A Review of Intelligent Control Techniques in HVAC Systems Hossein Mirinejad, Karla Conn Welch, Member, IEEE, and Lucas Spicer, Student Member, IEEE
comfort Abstract-- Two main objectives in the control of Heating, Ventilating and Air Conditioning (HVAC) systems are increasing thermal comfort and reducing energy consumption. Achieving these purposes requires a suitable control system design. In this paper, a thorough review of intelligent control techniques used in HVAC systems to date is completed. Such an overview provides
is
inherently
subjective
since
all
humans
have
different definitions for their comfort [2]. Because
of
the
problems
stated
above,
including the
impossibility of developing an accurate mathematical HVAC model, the necessity of using multi-criteria control in HV AC systems,
and
subjectivity
of
thermal
comfort;
intelligent
an insight into artificial intelligence methods for the control of
control strategies are a promising alternative for achieving
HVAC systems and can offer scholars and HVAC learners
superior
comprehensive information about a variety of soft computing
conventional control methods.
techniques in the field of HVAC. This information can in turn allow for improved designs of a proper controller for their work.
Index Terms-- fuzzy control, HVAC, neural network, thermal comfort
results
in
HVAC
applications
compared
to
The paper is organized as follows: In the next section, different types of intelligent control methods for HV AC systems are introduced. In particular, the combination of neural
networks
and
evolutionary
algorithms
with
fuzzy
systems are explained. In Section III, a variety of intelligent control techniques used in the area of HV AC systems are
I. INTRODUCTION
A
FTER the energy crisis in the 1970s, energy conservation has
been
considered
as
a
major
parameter
in
all
buildings. Based on surveys, the energy consumption in the
HVAC equipment in all residential, commercial, and industrial
discussed; and finally, a proper solution for designing the controller for HV AC systems is concluded. II. INTELLIGENT CONTROL METHODS FOR HVAC SYSTEMS
buildings constitutes about 40% to 50% of the world's energy
Recently, many studies have explored the use of intelligent
consumption [1]-[6]. Thus, in recent years, many techniques
and soft computing methods in the application area of HV AC
have been considered for reducing the energy consumption in HVAC systems. In comparison with conventional controllers (e.g., On-Off and PID controllers), intelligent controllers can notably thermal
save
energy
comfort
to
in
buildings
occupants
while
providing
simultaneously,
more
thereby
achieving better performance in the two major objectives of HVAC systems. In order to successfully control HVAC systems, their unique features and characteristics must be taken into account. In fact, an HV AC system is a complex, nonlinear, multi-input multi-output (MIMO) system with interrelated variables (air temperature,
relative
humidity,
air
velocity,
etc.)
and
is
exposed to various disturbances and uncertainties (external air temperature, occupants' activities, etc.); HVAC systems also have different time lags and inertia which are as inherent part of all thermal systems [2]. Therefore, it is a challenging task to find a mathematical model to accurately describe the process over a wide operating range [2]. Also, different criteria and parameters
like
variable
air
volume
and
controlled
air
temperature need to be considered for the control of HVAC systems [4]. Furthermore, it should be noted that thermal
systems.
Since
the
inputs
and
outputs
of
Fuzzy
Logic
Controllers (FLCs) are real variables mapped with a nonlinear function,
they
are
appropriate
for
various
engineering
problems especially for complex problems where classical control
methods
do
not
achieve
comparatively-favorable
results [4]. The main advantage of FLCs as compared to conventional controllers resides in the fact that no mathematical modeling is required for the design of the controller [2] ,[5], [7]. The essential part of a fuzzy controller is a Knowledge Base
(KE).
The KB is comprised of if-then rules (Rule Base), membership functions (MFs) and scaling factors (Data Base) designed based on knowledge from a human expert or based on learning and
self-organization
methods
which
do
not
require
a
mathematical model of the system. In addition, because the human sensation of thermal comfort is subjective and self reports can vary between occupants and over time, FLCs based on linguistic rules instead of inflexible reasoning are well adapted to describe HVAC systems and hence apt to increasing thermal comfort [2], [7], [8]. Furthermore, the use of rule-based controllers, especially FLCs, would enable the implementation of multi-criteria control strategies [4]. Thus, the FLC can be applied for the purpose of control of an HV AC
H. Mirinejad, K. C Welch, and L Spicer are with the Department of Electrical and Computer Engineering, University of Louisville, Louisville, KY 40292 USA (e-mail:
[email protected];
[email protected]; IbspicOl @louisville.edu).
