Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/255567860
CITATIONS READS
13 2,000
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Engin Yesil on 18 October 2014.
Istanbul Technical University, Electrical and Electronics Faculty, Control Systems Division,
34390, Maslak, Istanbul, TURKEY
E-mail: {yesil, gkaya, eksin}@elk.itu.edu.tr
Fax: +902122856700
Abstract: The aim of this study is to examine various studies on fuzzy PID
controllers in literature and to classify these fuzzy controllers into categories . There
exist three major categories: Direct action (DA) type fuzzy PID controllers, fuzzy
gain scheduling (FGS) type fuzzy PID controllers, and hybrid type fuzzy PID
controllers. The DA type fuzzy PID controllers are further classified according to
the number of the input variables; namely single input, two input, and three input
fuzzy PID controllers.
In the rest of the paper, various studies on fuzzy PID variable to generate three separate fuzzy proportional
controllers in literature will be examined and they will action [12, 15].
be classified according to the above mentioned three
major categories.
To obtain proportional, integral and derivative control Figure 8. A fuzzy PID controller formed by
action all together, it is intuitive and convenient to combining a fuzzy PD and a fuzzy I controller
combine PI and PD actions together to form a fuzzy
PID-type controller [6, 21-24]. Therefore, the
formulation of fuzzy PID controller can be achieved
by combining fuzzy PI and PD controllers with two
distinct rule-bases. Another and a simpler way of
construction a fuzzy PID controller is combining
fuzzy PD controller with an integrator and a
summation unit at the output. These two cases are
given in Figure 6 and in Figure 7, respectively. It is
obvious that the fuzzy PID controller given in Figure
7 has less number of rules and scaling factors
compared to the fuzzy PID that is given in Figure 6
[25].
Figure 13. The closed-loop control structure for Besides the form of the fuzzy rule set stated in (1), an
parameter adaptive PID type fuzzy logic controller via alternative form is utilized in [14], where the
relative rate observer Mamdani fuzzy model is implemented. In that study, a
new scheme utilizes fuzzy rules and reasoning to
The mechanism given in Figure 11 ameliorates the determine the control signal. It is demonstrated that
response after the occurrence of the first peak value the human expertise on PID gain scheduling can be
and therefore it outperforms worst than the other two represented in fuzzy rules. Furthermore, better control
mechanisms. However, one should keep in mind that performance can be expected in the proposed method
the mechanism given in Figure 11 has no parameter to than that of the PID controllers with fixed parameter
be adjusted [41]. since this type of fuzzy PID controllers can adapt
themselves to varying environments. The fuzzy PID
3. FUZZY GAIN SCHEDULING TYPE structure used in [14] is given in Figure 14. The main
CONTROLLERS difficulty in using this category of fuzzy PID
controllers is that the analysis task is relatively tough,
The fuzzy gain scheduling (FGS) is a rule-based as it is hard to acquire the equivalent nonlinearity of
scheme for gain scheduling [14, 42, 43]. This type of the fuzzy knowledge base. The issue of stability is
fuzzy PID controller is composed of the conventional also addressed in [14]. As in CGS, global stability
PID control system in conjunction with a set of fuzzy cannot be ensured, because the linear controller
rules (knowledge base) and a fuzzy reasoning parameters are functions of time. It is suggested that a
mechanism. The PID gains are tuned on-line in terms supervisor be included to monitor the stability of the
of the knowledge base and fuzzy inference, and then control system and take appropriate actions when
the PID controller generates the control signal. FGS is instability is identified.
implemented in essentially the same way as
conventional gain scheduling (CGS), expect that the
operating regions are associated with overlapping
membership functions of the fuzzy sets defined in the
scheduling variable space and that a fuzzy inference
mechanism is used to dynamically interpolate the
controller parameters around region boundaries based
on known local controller parameters. Hence, it is able
to express and use incomplete knowledge of the
controlled system and provide smooth transitions from
region to region.