Design of Intelligent Air Conditioner Controller
using Fuzzy Logic
S. Ahsan ur Rehman Omer
HITEC University Taxila, Pakistan
Email: ahsan403403@gmail.com
Abstract— The paper presents the algorithm for intelligent air
conditioner controller. This algorithm is based on fuzzy logic.
The implementation is done on arduino based microcontroller
with temperature and humidity as two inputs and power as
output. Air conditioners based on such controller are more
energy efficient as compared to conventional air conditioners.
The experimental results have also shown the superiority of such
controller.
Keywords—Energy efficient control, fuzzy inference system,
temperature and humidity sensor, triangular membership
function
I. INTRODUCTION
Air conditioning systems are major electrical loads in
houses, auditorium halls, indoor stadiums and other such
buildings. Conventional air conditioners consume constant
wattage when they are operating [1]. When the desired
temperature reaches, the outside unit turned off. There are two
states of operation; either on or off. The operating states can be
increased using fuzzy logic for reducing the consumption of
electricity without affecting air conditioning process. This
paper focuses on the development of air conditioner controller
using fuzzy logic.
In fuzzy logic, the output of the system is evaluated based
on the system’s input/inputs. The inputs are fuzzified using
triangular membership function. Based on these values
fuzzified output is produced using fuzzy rules. The fuzzified
output is converted to crisp output by ‘center average using
singleton value’ defuzzification method [2].
The focus of this research is to develop an intelligent air
conditioning system. If the temperature is very hot, the motor
of air conditioner compressor should run at high speed and as
the temperature lowers the compressor motor should run at
lower speeds. This process continues until the desired
temperature and humidity is reached. At this stage the
compressor will be consuming a little electricity for
maintaining the desired temperature and humidity and hence
saving the electricity.
978-1-5090-3310-2/17/$31.00 ©2017 IEEE
Ejaz Muhammad
HITEC University Taxila, Pakistan
Email: dean.qac@hitecuni.edu.pk
The air conditioning process can be controlled using
various methods. Some of these methods are on/off control,
neural network control, fuzzy control, robust control and many
more. A detailed review of these control methods is given by
Afram et al [3].
Shumit Saha et al developed a simulation based model of
an air conditioning system using fuzzy logic. They defined
two input variables namely temperature and humidity and the
output is the compressor motor speed. Gaussian membership
functions have been used for fuzzification of inputs and output
with eighteen fuzzy rules. If the room temperature reaches the
desired level, the fuzzy based controller switches off the
compressor motor and hence reduces the energy consumption
[4]. Vikas Kumar et al developed fuzzy logic based controller
for maintaining temperature, humidity and other hygienic
conditions as desired for operating rooms (operation theaters)
[5]. They used temperature, humidity, oxygen level and
number of particles in the room as input parameters. The
speed of air conditioner motor and speed of exhaust motor
were used as output parameters. Air conditioners control,
using fuzzy logic with two input parameters, is also developed
by Yordanaova et al [6].
The Rest of this paper is organized as follows: section II
describes the designed algorithm and simulations done on
matlab. Section III explains the experiments done using
various electronic and electrical components. Results are
discussed in section IV. Section V concludes the paper.
Section VI discusses triangular membership function.
II. ALGORITHM AND SIMULATIONS
The process of air conditioning control is shown in figure
1. In this figure, set point is the user defined temperature and
humidity. Controller output is power to be supplied to the
compressor of air conditioner.
An algorithm based on fuzzy logic is developed for
experimentation purpose [7]. Flow chart depicting the detailed
algorithm is shown in figure 2.

Fig. 1 Air conditioning control process
Fig. 2 Flowchart of proposed algorithm
Simulations of the proposed controller is done on Matlab
using FIS (Fuzzy Inference System) toolbox. ‘Temperature’
and ‘humidity’ are the inputs of the system and ‘power’ to be
supplied to compressor of air conditioner is the output. The
ranges of ‘temperature’, ‘humidity’ and ‘power’ are ‘-5 ˚C to
55 ˚C’, ‘0 to 100 % relative humidity (RH)’ and ‘0 to 250
watts’ respectively. Temperature range of ‘-5 ˚C to 55 ˚C’ is
divided to five ranges for evaluation of membership grade
values of five triangular membership functions. The five
triangular membership functions depicting ‘VCold’, ‘Cold’
‘Warm’ ‘Hot’ and ‘VHot’ temperatures are shown in figure 1.
In this figure, x-axis shows the variable and y-axis depicts the
membership function values. The ranges of ‘VCold’, ‘Cold’
‘Warm’, ‘Hot’ and ‘VHot’ are ‘-5 to 10 ˚C’, ‘-5 to 25 ˚C’, ‘10
to 40 ˚C’, ‘25 to 55 ˚C’ and ‘40 to 55 ˚C’ respectively.
