International Journal of Computer and Information Technology (ISSN: 2279 0764)

Volume 02 Issue 02, March 2013

Fuzzy Logic Models for Selection of Machining

Parameters in CAPP Systems

Saleh M. AMAITIK
Industrial and Manufacturing Systems Engineering Department
Benghazi University
Benghazi, Libya

AbstractFuzzy logic is a mathematical theory of inexact Very little literature is available in application of fuzzy
reasoning that allows us to model the reasoning process of logic in process planning [6]. El-Baradie [4] is one of the first
humans in linguistic terms. It is very suitable in defining the to suggest a fuzzy logic model for machining data selection. He
relationship between the system inputs and the desired system described the development stages of a fuzzy logic model for
outputs. This paper presents fuzzy logic models to select metal cutting. The model is based on the assumption that the
machining parameters (cutting speed and feed rate) in automated relationship between the hardness of a given material and the
process planning (CAPP) systems. Each model utilizes two-input recommended cutting speed is an imprecise relationship, and
and two-output variables which are partitioned into several fuzzy can be described and evaluated by the theory of fuzzy sets. The
sets according to their minimum and maximum values allowed to
model has been applied to data extracted from the Machining
control the model. A set of fuzzy rules have been constructed for
each model, based on the knowledge extracted from machining
Data Handbook, and a very good correlation was obtained
data handbooks. Once the rules are evaluated the variables are between the handbook data and that predicted using the fuzzy
defuzzified and converted into the corresponding output logic model. Wong et al [7] have suggested a new fuzzy model
variables (cutting speed and feed rate). An example is given to for machinability data selection, which is different from El
demonstrate and verify the application of the developed fuzzy Baradie [4]. The model suggested by El Baradie [4] was a one-
models. The results obtained are compared with the inputone-output fuzzy relationship by considering the depth
corresponding ones obtained from machining data handbook and of cut as a discrete parameter. Whilst Wong et al. [7] showed
shown good fit. the feasibility of incorporating the depth of cut as one of the
continuous parameters required to determine the cutting speed.
Keywords- Fuzzy logic; machining parameters; process planning; Hashmi et al [8] have developed a fuzzy logic model used to
CAPP select cutting speeds for three different materials in drilling
operation. The relationship between a given material hardness
I. INTRODUCTION and drilling speed was described and evaluated by fuzzy
relation for different cutting tool materials and different hole
In a Computer Aided Process Planning (CAPP) diameters and feed rates.
system, the machining parameters have to be selected
automatically. The approaches used in the existing In this research work, several fuzzy logic models are
computerized selection systems for the machining parameters developed to select machining parameters in drilling and
fall into one of the four categories, namely data storage and milling type operations. Each model can utilize two input
retrieval, empirical method, expert system and mathematical variables, two output variables, five fuzzy sets, nine workpiece
modeling. The storage and retrieval procedure requires a large materials, and two tool materials combinations. In order to
amount of memory space for storing data, and the data it explain the steps involved in the development of these models,
provides are often too conservative and not optimal. The there are three basic components of fuzzy models which have
systems based on empirical approaches reduce the data to an to be described. These components are fuzzification of input
empirical form whereas the expert system approach uses the and output variables, fuzzy rules applications, and
stored data or the empirical equations along with a knowledge defuzzification of the output variables. The following sections
base [1]. The mathematical modeling is using optimization will explain detailed steps on how these components of the
algorithms with constraints. It is difficult for traditional fuzzy model are implemented.
optimization algorithms to solve this problem because of the
problems of convergence speed and accuracy [2]. Recently, II. FUZZIFICATION OF INPUT AND OUTPUT VARIABLES
process planners have started using artificial intelligent
techniques, such as neural networks, fuzzy logic and genetic Fuzzification is a mathematical procedure for converting an
algorithms to select the machining parameters and have made element in the universe of discourse into the membership value
some progress [2, 3, 4, 5]. of a fuzzy set. Fuzzification dividing the input and output
variables into fuzzy regions (sets). The first step of

