Вы находитесь на странице: 1из 22






Fuzzy controllers are used to control consumer products such as

washing machines, video cameras, rice cookers, as well as industrial
processes, such as cement kilns, underground trains, and robots. Fuzzy
control is a control method based on fuzzy logic. Fuzzy logic can be
described simply as computing with words rather than numbers, fuzzy control
can be described simply as control with sentences rather than equations. A
fuzzy controller can include empirical rules, and that is especially useful in
operator controlled plants. In a rule based controller the control strategy is
stored in a more or less natural language.The control strategy is isolated in a
rule base opposed to an equation based description. A rule based controller is
easy to understand and easy to maintain for a non-specialist end user. A
measurable disturbance is being compensated. It requires a good model, but if
a mathematical model is difficult or expensive to obtain, a fuzzy model may
be useful.

The advantages of fuzzy logic controllers over the conventional PI

controller are that they do not need an accurate mathematical model; they can
work with imprecise inputs, can handle nonlinearity, and may be more robust
than the conventional PI controller. Use of fuzzy logic for minimization of
harmonics and improvement of power quality is not a new issue rather

various authors have introduced some innovative methodologies using these

tools (Suresh Kumar et al 2011).

In fuzzy Logic Toolbox software, fuzzy logic should be interpreted

as FL, that is, fuzzy logic in its wide sense. The basic ideas underlying FL are
explained very clearly and insightfully in Foundations of Fuzzy Logic. What
might be added is that the basic concept underlying FL is that of a linguistic
variable, that is, a variable whose values are words rather than numbers. In
effect, much of FL may be viewed as a methodology for computing with
words rather than numbers. Although words are inherently less precise than
numbers, their use is closer to human intuition. Furthermore, computing with
words exploits the tolerance for imprecision and thereby lowers the cost of
solution.Another basic concept in FL, which plays a central role inmost of its
applications, is that of a fuzzy if-then rule or,simply, fuzzy rule. Although
rule-based systems have a long history of use in Artificial Intelligence (AI),
what is missing in such systems is a mechanism for dealing with fuzzy
consequents and fuzzy antecedents. In fuzzy logic,this mechanism is provided
by the calculus of fuzzy rules.The calculus of fuzzy rules serves as a basis for
what mightbe called the Fuzzy Dependency and Command Language(FDCL).
Although FDCL is not used explicitly in the toolbox, it is effectively one of
its principal constituents. In most of the applications of fuzzy logic, a fuzzy
logic solution is, in reality, a translation of a human solution intoFDCL.A
trend that is growing in visibility relates to the use of fuzzy logic in
combination with neuron computing and genetic algorithms. More generally,
fuzzy logic, neurocomputing, and genetic algorithms may be viewed as the
principal constituents of what might be called softcomputing. Unlike the
traditional, hard computing, softcomputing accommodates the imprecision of
the real world.


The term multilevelstarts with the introduction of the three-level

converter. By increasing the number of levels in a given topology, the output
voltages have more steps generating a staircase waveform, which approaches
closely the desired sinusoidal waveform and also offers reduced harmonic
distortion. One promising technology to interface battery packs in electric and
hybrid electric vehicles are multilevel converters because of the possibility of
high VA rating and low harmonic distortion without the use of a
transformer(Mohemad et al 2008). There are three main topologies for
multilevel inverters. These topologies are diode clamped multi-level inverters
(DCMI), flying-capacitor multi-level inverters (FCMI) and cascaded multi-
level inverters with separate DC sources.