Triangular membership function is elaborated in appendix.
Similarly, relative humidity (RH) is divided to five triangular
membership functions as represented in figure 2. These
functions are represented by ‘VDry’, ‘Dry’, ‘Normal’, ‘Wet’
and ‘VWet’. The ranges of these membership functions are ‘0
to 25 percent RH’, ‘0 to 50 percent RH’, ‘25 to 75 percent
RH’, ‘50 to 100 percent RH’ and ’75 to 100 percent RH’
respectively.
Fig. 1 Membership functions for temperature
Fig. 2 Membership functions for relative humidity
 Fuzzy control make use of rules, in the form of sentences, 22. If (temperature is VHot) and (humidity is Dry) then
to control a process [8]. The number of fuzzy rules depend on (power is Normal)
possible combinations of membership functions of inputs. As
there are five membership functions for temperature and five 23. If (temperature is VHot) and (humidity is Normal) then
for humidity; twenty five fuzzy rules will be sufficient. These (power is High)
combinations of input membership functions will determine the 24. If (temperature is VHot) and (humidity is Wet) then
appropriate rule/rules to be fired. These rules are: (power is VHigh)
1. If (temperature is VCold) and (humidity is VDry) then 25. If (temperature is VHot) and (humidity is VWet) then
(power is High) (power is VHigh)
2. If (temperature is VCold) and (humidity is Dry) then These rules will yield a membership function of output
(power is High) (power). The power range of 0 to 250 watts is divided to five
3. If (temperature is VCold) and (humidity is Normal) triangular membership functions named as ‘VLow’, ‘Low’,
then (power is VHigh) ‘Normal’ ‘High’ and ‘VHigh’. The membership grade value of
output function, known as fuzzified output, is determined using
4. If (temperature is VCold) and (humidity is Wet) then ‘AND’ fuzzy relation of membership functions of two inputs
(power is VHigh) [9]. Centroid method of defuzzification is used to convert
fuzzified output to crisp value.
5. If (temperature is VCold) and (humidity is VWet) then
(power is VHigh)
6. If (temperature is Cold) and (humidity is VDry) then
(power is Normal)
7. If (temperature is Cold) and (humidity is Dry) then
(power is High)
8. If (temperature is Cold) and (humidity is Normal) then
(power is High)
9. If (temperature is Cold) and (humidity is Wet) then
(power is VHigh)
10. If (temperature is Cold) and (humidity is VWet) then
(power is VHigh)
11. If (temperature is Warm) and (humidity is VDry) then Fig. 3 Output membership functions
(power is VLow) Some results of simulations are discussed here. The power
12. If (temperature is Warm) and (humidity is Dry) then is 50 Watts when temperature is set at 25 ˚C and humidity at
(power is VLow) 50 percent relative humidity. If temperature is 35 ˚C and
relative humidity is 32 %; the output is 152 watts. When
13. If (temperature is Warm) and (humidity is Normal) simulation is performed while setting temperature 37.1 ˚C and
then (power is VLow) relative humidity 60.5 %; the output is 200 watts. These
14. If (temperature is Warm) and (humidity is Wet) then results show that the output power increases as temperature
(power is Low) and relative humidity rises.
15. If (temperature is Warm) and (humidity is VWet) then
(power is Normal) III. DESIGN METHODOLOGY AND EXPERIMENTATION
16. If (temperature is Hot) and (humidity is VDry) then Arduino coding, based on fuzzy logic, has been developed
(power is Normal) to perform experiments using microcontroller [10] [11].
17. If (temperature is Hot) and (humidity is Dry) then The experiments are performed using microcontroller,
(power is Normal) sensor for temperature and humidity measurement, an
alternating current motor and a specially designed hardware
18. If (temperature is Hot) and (humidity is Normal) then for controlling alternating current. The digital sensor is used to
(power is High) detect temperature and humidity of the room/hall. These
19. If (temperature is Hot) and (humidity is Wet) then readings are fuzzified, using triangular membership function,
(power is High) in the microcontroller. Fuzzy rules have been hardcoded in the
microcontroller. The output is produced using these rules. This
20. If (temperature is Hot) and (humidity is VWet) then
is the fuzzified output. Centre average method of
(power is VHigh)
defuzzification is employed to find the crisp value of output.