www.ijcit.com 279
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

fuzzification process is to define the fuzzy sets in the input and rules is known as Mamdani fuzzy rules [9, 11]. A simple but
output variables. The possible domain interval of both the representative Mamdani fuzzy rule describing the selection of
inputs and outputs are divided into a number of regions in such cutting speed in drilling operation is given as:
away that they overlap each other. The length of region may
differ for each variable and one membership function is IF Material Hardness is Soft AND Hole Diameter is Medium
assigned to each region [9, 10]. THEN Cutting Speed is High.

The input variables in this research work are material

hardness, hole diameter, depth of cut, and thread pitch. The
output variables are cutting speed and feed rate. Table I
presents the inputs, outputs, and the domain intervals of the
variables used in the developed fuzzy models as well as the
range of each variable. The universe of input and output
variables have been partitioned according to their minimum
and maximum values allowed controlling the models. Table II
shows the fuzzy sets of the variables and their associated
values and labels. The number of fuzzy sets for each input and
output variable is five sets. Each fuzzy set has a defined
linguistic term and specified range which can be modified to
control the fuzzy model. Figure 1. A triangular membership function of a fuzzy set
Although scientific publications have suggested many TABLE II. FUZZY SETS OF INPUT AND OUTPUT VARIABLES
different types of membership functions of fuzzy logic,
standard membership functions are used in most practical Fuzzy set
Range
Symbol
applications [11]. In this research work, a triangular a b c
membership function is used for all input and output variables. Material hardness
Very soft 0 0 150 VS
It is defined by the following equation: Soft 0 150 250 S
Medium 150 250 350 M
Hard 250 450 H
0 ( x a)
Very hard 350 450 550 VH
x a Hole diameter
a (a x b) Very small 0 0 6 VS
Triangle( x, a, b, c) bc Small 0 6 13 S

x (b x c) Medium 6 13 26.5 M
c b (c x ) Large 13 26.5 55 L
0 Very large 26.5 55 80 VL
Depth of cut
It has three parameters 'a' (minimum), 'b' (middle), and 'c' Very small 0 0 1 VS
Small 0 1 2.5 S
(maximum) that determine the shape of the triangle. Figures 1
Medium 1 2.5 4 M
shows a triangular membership function of a fuzzy set. Figures Large 2.5 4 6 L
2 and 3 show the membership functions for the input and Very large 4 6 10 VL
output variables of the developed fuzzy models. Thread pitch
Very short 0 0 0.5 VS
Short 0 0.5 1.5 S
TABLE I. DOMAIN INTERVALS OF INPUT AND OUTPUT VARIABLES
Medium 0.5 1.5 2.5 M
Variable Range Unit Long 1.5 2.5 3.5 L
Input variables Very long 2.5 3.5 4.5 VL
Material hardness 0 550 BHN
Hole diameter 0 80 Mm Where material hardness and hole diameter are input
Depth of cut 0 10 Mm
Thread pitch 0 4.5 Mm
variables and cutting speed is output variable. "Soft",
Output variables "Medium", and "High" are fuzzy sets, and the first two are
Cutting speed 0 250 m/min called input fuzzy sets while the last one is called output fuzzy
Feed rate 0 1.5 mm/rev or mm/tooth set. The variables as well as linguistic terms, such as High can
be represented by mathematical symbols.
III. FUZZY RULES KNOWLEDGE BASE
Thus, a Mamdani fuzzy rule for a fuzzy model involving
A fuzzy model uses fuzzy rules, which are linguistic IF- two input variables and two output variables can be described
THEN statements involving fuzzy sets and fuzzy inference. as follows:
Fuzzy rules play key role in representing expert modeling IF X1 is A AND X2 is B THEN Y1 is C, Y2 is D
knowledge and experience and in linking the input variables of Where X1 and X2 are input variables, and Y1 and Y2
fuzzy models to output variables. The most used type of fuzzy are output variables. A, B, C, and D are fuzzy sets, and AND