978-1-4673-1835-8112/$31.00 ©2012 IEEE
system,
because
it
can
properly
address
the
challenges
inherent in that problem. A fuzzy Rule Base
(RE)
is normally constructed by
2
formulating an expert's implicit knowledge of the underlying
process of automatically generating KB in FSs. In Genetic
process into a set of linguistic variables and if-then rules. For a
Learning, all properties of RB, MFs, or both, are constructed
complex system such as an HV AC system, constructing the
without prior knowledge in respect of either, or both [19].
fuzzy RB from heuristic information is often based on a
Since Genetic Learning is concerned by generating a fuzzy
tedious and unreliable trial-and-error approach [9]. Using soft
system RB, it is a more difficult task rather than Genetic
computing methods is a general solution for the automatic
Tuning which optimizes a performance of a FS that already
construction of RB in fuzzy systems (FSs). Soft computing
operates at least approximately correct [9]. Fig. 2 shows the
methods such as Artificial Neural Networks (ANNs) and
difference between Genetic Learning and Genetic Tuning
evolutionary techniques are two different approaches for the
approaches in GFSs.
automatic construction of RB in FSs. They also can be applied
Genetic Fuzzy System (GFS)
for optimizing the FS parameters, i.e. MFs and Scaling Gains (SGs).
Fuzzy System Construction (Generating RB and/or M F )
Genetic Leaming
The first approach combines the learning capability of neural networks with the knowledge representation of fuzzy logic. This method is described by the general term Neuro Fuzzy System (NFS) and is often used in problems whose goal is minimizing the error between the output of the FS and the target value. In NFSs, learning is used to produce the fuzzy RB or adaptively adjust the rules in the RB. It also can be used
Optimization ofa Fuzzy RB
to optimize the fuzzy data base including MFs and SGs in a
Genetic Tuning
FS [10]-[14].
m Tuning MFs &
The second approach applies evolutionary algorithms to
G
automate the knowledge acquisition stage in FS design. FSs in which evolutionary algorithms are used for the automatic tuning/learning the FS components, are known as Genetic Fuzzy Systems (GFSs). The automatic design of a fuzzy controller can be interpreted as an optimization problem where
Fig. 2. Difference between Genetic Learning and Genetic Tuning methods in Genetic Fuzzy Systems
III. ApPLICATION OF INTELLIGENT CONTROL TECHNIQUES IN HV AC SYSTEMS
the Genetic Algorithm (GA) finds a best solution or a set of optimum solutions on the space of potential solutions. Fig. 1
Intelligent controllers in HVAC applications have been
shows two different approaches for constructing the KB in
extensively studied by many authors. Some research directly
FSs.
uses intelligent controllers for the control of HVAC systems
Knowledge Base (K8) Construction Re triction on Comp�cat ed Application
Automatic Generating Rules
Combination of Fuzzy System and Soft omputing
[1],
[4],
[20]-[22],
while others apply intelligent (fuzzy)
methods to improve the work of current traditional PID controllers [23]-[29]. In the latter case, intelligent techniques are used for auto-tuning the PID controllers to relieve the difficulty of manually adjusting the PID gains. Liang and Du presented the design of an intelligent comfort control
system
by
combining
the
human
learning
and
minimum power control strategies for an HVAC system [1]. In their work, a minimum power control strategy including both
balancing the
input
power
of HVAC
devices and
reducing energy consumption was used. The Predicted Mean Vote
(PMV);
a
thermal
comfort
model
including
six
parameters: air temperature, radiant temperature, air velocity, relative humidity,
occupants' activity level,
and clothing
Fig. I. Knowledge base construction in fuzzy systems
insulation originally proposed by Fanger [20]; was used.
