Beruflich Dokumente
Kultur Dokumente
K. RAM CHARAN
Assistant Professor in the department of Electrical & Electronics Engineering,
L.B.R College of Engineering, Mylavaram
Andhra Pradesh, India
K.RAMALINGESWARA PRASAD
Assistant Professor in the department of Electrical & Electronics Engineering,
L.B.R College of Engineering , Mylavaram
Andhra Pradesh, India
http://www.lbrce.ac.in/eee_Ramalingeswara.html
Abstract:
This paper presents the comparison of tracking control performance of, parallel combination of fuzzy PI & PD
controllers and three input fuzzy PID controller. The analytical expressions of Parallel structure of Fuzzy PI & PD
Controller is derived via triangular membership functions by using Mamdani’s minimum inference method with
Zadeh fuzzy logic AND, Lukasiewicz fuzzy logic OR and center of gravity defuzzification method. (In Simulation
the tracking Control performance of Non-linear Systems using the fuzzy PID and parallel combination of fuzzy
PI&PD controllers by considering triangular, gaussian, gbell, trapezoidal and pi input membership functions.)
Keywords: Fuzzy control systems, membership functions, non-linear systems, PID controllers.
1. Introduction
Since the first successful application of the idea of the fuzzy sets [1] to the control of dynamic plant by Mamdani
and Assilian [2], there has been considerable worldwide interest in the subject of “Fuzzy Control System
Engineering”. It has been known that it is possible to control many complex systems effectively by human operators
who have no knowledge of their underlying dynamics, while it is difficult to achieve the same with conventional
controllers. It is this fact which has ultimately led to the prospective development of fuzzy control in a variety of
applications [3], most of these applications have been based on the intuitive implementation of domain experts’
experience.
In recent years, Fuzzy logic controllers (FLC), especially fuzzy proportional –integral-derivative controllers
have been widely used for industrial process owing to their heuristic nature associated with simplicity and
effectiveness for both linear and non linear systems.
Fuzzy logic controllers can be divided into three main categories: (1) Fuzzy logic controllers, in which
fuzzy inference and if -then logic rules are employed without using precise mathematical models;(2) Fuzzy logic
controllers, whose structure is designed based on fuzzy logic with precise mathematical models.(3) fuzzy logic and
convention technique combined controllers.
Conventional PID controllers generally do not work well for non-linear systems, and particularly complex
and vague systems which do not have precise mathematical models. In order to overcome these difficulties, a class
of non-conventional fuzzy PID controllers has been used by us.
The structure of parallel combination of PI & PD preserves the basic properties and merits of the
Conventional PID controller, but has a simple configuration similar to the fuzzy PI and PD controllers. It has four
simple fuzzy if-then rules for each of the PI and PD controllers. Similarly the structure of the three input Fuzzy PID
preserves the basic properties and merits of the Conventional PID controllers, but it has eight simple fuzzy if-then
rules.
The present paper should be an attempt to understand what the parallel structure of fuzzy PID controller is
and how well it can perform in comparison with the three input fuzzy PID controller. Here it has demonstrated that
the parallel combination of fuzzy PI & PD controller is better than the three input fuzzy PID controller for nonlinear
systems, this is not a defect of the fuzzy PID Controller. The fuzzy PID controllers were designed for both linear and
non-linear systems. The parallel combination fuzzy PI & PD controller developed has a new structure that is not a
simple modification of the conventional version.
This paper organized as follows, section II describes the principal components of parallel combination of
fuzzy PI & PD controller. Section III describes fuzzification algorithm and fuzzy control rules, section IV describes
the defuzzification algorithm, section V describes analytical input output relationship and section VI describes
finally computer simulations.
Where K cp and K dc are the proportional and derivative gains, respectively, and Y(s) is the output signal. This
1
equation can be transformed into the discrete version by applying the bilinear transformation s 2 T z 1 z 1,
where T>0, is the sampling, period, which results in the following form:
2( z 1) c
u PD ( z ) ( K pc
1
K d )Y ( z ) . (5)
T ( z 1)
The above equation (5) is simplified as,
u PD ( nT ) K 1p d ( nT ) K d y ( nT ), (6)
Where
uPD(nT) uPD(nT T)
uPD(nT) ,
T
y ( nT ) y ( nT T )
d ( nT ) ,
T (7)
y ( nT ) y ( nT T )
y ( nT ) .