www.ijcit.com 280
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

is a fuzzy logic operator. The IF-part "X1 is A AND X2 is B" is workpiece materials, and two tool materials combinations.
called the rule premise, whereas the remaining part, THEN- With these numbers of parameters, each fuzzy logic model
part "Y1 is C, Y2 is D", is called the rule conclusion. needs a maximum number of rules about 450 rules. For the
purpose of explanation on how these rules are utilized, Table 4
shows the fuzzy rules in a tabulated form used to select cutting
speed and feed rate for twist drilling of carbon steel workpiece
with high speed steel tool.

(a) Material hardness membership function

(e) Cutting speed membership function

(b) Hole diameter membership function

(f) Feed rate membership function

Figure 3. Membership functions for output variables

TABLE III. SUMMARY OF DEVELOPED FUZZY LOGIC MODELS

(c) Depth of cut membership function Fuzzy logic model Input variables Output variables
Hole diameter Cutting speed
Twist drilling model
Material hardness Feed rate
Hole diameter Cutting speed
Spade drilling model
Material hardness Feed rate
Center drilling,
Hole diameter Cutting speed
Counterboring, and
Material hardness Feed rate
countersinking model
Hole diameter Cutting speed
Reaming model
Material hardness Feed rate
Depth of cut Cutting speed
Boring model
Material hardness Feed rate
Thread pitch Cutting speed
(d) Thread pitch membership function Tapping model
Material hardness Feed rate
Depth of cut Cutting speed
End milling
Figure 2. Membership functions for input variables Material hardness Feed rate
Depth of cut Cutting speed
Side milling
A set of fuzzy rules have been constructed for each fuzzy Material hardness Feed rate
model, based on the knowledge extracted from machining data Depth of cut Cutting speed
Plane milling
Material hardness Feed rate
handbooks [12]. In this research work, nine fuzzy models have
been developed as shown in Table III. Each model can utilize Table IV shows the example of tabulated rules between
two input variables, two output variables, five fuzzy sets, nine work material hardness (bhn) and hole diameter (d) and the

www.ijcit.com 281
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

corresponding cutting speed and feed rate. The first column the truncation of the output fuzzy set. This method is applied to
denotes the fuzzy sets for the workpiece material hardness (VS, all rules to obtain the final result which gives the final shape of
S, M, H, VH) starting from very soft to very hard. The first row the output fuzzy membership function after aggregation of all
denotes the fuzzy sets for hole diameter varying from very the rules, respectively. Then the union operation is applied to
small to very large (VS, S, M, L, VL). The contents of the table all the output fuzzy sets to yield the final fuzzy set [10].
are the outputs yielded, which are the cutting speed (v) and
feed rate (f) for this model. The AND used in the rules will V. DEFUZZIFICATION OF OUTPUT VALUES
apply to the fuzzy AND operation. A few examples of fuzzy
rules (from Table IV) in a Mamdani form are presented as Defuzzification is a mathematical process used to convert a
follows: fuzzy set or fuzzy sets to a real number. It is necessary step
because fuzzy sets generated by fuzzy inference in fuzzy rules
R1: IF bhn is Very soft AND d is Very small must be somehow mathematically combined to come up with
THEN v is Medium, f is Very slow one single number as the output of a fuzzy model [10].

