Knowledge Base

Fuzzy rule base [22] is considered as the core component of the Fuzzy Logic controller
which stores all the information necessary to control the plant and it comes from the expert
operator. Therefore, it is considered as the central part of the fuzzy logic controller as it helps the
controller to make intelligent decisions by itself using the fuzzy rule base to make correct control
action.
The fuzzy rule can be divided into two parts i.e. IF (antecedent) THEN (consequent)
where antecedent part defines the condition and consequent part gives the corresponding
corrective control action [25]. The number of rules in a rule base depends on the number of input
and output variables and the number of membership function attached to each variable.
The knowledge base consists of the data base and the linguistic control rule base. The
database contains information about the linguistic control rules and the data manipulation
function for the controller. The rule base consists of IF-THEN rules which specify the control
action defines by the expert member. The fuzzy logic controller looks at the input signals and by
using the rules defined, it determines the appropriate corrective control output signals (control
actions). The main methods of developing a rule base are
1.
2.
3.
4.

Developing the rules using the knowledge of an expert for applying the control actions.
Modeling the control action of the operator.
Modeling the process.
Developing an optimized fuzzy controller.
The initial rules are obtained from the expert which is related to the physical

consideration of the controller, the main objective to be considered by the fuzzy logic controller
to achieve the control actions are:
1. Adjusting the control output by removing any significant error in the process output.
2. Corrective control actions are carried out to make a smooth transition of input to the
nearest output reference value.
3. Keeping the process output value under control within the specified value ie., preventing
it from exceeding the user specified limit.
4. By considering the two dimensional matrix of the input variables, each subspace is
associated with a fuzzy output situation.
Inference Engine

Inference Engine works based on the fuzzy IF THEN rules and fuzzy reasoning. It has
been applied to a wide range of applications, such as automatic control, decision analysis, expert
system robotics, etc. The fuzzy inference system consists of three conceptual components:
1. A rule base contains a selection of fuzzy rules
2. A database or dictionary defines the membership function used in the fuzzy
rules.
3. A reasoning mechanism performs the inference procedure upon the rules and
given facts to derive a reasonable output or conclusion.
With crisp inputs and outputs, a non linear mapping from its input space to output space
is implemented by fuzzy inference system using IF THEN rules, which describes the local
behavior of the mapping. The antecedent part of the fuzzy IF-THEN rules defines the input
region of the input space and the consequent part specifies the output part in the fuzzy region.
There are mainly used inference systems are:
1. Mamdani Fuzzy Inference system and
2. 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 method of FIS is universally accepted (Sivanandam and
Deepa 2010). The Sugeno model cannot strictly follow the compositional rule of inference in its
fuzzy reasoning mechanism.
Decomposition and aggregation operations are done in the inference engine. If the rule is
compound in nature, it is converted into a simple canonical rule form. There is more than one
rule available in the rule based system. In aggregation, the overall consequents from the
individual consequents are obtained.
5.5.8 Defuzzification
The control output from the fuzzy inference engine is fuzzy in nature, it is converted to
crisp output to control the system under study. The technique which converts the fuzzy output of
the controller into its corresponding crisp value is known as defuzzification process. There are
many types of defuzzification methods available such as centre of area, bisector of area, mean
of maxima, smallest of maxima and largest if maxima. However, centre of area is the most widely
adopted defuzzification method. The reasons for using this method are:
1) It is fairly simple and requires less computational time and effort.

2) This method takes into account each and every activated rule of the fuzzy rule base as
sum operator is used unlike other methods using max operator in which the rule with
highest value of the MF is considered for the output [25].