T
Again, we have,
uPD (nT) uPD (nT T ) TuPD (nT).
uPD(nT) becomes the fuzzy control action in the new design, where we will use K uPD as a fuzzy control gain,
which will be determined later in the design and rewrite it as
u PD ( nT ) u PD ( nT T ) K uPD u PD ( nT ) , (8)
The parallel combination of fuzzy PI & PD control system (9) is shown in Fig. 2. This fuzzy PI-PD controller is
obtained by inserting the fuzzy PI and the fuzzy PD controllers in the conventional PI-PD controllers [6].
3.1. Fuzzification
We Fuzzify the PI and PD components of the Fuzzy PI-PD control system individually and then combine the desired
fuzzy control rules for each of them. The overall control law given in (9) is taken into consideration during
fuzzification. Similar to the fuzzy PI controller [7], the input and the output membership functions of the PI
component are shown in Fig.3.
The fuzzy PI controller employs two inputs, namely the error signal we * e and the rate of change of error
signal we *e . The fuzzy PI controller has a single output called the incremental control output and is denoted by
u PI (nT ) as shown in Fig. 3.
The fuzzy PD controller has two inputs, namely, average change of output wy * y and the rate of change of
output y * y . The fuzzy PD controller has a single output called the incremental control output and is denoted
w
by u PD (nT ) . The inputs to the fuzzy PD controller have to be fuzzified before being fed into the controller. The
membership functions for the two inputs and the output of the controller are shown in Fig. 4.
Fig. 3 functions for error, rate of change of error and incremental control output
Fig. 4 Membership functions for average output, rate of change of output and incremental control output
4. Defuzzification
In the defuzzification step, for both the fuzzy PI and PD controllers, the commonly used “centre of gravity” formula
[4], [6] is employed to defuzzify the incremental control of the fuzzy control law. The “centre of mass” formula for
the defuzzification reads as:
u (nT ) membership value of input output corresponding to the membership value of input
membership value of input
Superimposing the four individual input space divisions to form overall input space division for all rules in
fuzzy PI controller is shown in fig [6] and for fuzzy PD controller are shown in fig [7]. The control rules for the
fuzzy PI controller (R1)-(R4), with membership functions and input-combination (IC) regions together, are used to
evaluate appropriate fuzzy control laws for each region. Working through all regions, we obtain the following
formulas for the four IC regions are shown in Table 1.
formulas in above, we use the values and apply the straight-line formulas obtained from the geometry of Fig. 7. is
shown in Table 2. membership function of the incremental control output signal over the vertical wy * y -axis as
shown in Fig.
Table 2. Input spaces for Fuzzy PD Controller
Wu [Wy * y Wy *y ]
IC1
4 2Wy * y
Wu [W y* y Wy* y ]
IC2
4 2Wy* y
Wu [W y* y Wy* y ]
IC3
4 2Wy* y
Wu [W y* y Wy* y ]
IC4
4 2Wy* y
Therefore from Table 1 & Table 2, we have derived the possible 16 analytical expressions for Fuzzy PI-PD
controller and are listed in Table 3.
6. Computer Simulations
As a first example consider a non-linear system described by equation y =0.0001 y (t ) +u(t) and the unit step
responses are shown in Fig. 8 to Fig. 12.Computer simulations are done for the Fuzzy PI-PD controller, Parallel
combination of Fuzzy PI+PD Controller ,and three input Fuzzy PID Controller when triangular, Gbell, trapezoidal,
pi, and Gaussian membership functions in both input and output are used .
7. Conclusions
We have described the design principle of a Parallel combination of fuzzy PI & PD controller and also the analytical
input and output relations are derived using triangular membership functions. The parallel combination of fuzzy PI
& PD controller is a discrete time fuzzy version of the parallel combination of conventional PI & PD controller. The
results demonstrate that the parallel combination of fuzzy PI & PD controller shows the better control performance
than fuzzy PID controller.
8. References
[1] Zadeh, L. A (1965). Fuzzy sets. Inf. Control, 8 , pp-338-353.
[2] E .H Mamdani, S. Assilian. An experiment in linguistic synthesis with a fuzzy logic controller, internat. J. Man- machine stud.7 (1)(1974).
Pp1-13.