Based on the distinction between the Data Base (DB) and
designed to tune the user's comfort zone by learning the
the Rule Base (RB) in the KB of a FS, there are two different
specific user's comfort preference. The comfort zone first was
Based on the PMV model a human learning strategy was
approaches in GFSs:
used by Fountain et al. to design the control strategy for short
(i) One so called "Genetic Tuning" method is concerned
term occupancy in hotels [30]. The integration of comfort
about optimizing the performance of an already existing FS. In
zone with the human learning strategy was applied for thermal
fact, tuning a FLC is a process of optimizing both given MFs
comfort control. In fact, this system tunes the user's comfort
and input-output SGs,
in order to
zone, i.e. the optimal PMV reference value, instead of the
optimize the FLC output response. Different Genetic Tuning
i.e. DB optimization,
thermal sensation model. Therefore, the application procedure
methods are presented in [15]-[18]. (ii) The second one, named "Genetic Learning", describes a
will be more convenient. To overcome the nonlinear feature of the PMV calculation, a direct neural network (NN) controller
3
is designed. Then, based on the variable air volume (V AV)
take a long time to run, thus the selected tuning algorithm
method, a minimum power control strategy is proposed to
would need to converge quickly. These restrictions were overcome through the use of the objective weighting method,
further optimize the system operation for energy saving. Jian and Wenjian designed an Adaptive Neuro-Fuzzy (ANF) controller for the supply air pressure control loop in an
steady-state
GA,
and
reducing
the
population
size.
The
objective weighting method involved combining multiple
HVAC system [31]. They developed a simple RB for FS
objective functions into one overall objective function by
including
means of a vector of weights, which in this case were obtained
three
rules
and
then
applied
an
error
back
propagation learning rule combined with the least squares
from expert system designers. The steady-state GA involved
method to optimize the FS parameters by ANNs. In order to
selecting two of the best individuals in the population and
increase the capacity of the ANF controller to deal with the
combining them to obtain two offspring. This approach with
steady state error, they also added a conventional integral
the restrictions improved the convergence and simultaneously
controller to the system in a secondary loop. A comparison
decreased the number of evaluations.
between the ANF controller designed by Jian and Wenjian and Bi and Cai's PID controller for the supply air pressure loop
In
Alcal'a
et
al.'s
work,
a
genetic
tuning
strategy
considering an efficient multi-criteria approach has been
control demonstrated that the ANF controller combined with
proposed and then different FLCs have been constructed and
the
advantageous
tested in order to check the adequacy of such a control and
performance as the well-tuned PID controller under normal
tuning technique. Accurate simulation models were designed
secondary
loop
can
have
the
same
work conditions [32]. In addition, in case of large variation in
for two experimental test buildings, and both the genetically
the HV AC parameters, the ANF controller maintained much
tuned and original FLC were compared to a traditional On-Off
robustness.
controller for the same buildings over a 10 day period. Finally,
Arabinda developed a Neuro-Fuzzy Controller (NFC) with
the simulated controllers were implemented and tested in the
the aim of using a smaller number of fuzzy rules leading to a
physical experimental buildings. The results showed that the
savings in computational time. First, an FS with thirty-six
use of expert knowledge for the building of the simulation and
fuzzy rules was developed, then a few of these rules were used
KB accurately matched the physical systems and demonstrated
for
the FLC's superiority through its ability to match thermal
neural
network
training through
a
back
propagation
algorithm. They applied a three-layer neural network with
comfort
two, thirty, and one neuron in the first, second, and third layer
simultaneously reducing energy consumption by over 10%.
level
with
the
traditional
controller
while
respectively. Finally the proposed NFC by Arabinda was
In an extension of the work presented in [4], Gacto et al.
compared with Bi et al.'s PID controller and Jian and
introduced an advanced evolutionary Multi-Objective Genetic
Wenjian's ANF controller for supply air pressure loop control
Algorithm (MOGA) to effectively improve the performance
[32],[31]. The result demonstrated a noticeable improvement
and efficiency of GA tuning of FLCs for an HV AC System
in settling time and peak overshoot for the transfer function of
[21]. Their MOGA Algorithm has been adapted to improve its
the air supply model compared to ANF and PID controller.