21. If (temperature is VHot) and (humidity is VDry) then This output is converted to direct current (DC) voltage using
(power is Low) pulse width modulation of microcontroller [12]. The DC
voltage is converted to alternating current (AC) voltage using consumption of electricity. These results have been
opto-couplers, bilateral triode thyristor and full wave rectifier. demonstrated by simulation model and experimental model of
This AC voltage is to be supplied to the compressor of air air conditioner.
conditioner. Arduino is isolated from this high voltage using
optical-isolators (opto-isolator).
In this experimental setup, an induction motor of 250 watts VI. APPENDIX
is used for comparing the results with ‘Power’ being generated Triangular membership functions can be realized by three
by ‘matlab simulations’. Table 1 depicts comparison of results parameters (a, b, m). The parameters a, b and m are lower
being produced by experiments and matlab simulations. Table boundary, upper boundary and peak of this membership
2 shows the comparison of results being generated by function respectively. is the function value at any value of
emulating ‘temperature’ and ‘humidity’ and simulations.
independent variable .
Table 1 Comparison of results
Temp Humidity Simulation Experiment Percentage
(˚C) (% RH) Results Results Error
(Watts) (Watts)
29 32 50 52 0.01
30 32 51.8 52 0.00
35 32 152 128 9.6
43 24 127 133 2.4
47 24 126 133 2.8

Table 2 Comparison of results while emulating temperature

and humidity
Temp Humidity Simulation Experiment Percentage
(˚C) (% RH) Results Results Error
(Watts) (Watts)
0 20 126 118.6 2.9
10 30 150 140 4.0
20 20 41 32 3.6
30 30 50 45.5 1.8
40 20 127 125.5 0.01
50 40 166 150 6.4 REFERENCES
25 50 50 48 0.01 [1] "Home". PEL. N.p., 2016. Web. 19 Nov. 2016.
37.1 60.5 212 236 9.6 [2] Kozlova, L. P., and O. A. Kozlova. "Using of fuzzy set theory
when designing technical systems." Soft Computing and
Measurements (SCM), 2015 XVIII International Conference on.
IV. RESULTS IEEE, 2015.
[3] Afram, Abdul, and Farrokh Janabi-Sharifi. "Theory and
The results obtained from experimentation and simulations applications of HVAC control systems–A review of model predictive
demonstrate that there is a potential in saving electricity using control (MPC)." Building and Environment 72 (2014): 343-355.
fuzzy logic control for controlling air conditioners. Such [4] Saha, Shumit, et al. "Designing of an air-conditioning system
microcontrollers enable variable power supply to compressor using fuzzy logic with advantage of energy saving." Informatics,
of air conditioner. In contrast, conventional air conditioners Electronics & Vision (ICIEV), 2014 International Conference on.
IEEE, 2014.
use constant wattage when they are operating.
[5] Kumar, Vikas, Santosh Kumar, and Harshit Kansal. "Fuzzy logic
Percentage error is in the acceptable range and is controller based operating room air condition control system."
negligible in the most operating ranges of temperature and International Journal of Innovative Research in Electrical,
humidity. Electronics, Instrumentation and Control Engineering, 2014
[6] Yordanova, Snejana, Daniel Merazchiev, and Lakhmi Jain. "A
two-variable fuzzy control design with application to an air-
V. CONCLUSION conditioning system." IEEE Transactions on Fuzzy Systems 23.2
Conserving electricity is very important for developing and (2015): 474-481.
developed countries. Intelligent air conditioners can be [7] Kumari, Rajani, Sandeep Kumar, and Vivek Kumar Sharma. "Air
conditioning system with fuzzy logic and neuro-fuzzy
developed, by incorporating fuzzy logic, for saving electricity.
algorithm." Proceedings of the Second International Conference on
Such air conditioners are very energy efficient and are capable
to maintain desired temperature and humidity at lower
Soft Computing for Problem Solving (SocProS 2012), December 28-
30, 2012. Springer India, 2014.
[8] Jantzen, Jan. Foundations of Fuzzy Control: A Practical
Approach. John Wiley & Sons, 2013.
[9] Wang, Li-Xin. A Course in Fuzzy Systems. Prentice-Hall press,
USA, 1999.
[10] Driankov, Dimiter, Hans Hellendoorn, and Michael
Reinfrank. An introduction to Fuzzy Control. Springer Science &
Business Media, 2013.
[11] Du, Mingfang, et al. "Design of a new practical expert fuzzy
controller in central air conditioning control system." Computational
Intelligence and Industrial Application, 2008. PACIIA'08. Pacific-
Asia Workshop on. Vol. 2. IEEE, 2008.
[12] Akbar, M. Amirullah. "Simulation of fuzzy logic control for DC
servomotor using Arduino based on MATLAB/Simulink." Intelligent
Autonomous Agents, Networks and Systems (INAGENTSYS), 2014
IEEE International Conference on. IEEE, 2014.