Cascade Multilevel Inverter (CMLI) is one of the most important

topologies in the family of multilevel and multipulse inverters. It requires
least number of components with compare to diode-clamped and flying
capacitors type multilevel inverters and no specially designed transformer is
needed as compared to multipulse inverter. It has modular structure with
simple switching strategy and occupies less space(Jagdish Kumar et al 2008).
The CMLI consists of a number of H-bridge inverter units with separate dc
source for each unit and is connected in cascade or series as shown in

Each H-bridge can produce three different voltage levels: +Vdc, 0,

and Vdc by connecting the dc source to ac output side by different
combinations of the four switches T1,T2,T3 and T4. The ac output of each H-
bridge is connected in series such that the synthesized output voltage
waveform is the sum of all of the individual H-bridge outputs. By connecting
the sufficient number of H-bridges in cascade and using proper modulation
scheme, a nearly sinusoidal output voltage waveform can be synthesized. The

number of levels in the output phase voltage is 2s+1, where s is the number of
H-bridges used per phase(Mitali et al 2012).

Figure 3.1 Cascaded Multilevel Inverter



When designing a fuzzy logic controller, the information obtained

from the operator is more important than the dynamic mathematical model of
the system. The real problem in a control system is the output signal or error
signal in the physical environment. This information plays an important role
in the processes of closed loop systems because the closed loop systems are
commanded according to this information. The aim of fuzzy logic control is
to reduce the error of the system to a minimum. The size of the controller
input is associated with the size of error. The rate of change of error affects
the determination of the controller input. Therefore, as the linguistic changes,
the error and the change of error are used to give the decision according to the
controller rules. The most important state for solving a problem by using
fuzzy logic theories is the determination of the membership functions. In
many studies, it has been shown that the degree of membership of a fuzzy
logic set is directly related to the senses obtained from samples in some
applications. In other applications, it is related to statistical or mathematical
estimations under specific assumptions (Abdelkhalek et al 2008).

Recently, fuzzy logic controller (FLC) is used in power electronic

systems for adjustable motor drives and active power filter applications.
FLCs are widely used because of the following reasons (Altas et al 2007).

It does not need accurate mathematical model.

It can handle non-linearity.

It is more robust than conventional controllers.


3.3.1 Fuzzy Logic Controller Design Steps

Fuzzy logic controller is explained by TimothyJ.Ross(1997).The

steps in designing a fuzzy logic control system are as follows.

1. Identify the variables (inputs, states and outputs) of the plant.

2. Partitioning the universe of discourse or the interval spanned

by each variable into a number of fuzzy subsets, assigning
each a linguistic label (subset include all the elements in the

3. Assign or determine a membership function for each fuzzy


4. Assign the fuzzy relationships between the inputs or states

fuzzy subsets on the one hand and the outputs fuzzy subsets
on the other hand, thus forming the rule base.

5. Choose appropriate scaling factors for the input and the

output variables in order to normalize the variables to the
[0,1] or the [-1,1] interval(optional).

6. Fuzzify the inputs to the controller.

7. Use fuzzy approximate reasoning to infer the output

contributed to each rule.

8. Aggregate the fuzzy outputs recommended by each rule.

9. Apply defuzzification to form a crisp output.


In a non adaptive fuzzy logic controller, the methodology used and

the results of the nine steps mentioned above are fixed, whereas in an
adaptive fuzzy logic controller, they are adaptively modified based on some
adaption law in order to optimize the controller.

3.3.2 Fuzzy Controller

The performance of Fuzzy logic controller is well documented for

improvements of both transient and steadystate performances. The function of
fuzzy logic controller is very useful since exact mathematical model of it
isnot required. The schematic diagram of fuzzy logic control system is shown
in Figure 3.2.

Knowledge Base

Data | Rule

Base | Base

Input Fuzzification Inference Defuzzification Output


Figure3.2Functional block diagram of Fuzzy logic control

It can be divided into four main functional blocks namely

Knowledge base, Fuzzification, Inference mechanism and Defuzzification.
The knowledge base is composed of data-base and rule-base. The data-base,
consisting of input and output membership functions, provides information
for appropriate fuzzification operations, the inference mechanism and
defuzzification. The rule-base consists of a set of linguistic rules relating the
fuzzy input variables to the desired control actions. Fuzzification converts a
crisp input signal, into fuzzified signals that can be identified by level of

membership in the fuzzy sets. The inference mechanism uses the collection of
linguistic rules to convert the input conditions to fuzzified output. Finally, the
defuzzification converts the fuzzy outputs to crisp output signals (Kerrouche
et al 2009).