exploration ability for fast convergence. Additionally, in order
A GA method has been implemented by Alcal'a et al. to
to improve the algorithm's search ability they added an
develop a smartly-tuned FLC dedicated to the control of
intelligent crossover operator and a mechanism for incest
HVAC systems concerning both energy saving and thermal
prevention in GA Algorithm which maintains population
comfort [4]. In general, Alcal'a et al. recognized the benefits
diversity
of FLCs to implement expert knowledge and control of HVAC
objective framework. These improvements favored a quick
by
avoiding toward
not-useful
good
crossovers
solutions,
which
in
a
they
multi
system in the form of linguistic rules, as well as the difficulty
convergence
in actually gathering the appropriate expert knowledge for a
appropriate to solve the HV AC control problem; due to the
found
specific HVAC control problem. They obtained the initial KB
problem of tuning parameters with simulations, as was noted
from
engineering
by Alcal'a et al [4]. Overall their advanced GA method
knowledge which they subsequently tuned by the application
reduced the number of fuzzy rules, found the best combination
of automatic tuning techniques, in this case a GA. Adequate
of rules for the FLCs, and yielded better performance than
human
expert
experience
and
control
control with fewer rules was possible through the use of the
both traditional On-Off controllers and their previous work
expert knowledge of the system to partition the controller. In
using a mono-objective GA algorithm. Finally it should be
was made easier
noted that in general the runtime for each evaluation was also
because the modification of one parameter influenced a
improved from that in [4] through the use of the improved
smaller number of rules.
GA.
addition,
the
automatic
tuning
process
In the development of the GA method to tune the KB
Nowak and Urbaniak applied fuzzy control algorithms
control parameters two specific restrictions which influenced
combined
its design were mentioned. First, the evaluation of parameters
algorithms in the hierarchical structure for the control of an
would be made based on the evaluation of multiple objectives
HVAC system [21]. They used two different MPC methods:
with
the
Model
Predictive
Control
(MPC)
(such as energy consumption, thermal comfort, etc), which
Dynamic Matrix Control (DMC) and Generalized Predictive
would force a selection among optimizing different criteria.
Control (GPC) algorithms. Their hierarchical structure of
Second, the primary way to access the accuracy of a given
control
controller is through the use of simulations, which generally
conflicting goals (energy consumption and thermal comfort)
demonstrated
a
good
compromise
between
two
4
practically impossible to find an exact mathematical model for
of HV AC systems. In [33], Pargfrieder and Jorgl used a FLC involving seven
these systems. Therefore,
intelligent controllers,
especially
variables (five inputs and two outputs) and optimized it with
FLCs, could be a good alternative. Fuzzy control is well
an
energy
adapted to the subjective concept of thermal comfort and
consumption and to maintain a temperature setpoint, which
easily handles multi-criteria objectives of thermal comfort and
evolutionary
algorithm
to
mInImIze
the
also set aside some important criteria. In their work, three
energy saving without any need to mathematically model the
different intelligent controllers were produced for the same
system. In addition, fuzzy control methods can properly deal
HVAC system: a fuzzy controller with an adaptive power
with nonlinear systems with time lags, which are inherent
profile, the same fuzzy controller used again except with the
features of HV AC systems. Therefore, fuzzy control would be
fuzzy controller parameters optimized using GAs,
and a
the first choice for the control of HV AC systems. However,
Predictive
designing the fuzzy controller is initially difficult and needs
Control (GPC) is a technique for generating a sequence of
some heuristic information about the system. To relieve the
future control signals within each sampling interval in order to
difficulty of constructing the FLC and even for optimizing the
generalized
predictive
controller.