R2: IF bhn is Soft AND d is Small Every fuzzy model uses a defuzzifier, which is simply a
mathematical formula, to achieve defuzzification. For fuzzy
THEN v is Medium, f is Very slow models with more than one output variable, defuzzification is
carried out for each of them separately but in a very similar
R3: IF bhn is Soft AND d is Large THEN v is Low, f is Fast
fashion. In most cases, only one defuzzifier is employed for all
output variables, although it is theoretically possible to use
TABLE IV. FUZZY RULES TABLE FOR TWIST DRILLING CARBON STEEL different defuzzifiers for different output variables.
WORKPIECE WITH HSS TOOL
In this research work, the centroide or center of area (COA)
Hole diameter (d)
Material is used as a defuzzifier for all output variables of the developed
Very Small Medium Large Very
Hardness (bhn) models. In COA defuzzification the crisp value u* is taken to
small large
v M M M M M be the geometrical center of the output fuzzy value out(u),
Very soft
f VS S M F F where out(u) is formed by taking the union of all fuzzy rule
Soft
v L L L L L contributions. The center is the point which splits the area
f VS S M F F under the out(u) curve into two equal parts. The defuzzified
v L L L L L output is defined as:
Medium
f VS S M M V
v VL VL VL VL VL
Hard
f VS S S M F N

u
v VL VL VL VL VL
Very hard
f VS S S M M i out (ui )
IV. FUZZY INFERENCE u* i 1
N
Fuzzy inference is sometimes called fuzzy reasoning. It is
used in a fuzzy rule to determine the rule outcome from the
given rule input information. Fuzzy rules represent modeling

i 1
out (u i )

knowledge or experience. When specific information is
assigned to input variables in the rule premise, fuzzy inference Where the summation is carried over (discrete) values of
is needed to calculate the outcome for output variables in the the universe of discourse ui sampled at N points. COA is a
rule conclusion. In other words, for the general Mamdani fuzzy well-known and often used defuzzification method [10].
rule, the question about fuzzy inference is the following: Given
Xi = i, for all of i, where i are real numbers, what should Yi
be? For fuzzy modeling, after fuzzifying Xi at i and applying VI. FUZZY LOGIC ALGORITH
fuzzy logic AND operation on the resulting membership values In this research work, nine fuzzy models have been
in the fuzzy rule, we attain a combined membership value, , developed for selection of machining parameters. Each model
which is the outcome for the rule premise. Then, the question is utilizes two input variables and two output variables. Each
how to compute "Then-part" in the rule. Calculating "THEN" input and output variable is partitioned into five fuzzy regions
is called fuzzy inference. Specifically, the question is: Given , (sets). A fuzzy rule base for each fuzzy model is generated and
how should Yi be computed? [9]. saved in separate files. The rule base contains a maximum
A number of fuzzy inference methods can be used to number of rules about 450 rules for each model. The Max-Min
accomplish this task. In this research work, the Max-Min inference method is used as the fuzzy inference engine, and the
inference method is used. In this method, all the fuzzy AND center of area method is used as a defuzzifier for the output
operations are applied into all the input's value of the variables. The fuzzy logic algorithm developed in this research
corresponding fuzzy sets. Applying a fuzzy AND operation work is based on four main steps summarized as follows:
will yield a result that is the minimum of the fuzzy value of the
number of input variables. The aggregation of the rule will be

www.ijcit.com 282
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

A. Fuzzification of Input Values (Y j ) FSO ( y1 ) FSO ( y 2 ) ...... FSO ( y p ) .

1 2 m
Fuzzify the input values of X1, X2, Xn variables using the
fuzzification formula of triangular membership function: C. Defuzzification of Output Values
The crisp numbers for Y1, Y2, ., Yp can be calculated
using the center of area defuzzification method as follows;
x j ai
1 (ai x j bi )


bi a i
Yj
Y (Y )
j j

ci x j (bi x j ci ) (Y )
j
FSIi ( x j ) 1
ci bi
VII. MACHINING PARAMETERS SELECTION EXAMPLE
To demonstrate the application of the proposed fuzzy
0 otherwise models, an example is being presented to select cutting speed
and feed rate for twist drilling of carbon steel workpiece with