3.3.3 Fuzzification

Fuzzification converts a crisp input signal, into fuzzified signals

that can be identified by level of membership in the fuzzy sets. Fuzzy logic
uses linguistic variables instead of numerical variables. In a control system,
error between reference and output can be labeled as very small (VS),
Small(S), Medium(M), Large(L),Very Large(VL). The process of converting
a numerical variable (real number) convert to a linguistic variable (fuzzy
number) is called fuzzification.

3.3.4 Inference Mechanism

It works on the concept of fuzzy if-then rules and fuzzy reasoning.

It has a fruitful outcome in many fields such as automatic control, data
classification, decision analysis, expert systems, robotics etc. It implements a
nonlinear mapping from its input space to output space. The control rules are
evaluated by inference mechanism. There are mainly two forms of inference
mechanism available(Nagewara Rao et al 2010).

Mamdani Fuzzy Inference System

Sugeno Fuzzy Inference System

The basic difference between the two methods is in obtaining the

consequent of fuzzy rules. Mamdani FIS uses fuzzy sets for the rule
consequents and Sugeno FIS uses linear functions of input variables.
Mamdani FIS is universally accepted.

Decomposition and aggregation operations are done in the

inference system. If the rule is compound in nature, it is converted into a
simple canonical rule form. There should be more than one rule available in
the rule base system. In aggregation, the overall consequents from the
individual consequents are obtained.

3.3.5 Knowledge Base

The knowledge base is composed of data-base and rule-base. The

data-base, consisting of input and output membership functions, provides
information for appropriate fuzzification operations. Rule base involves
defining the rules represented as a set of if then rules, for example if (alpha0
is very small VS) and (alpha1 is very small VS) and (alpha2 is very small
VS) then (V1 is very large VL) (V3 is large L) (V5 is medium M). Database

The Database stores the definition of the membership function

required by fuzzifier and defuzzifier. The membership to accommodate
various degrees of membership on the real continuous interval is [-1, 1].
But there are infinite number of values in between the end points [-1, 1],
which can represent various degrees of membership for an element. A crisp
set has a unique membership function, whereas a fuzzy set can have an
infinite number of membership functions to represent it. The main power and
strength of membership function is that it employs some amount of overlap.

The various form of membership function are

membership function

S membership function

L membership function

Triangular membership function

Gaussian membership function

Trapezoidal membership function Rule Base

In conventional controllers, we have control gains or control laws

which are combination of numerical values. In FLC, the equivalent term is
rules and they are linguistic in nature.

IF premise (antecedent- input) THEN conclusion (consequent-

output). This form of knowledge representation, characterized as shallow
knowledge, is quite appropriate in the context of linguistics because it
expresses human empirical and heuristic knowledge in ones own language of
communication. The IF-THEN rules can be made as compound rules by
adding AND in between IF-THEN. Then, the fuzzy rules become as
compound rules like IF..ANDAND..THEN.

There are mainly four approaches in the formation of fuzzy rules

From the experience and control engineering knowledge of

the experts.

From the behavior of the human operator working in the


Using a fuzzy model of a process.

Obtaining a relationship through experience or simulation

with a learning process.

A typical rule can be written as follows;

If e is A and ce is B then output is C

Where A, B, C are the labels of linguistic variables of error (e),

change of error (ce) and output respectively.

Here e, ce and output represents degree of membership.

3.3.6 Defuzzification

The output generated by fuzzy controller must be crisp which is

used to control the drive unit and thus accomplished by the defuzzification
block. The rules of FLC generate required output in a linguistic variable
(Fuzzy Number), according to real world requirements, linguistic variables
have to be transformed to crisp output (Real number). The choices available
for defuzzification are numerous. So far the choice of strategy is a
compromise between accuracy and computational intensity.