Generalized
optimize the control effort [34-36]. First introduced by Clarke
FLC structure, fuzzy control can be combined with soft
et al. [34], GPC has a considerable robustness; however it
computing methods such as evolutionary algorithms or neural
needs a high calculation time due to minimization of a
networks. Such a combination can be a proper choice for the
complex cost function [37]. Pargfrieder and Jorgl showed that,
control of MIMO, nonlinear HVAC systems.
in comparison
with
controllers used
in
present
building V. REFERENCES
automation, their intelligent controllers can decrease the user discomfort noticeably while saving more energy [33]. traditional PID controllers, Soyguder and Alli used two PID controllers for the control of two different damper gap rates (temperature and humidity control dampers) in which the PID gains
(Kp-KrKD)
were obtained by using fuzzy sets for the
same HV AC system [38]. The damper gap rates of an HVAC system were predicted by using the Artificial Neural Fuzzy Interface System (ANFIS) method. They showed that faster and simpler control solutions can be obtained using ANFIS for predicting the damper gap rates of the system. Another case of such systems is in [39] where a fuzzy self tuning PID controller with ideas from the biological immune system was developed. In fact, Wang et al. combined the capability of universal approximation of fuzzy systems and a new
feedback
control
law
inspired
from
the
1 Liang, and R Du, "Design of intelligent comfort control system with human learning and minimum power control strategies," Energy Conversion and Management, vol. 49,pp. 517-528,2007. [2] H Mirinejad, S.H Sadati, M. Ghasemian, and H Torab, "Control techniques in heating, ventilating and air conditioning systems," Journal o/Computer Science, vol. 4,no. 9,pp. 777-783,2008. [3] S. Li,X. Zhang, 1 Xu,and W. Cai,"An improved fuzzy REF based on cluster and its application in HVAC system," in Proc. 2006 World Congress on Intelligent Control and Automation, Dalian, China, pp. 6455-6459. [4] R Alcal'a, 1M. Ben'ttez, 1 Casillas, O. Cordo'n, and R Pe'rez, "Fuzzy control of HVAC systems optimized by genetic algorithms, Applied Intelligence, vol. 18,pp. 155-177,2003. [5] S. Soygudera, M. Karakoseb, and H. Allia, "Design and simulation of selt:tuning PID-type fuzzy adaptive control for an expert HVAC system," Energy Systems with Applications, vol. 36, no. 3, pp. 45664573,2009. [6] S. 1 Li,W. 1 Cai,X. Q Zhang, and 1 X. Xu, "Modeling of HVAC based on fuzzy RBF neural network," in Proc. 2006 IEEE Con[ on Industrial Electronics and Applications, pp. 1-5. [7] M. Hamdi, and G. Lachiver, "A fuzzy control system based on the human sensation of thermal comfort," in Proc. 1998 IEEE International Con[ on Fuzzy System, Washington DC, USA, pp. 4-9. [8] Y.c. Chioua, and L.W. Lan,"Genetic fuzzy logic controller: an iterative evolution algorithm with new encoding method," Fuzzy Sets Syst., vol. 152,no. 3,pp. 617-635,2004. [9] F. Hoffmann, "Evolutionary algorithms for fuzzy control system design," in 2001 Froc. o/the IEEE, vol. 89,no. 9,pp. 1318-1333. [10] D. Nauck, F. Klawonn, and R Kruse, Foundations 0/ Neuro-Fuzzy Systems, Chichester: John Wiley & Sons,1997. [11] B. Egilegor, .I. P. Uribe, G. Arregi, E. Pradilla, and L. Susperregi, "A fuzzy control adapted by a neural network to maintain a dwelling within thermal comfort, 5th International IBPSA Conference, Building [I]
As an example of using intelligent methods for auto-tuning
biological
immune system (T cells) to tune the PID gains automatically. Simulation results demonstrated the effectiveness of their fuzzy immune self-tuning PID controller in comparison with a traditional PID controller in terms of overshoot, rise time, and settling time of the system response. In contrast to some of the work described where intelligent methods were used to tune traditional PID gains, Hongli et al. used correspondence between PID gains and FLC parameters in order to derive a fuzzy controller from a PID controller [40]. This method could be useful, because constructing the fuzzy rules and initiating the fuzzy controller is rather difficult
"
[12]
in a complex HVAC system [41]. However, finding the PID gains is easier using Ziegler-Nichols or Astrom's modified Ziegler-Nichols tuning methods [42],
[43]. Their
results
showed the designed Fuzzy-PID controller performed better than a traditional PID at tracking the room temperature [40].