high speed steel tool. Consider the situation where work
material hardness, BHN, is equal to 275 and hole diameter, D,
Where is equal to 15 mm. Using the fuzzification formula of triangular
FSIi ( x j ) are the degree of membership value of input membership function, the crisp material hardness value of 275
belongs to fuzzy set Medium to a degree of 0.75 and to fuzzy
variables Xj to the fuzzy set FSIi. set Hard to degree of 0.25. Similarly, crisp hole diameter value
Xj are the input variables to the fuzzy model (j = 1, 2, ., n, n of 15 belongs to fuzzy set Medium to degree of 0.85 and to
= number of input variables). fuzzy set Large to degree of 0.15, see Figure 4.
ai are the minimum values of the fuzzy set FSIi.
bi are the middle values of the fuzzy set FSIi.
ci are the maximum values of the fuzzy set FSIi.
FSIi are the fuzzy sets associated with the input variables Xj ( i
= 1, 2, .., m, m = number of fuzzy sets for Xj variable).

B. Fuzzy Inference
Pass the fuzzified values to the fuzzy rules knowledge base (a) Material hardness membership function
through the inference method used. The evaluation form for
possible rules can be written as follows;
IF X1 is FSI1 AND X2 is FSI2 AND Xn is FSIm
THEN Y 1 is FSO 1 , Y 2 is FSO 2 , .., Y p is FSO q
Where
Yj are the output variables of the fuzzy model (j = 1, 2, .., p, p =
number of output variables.
FSOi are the fuzzy sets associated with the output variables Yj
(i = 1, 2, ., q, q = number of fuzzy sets for Yj variable).
Use the Max-Min inference method (i.e. the min "^" (b) Hole diameter membership function
interpretation of AND). The degree of fulfillment of each fired
rule can be calculated as follows;
Figure 4. Fuzzification of the input values
DOFk FSI1 ( x1 ) FSI2 ( x2 ) ........ FSIm ( xn ) .
These fuzzified values are then passed to the fuzzy rules
Where k is the number of fired rules contributed for output knowledge base through the inference method used. The only
variable calculation. rules that will have degree of fulfillment, DOF (it is a measure
of the degree of similarity between the input and the premise of
The fuzzy output (Yj) is the union (max ) of the the rule) greater than zero will fire up. In the case of this
contributions DOFk and can be written as follows; example the following four rules will fire up (see Table IV).
R1 IF bhn is M and d is M THEN v is L, f is M

www.ijcit.com 283
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

R2 IF bhn is M and d is L THEN v is L, f is M (v) = low(v) low(v) verylow(v) verylow(v)

R3 IF bhn is H and d is M THEN v is VL, f is S (f) = medium(f) medium(f) slow(f) medium(f).
R4 IF bhn is H and d is L THEN v is VL, f is M
Selecting crisp numbers for cutting speed, v* and feed rate,
Using the Max-Min inference method of DOF (i.e. the min f representative of (v) and (f) is the process of
*

(^) interpretation of AND), each rule contributes the machining defuzzification. Using COA defuzzification, the crisp outputs
parameters values. The degree of fulfillment of each rule can of the cutting speed, v* and feed rate, f* are obtained as follows;
be calculated as follows (see Figure 4 for evaluation steps);
R1: DOF1 = medium(275) ^ medium(15) = 0.75 ^ 0.85 = 0.75.
v*
v. (v) f *

f . ( f )
(v) ( f )
Contributes Low to cutting speed and Medium to feed rate.
Similarly, rules R2, R3, and R4 can written as
R2:DOF2 = medium(275) ^ large(15) = 0.75 ^ 0.15 = 0.15.
Contributes Low to cutting speed and Medium to feed rate. According to the input values of material hardness, BHN =
275 and hole diameter, D=15 mm, it is found that the cutting
R3:DOF3 = hard(275) ^ medium(15) = 0.25 ^ 0.85 = 0.25. speed v* =13.56 m/min and feed rate f* = 0.2 mm/rev. The
Contributes Very Low to cutting speed and Slow to feed rate. corresponding values of cutting speed and feed rate obtained
from Machining Data Handbook [12, 13] are v = 15 m/min and
R4:DOF 4 = hard(275) ^ large (15) = 0.25 ^ 0.15 = 0.15. f = 0.22 mm/r. These values show a good correlation between
Contributes Very Low to cutting speed and Medium to feed Machining Handbook recommended values of cutting speed
rate. and feed rate and those predicted by fuzzy logic model.
The fuzzy outputs (v) and (f) are the union (max) of these
four contributions for each parameter; that is