Many defuzzification strategies are available, such as, the weight

average criterion, the mean-max membership, and center-of-area (centroid)
method. The defuzzification technique used here is based upon centroid


Let X be a collection of objects denoted generically by{x}, which

could be discrete or continuous, X is called the universe. If an element in the
universe, say x, is a member of fuzzy set Athen mapping is given as

( ) [0,1]

= [ , ( )|

The basic fuzzy set operations needed for evaluation of fuzzy rules are AND
),OR( ) and NOT ).

AND -Intersection: = min [ ( ), ( )]

OR -Union: = max [ ( ), ( )]

NOT -Complement: =1 ( )



Fuzzy logic controller is designed for the optimum selection of

switching angles for harmonic mitigation in multilevel inverter. The
following steps were involved in the design procedure.

1. The input variables are the switching angles and the output
variables are the harmonic voltage.

2. The universe of discourse for the input and output variables

are portioned into five intervals. The linguistic variables
are Very Small(VS), Small(S), Medium(M), Large(L),
Very Large(VL).The range for the input is [0,90] and
is defined as VS (0,18), S(18,36), M(36,54), L(54,72),
VL(72,90); range for the control output is [-300,300] and is
defined as VS(-300,-180), S(-180,-60), M(-60,60), L(60,180),

3. Triangular membership function is taken for the input

and output variables and they are designed as shown in
Figures 3.3, 3.4.

4. The Rule base stores the linguistic control rules required by

rule evaluator (decision making logic).In the proposed work
three inputs (alpha0,alpha1 and alpha2) were used with each
input having five membership functions (Very Small(VS),
Small(S), Medium(M), Large(L),Very Large(VL)).So totally
125 rules were formed(Number of inputs*square of number
of membership functions).The rules used are shown in
Appendix 1.

5. No normalization is done for the input and output variables.

6. The inputs are applied.

7. Fuzzy approximate reasoning is applied to the above inputs.

8. In the aggregation of fuzzy outputs and operation is

performed on the rule outputs.

9. Centroid method of defuzzification is used in this work.


Figure 3.3 Membership function for the input


Figure 3.4 Membership function for the output



The implementation of the Fuzzy logic Control technique is

performed in the working platform of MATLAB (version 7.10). The
technique is implemented in a seven level three-H bridge inverter in such a
way that it can eliminate the 3rd order and 5th order harmonics and so it can
minimize the total harmonic distortion.The output voltage of each bridge is
shown in Figure3.5,3.6 and3.7. The output voltage obtained for optimal
switching angles for the multilevel inverter are given in Figure3.8.Output
voltage Vs time plot is given in Figure 3.9. FFT analysis of inverter output
voltage is shown in Figure 3.10. The Total Harmonic Distortion and the
computational time for modulation index of 0.5 are listed in the Table 3.1.

Figure 3.5 Output waveform of inverter bridge 1


Figure 3.6 Output waveform of inverter bridge 2


Figure 3.7 Output waveform of Inverter Bridge 3


Figure 3.8 Output waveform of the Inverter using Fuzzy Logic Controller

Figure 3.9 Output Voltage Vs Time plot of the Inverter using Fuzzy
Logic Controller

Figure 3.10 FFT Analysis of Inverter with Fuzzy Logic Controller


Table 3.1 Results of multilevel inverter with Fuzzy Logic Controller

Total harmonic Computational

distortion (THD) time in Seconds

Fuzzy Logic 1.8468 0.0795

BasilM.Saied et al (2008) found 3 rd, 5th,7th, 9th and 11th harmonics

with the help of SHE-PWM technique on a single phase 3 level H bridge
inverter using Neural network and Fuzzy Logic. The 3rd and 5th harmonic
voltages for Fuzzy Logic controller were reported as 3% and 1%
respectively.Hence it is clear that the proposed Fuzzy Logic Controller used
in this thesis gives better result with the THD of 1.8%.


The application of Fuzzy logic controller in harmonic reduction of

a multilevel inverter has been explained in detail. The design steps for Fuzzy
Logic Controller are given and their application in the proposed system is
clearly explained. The designed Fuzzy Logic Controller is tested in the 7 level
H bridge inverter and the simulated results are obtained.