[13]
[14] [15]
IV. CONCLUSION A variety of intelligent control methodologies as applied to
[16]
HVAC systems were reviewed and investigated in the present study. Since HV AC
systems are nonlinear,
MIMO with
interrelated parameters, and exposed to uncertainties, it is
[17]
Simulation, Sept 1997. K. P. Arabinda, "Development of neuro-fuzzy controller for applications to HVAC system, inverted pendulum and other processes," Intern. Journ. Computational Cognition, vol. 6,pp. 1-6,June 2008. Y. Wu, B. Zhang, .I. Lu, and K. L. Du, "Fuzzy logic and neuro-fuzzy systems: a systematic introduction," Intern. Journ. ArtifiCial Intelligence and Expert Systems (IJAE), vol. 2,no. 2,pp. 47-80,2011. R. Fuller, Introduction in Neuro-Fuzzy Systems, Series: Advances in Intelligent and Soft Computing, vol. 2,2000. F. Bolala and A Nowr, "From fuzzy linguistic specifications to fuzzy controllers using evolution strategies," 1995 IEEE International Con[ on Fuzzy Systems, pp.l 089-1094. F. Herrera, M. Lozano, and .I. L. Verdegay, "Tuning fuzzy logic controllers by genetic algorithms," Intern. Journ. Approximate Reasoning, vol. 12,pp. 299-315,1995. H. Surmann, A Kanstein, and K. Goser, "Self-organizing and genetic algorithms for an automatic design of fuzzy control and decision
5
[18]
[19]
[20] [21]
[22]
[23]
[24] [25]
[26] [27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35] [36]
[37]
[38]
[39]
[40]
[41]
systems," 1993 European Congress on Fuzzy and Intelligent Teclm., Aachen,Germany,pp. 1097-1104. O. Cord'on and F. Herrera, "A three-stage evolutionary process for learning descriptive and approximative fuzzy logic controller knowledge bases from examples," Intern. Journ. Approximate Reasoning, vol. 17, no. 4,pp. 369-407,1997. O. Cordon, F. Herrera, F. Hoffmann, L. Magdalena, and F. Gomide, "Ten Years of Genetic Fuzzy Systems: Current Framework and New Trends," Fuzzy Sets Syst., vol. 141,pp. 5-31,2003. P. O. Fanger, Thermal comfort analysis and applications in environmental engineering, 1st edition,New York: McGrow Hill,1972. M. Nowak and A. Urbaniak, "Utilization of intelligent control algorithms for thermal comfort optimization and energy saving," 2011 IEEE Carpathian Control Con! (ICCC), pp. 270-274. M. J Gacto, R. Alcala, and F. Herrera, "Evolutionary multi-objective algorithm to effectively improve the performance of the classic tuning of fuzzy logic controllers for a heating, ventilating and air conditioning system," 2011 IEEE International Workshop on Genetic and Evolutionary Fuzzy Systems (GEFS), pp. 73-80. H. X. Li, L. Zhang,K. Y.Cai,and G. Chen,"An improved robust fuzzy PID controller with optimal fuzzy reasoning," IEEE Trans. SYs., Man, and Cybern.. vo1.35, no. 6,2005. W Z. Qiao and M. Mizumoto, "PID type fuzzy controller and parameters adaptive method," Fuzzy Sets Syst., vol. 78,pp. 23-35,1996. H X Li and G. Chen, Feature-based integrated design of fuzzy control systems, Intelligent Systems: Techniques and Applications, vol. VI, C T Leondes (ed),CRC Press,Florida,USA,pp. VI253-VI283,2003. M. Mizumoto, "Realization of PID controls by fuzzy control methods," in Froc. 1992 IEEE Con! Fuzzy Systems, pp. 709-715. H A. Malki, H. Li, and G. Chen, "New design and stability analysis of fuzzy proportional-derivative control systems," IEEE Trans. Fuzzy SYst., vol. 2,pp. 345-354,Apr. 1994. B. M. Mohan and A. V Patel, "Analytical structures and analysis of the simplest fuzzy PD controllers," IEEE Trans. Syst.. Man, CybernB, vol. 32,no. 2,pp. 239-248, Apr. 2002. H Ying, "Practical design of nonlinear fuzzy controllers with stability analysis for regulating processes with unknown mathematical models," Automatica, vol. 30,no. 7,pp. 1185-1195,1994. M. Fountain, G. Brager, E. Arens, F. Bauman, and C. Benton, "Comfort control for short-term occupancy," Energy Build, vol. 