Figure 5. Evaluation of the fuzzy algorithm of the example

www.ijcit.com 284
 International Journal of Computer and Information Technology (ISSN: 2279 0764)
Volume 02 Issue 02, March 2013

[3] Chen, Y.-T. & Kumara, S.R.T., Fuzzy logic and neural networks for
VIII. CONCLUSIONS design of process parameters: a grinding process application. Int. J.
Prod. Res., 36, 395-415, 1998.
This paper presented several fuzzy logic models for [4] EL Baradie, M. A., A fuzzy logic model for machining data selection,
selecting machining parameters (cutting speed and feed rate) in Int. J. Mach. Tools and Manuf., 37, 1353-1372, 1995.
drilling and milling type machining operations. Indeed, the [5] Wong, S.V. & Hamouda, A. M., Machinability data representation
proposed fuzzy models have been implemented in a CAPP with artificial neural networks". Journal. Mater. Process. Technol., 138,
system developed by the author in [14]. The results obtained 538-544, 2003.
showed that fuzzy logic approach can provide a promising [6] Singh, N., & Mohanty, B. K., A fuzzy approach to multi-objective
approach for automated knowledge acquisition and can be routing problem with application to process planning in manufacturing
systems. Int. J. Prod. Res., 29, 1161-l 170, 1991.
advantageously used in the building of new generation CAPP
systems. This is due to its ability to cope with the dynamic [7] Wong, S.V., Hamouda, A.M.S. & EL Baradie, M.A., Generalized fuzzy
model for metal cutting data selection. J. Mater. Process. Technol., 89-
changes of manufacturing systems. In addition, fuzzy logic 90, 310-317, 1999.
approach utilizes the knowledge of machining data and avoids [8] Hashmi, K., Graham, I.D. & Mills, B., Fuzzy logic based data selection
complex optimization procedures hence it may be more for drilling process. J. Mater. Process. Technol., 108, 55-61, 2000.
acceptable to process planners. [9] Ying, H., Fuzzy control and Modeling: Analytical foundations and
applications, IEEE, New York, 2000.
ACKNOWLEDGMENT [10] Tsoukalas, L.H. & Uhrig, R.E., Fuzzy and neural approaches in
Engineering, John Willey & Sons, Inc., New York, 1997.
The author would like to express his gratefulness and [11] Berkan, R.C. & Sheldon, L.T., Fuzzy systems design principles:
appreciation to Libyan Ministry of Higher Education and building fuzzy IF-THEN rule bases, IEEE Press, New York, 1997.
Scientific Research and Benghazi University for their support. [12] Machining Data Handbook, 3rd ed., 12. Metcut Research Associates
Inc., Cincinnati, 1980.
REFERENCES [13] Boaisha, S. M. & Amaitik, S.M., "Forecasting Model Based on Fuzzy
Time Series Approach", Proceedings of the 10th International Arab
[1] Shunmugam, M.S., Reddy, S.V. & Narendran, T.T., Selection of Conference on Information Technology - ACIT 2010, Benghazi, Libya,
optimal conditions in multi-pass face-milling using a genetic algorithm. December 14-16, 2010.
Int. J. Mach. Tools and Manuf., 40, 401-414, 2000.
[14] Amaitik, S.M. & Kilic, S.E., An intelligent process planning system for
[2] Liu, Y. & Wang, C., A modified genetic algorithm based optimization prismatic parts using STEP features. The International Journal of
of milling parameters. Adv. Manuf. Tech.,15, 296-799, 1999. Advanced Manufacturing Technology 31 (9), 978-993, 2006.

www.ijcit.com 285