Network [5] Arnold Reynaldi, Samuel Lukas and
Controller Helena Margaretha, “Backpropagation and levenberg-marquardt
algorithm for training finite element neural network,” Proceedings
RMSE 8.64 x 10-5 7.21 x 10-4 of the 2012 sixth UKSIM/AMSS european symposium on
computer modeling and simulation, pp. 89-94.
[6] Kişi Özgür and Uncuoğlu Erdal, “Comparison of three back-
VII. CONCLUSION propagation training algorithms for two case studies,” Indian
The MATLAB based neural network controller and a PID journal of engineering and material sciences, vol 12 october 2005,
controller was designed for the control of DC motor model. A pp. 434-442.
comparison study was done on performance between both [7] Kangbeom Cheon, Jaehoon Kim, Moussa Hamadache, and
controllers and effect of change in inertia of rotor on the Dongik Lee “On replacing pid controller with deep learning
controller output to verify the application of machine learning controller for dc motor system,” Journal of automation and control
in control systems. From the simulation results, it has observed engineering Vol. 3, No. 6, December 2015, pp. 452-456.
that accuracy of NNPC is slightly less compared to PID [8] Emre Sariyildiz, Haoyong Yu and Kouhei Ohnishi “A practical
controller. At the same time NNPC has better control on dc tuning method for the robust pid controller with velocity feed-
motor with minor delay for ten percent increase in moment of back,” Machines Vol. 3, 2015, pp. 208-222.
inertia of rotor, where the PID controller fails and thereby, [9] K. Åström, K.J. and T. Hägglund, “PID controllers: theory,
explains that, there will not be any large change in controller design, and tuning,” Instrument society of America, Research
output due to reduction in performance of equipment for a triangle park, 1995.
[10] Skogestad S., “Simple analytical rules for model reduction and pid
neural network controller. The results demonstrate the
controller tuning,” Journal of process control, Vol. 13, 2003, pp.
feasibility and advantage of using machine learning algorithms 291309.
in control systems. The future work will be to improve the [11] Soloway D., and P. J. Haley, “Neural generalized predictive
preciseness of the above neural network controller by more control,” Proceedings of the 1996 IEEE international symposium
stable, robust machine learning algorithm with less time on intelligent control, 1996, pp. 277–281.
complexity and a real-time implementation of the controller in [12] Hagan, M.T., H.B. Demuth, and M.H. Beale, Neural Network
a DC motor speed control system. Design, Boston, MA: PWS Publishing, 1996.

IJERTV7IS040033 www.ijert.org 37
(This work is licensed under a Creative Commons Attribution 4.0 International License.)