21, pp. 1-13, 1994. W. Jian and C. Wenjian, "Development of an adaptive neuro-fuzzy method for supply air pressure control in HVAC system," 2000 IEEE International Co,?! on Systems, Man, and Cybernetics, vol. 5, pp. 3806-3809. Q. Bi, W. Cai and et ai, "Advanced controller auto-tuning and its application in HVAC systems," Control Engineering Practice, vol.8,pp. 633-644,2000. J. Pargfrieder and H. Jorgl, "An integrated control system for optimizing the energy consumption and user comfort in buildings," in Froc. of the 2002 IEEE International Symposium on Computer Aided Control System Design, Glasgow,Scotland,pp. 127-132. D. W. Clarke, C. Mohtadi, and P. S. Tuffs, "Generalized predictive control: Part 1, the basic algorithm, and part 2, extensions and interpretations," Automatica, vol. 23, no. 2,pp. 137-160, 1987. D. W. Clarke and C. Mohtadi, "Properties of generalised predictive control,"Automatica, vol. 25,no. 6,pp. 859-875,1989. D. W Clarke, "Application of generalised predictive control to industrial processes," IEEE Contr, Syst, Mag., vol. 8, pp. 49-55, Apr. 1988. R. Kennel, A. Linder, and M. Linke, "Generalized Predictive Control (GPC) - Ready for use in drive applications?," 2001 IEEE Power Electronics Specialists Co,?! PESC, Vol. 4,pp. 1839-1844. Soyguder and H. Alii, "An expert system for the humidity and temperature control in HVAC systems using ANFIS and optimization with Fuzzy Modeling Approach," Energy and Buildings, vol. 41 pp. 814-822,2009. J. Wang, C. Zhang, and Y. Jing, "Fuzzy immune self-tuning PID control of HVAC system,", in proc, 2008 IEEE International Conference on Mechatronics and Automation. ICMA, pp. 678-683. L. Hongli; D. Peiyong; and J. Lei, "A novel fuzzy controller design based-on PID gains tor HVAC systems," in Froc, 2008 World Congress on Intelligent Control and Automation, pp. 736-739. J. H. Lilly,Fuzzy Control and Identification, Hoboken,NJ: Wiley,2010.
[42] G. F. Franklin, J. D. Powell, and A. Emami-Naeini , Feedback Control of Dynamic Systems, 5th ed.,Englewood Cliffs: Prentice Hall,2006 . [43] K. J Astrom and T Hagglund, PID controllers: theory, design, and tuning, 2nd ed. ISA: The Instrumentation, Systems, and Automation Society,1995.
VI. BIOGRAPHIES Hossein Mirinejad received the M.S. degree in mechatronics engineering from K. N. Toosi University of Science and Technology, Tehran, Iran, in 2008. He has been selected as one of the elite university students by Iranian National Institute of Elites in 2009. He was also honored to receive the fellowship from the University of Louisville, Louisville, KY, in 2011 where he is currently working toward the Ph.D. degree in electrical and computer engineering. His research interests include energy systems, building automation and intelligent controls. Karla Conn Welch (S'99, M'IO) received the Ph.D. degree in electrical engineering and computer science from Vanderbilt University, Nashville, TN in 2009. In 2010, she joined the University of Louisville, Louisville,KY,where she is currently an Assistant Professor in the Electrical and Computer Engineering Department. Her current research interests include machine learning, a±Iective computing, human-machine interaction, and the human impact on energy systems.
Lucas Spicer (S'08) received the B.S. degree in electrical and computer engineering from the University of Louisville, Louisville, KY, in 2011, where he is currently working toward the M.Eng. degree in electrical and computer engineering. His current research interests include machine learning, robotics, and intelligent algorithms, especially those that deal with embedded systems He has competed in the annual IEEE SoutheastCon Hardware Competition since 2009, and is currently the for University of Louisville's NASA University team Student Launch Initiative competition team.
