Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Neuro-Fuzzy Programming to Finding Fuzzy Multiple

Objective Linear Programming Problems
1 2
Anil Kumar Yadav Dr. Savita Mishra
1 2
Research Scholar Assistant professor
University Department of Mathematics, Department of Mathematics, The Graduate School College
Kolhan University, Chaibasa, Jharkhand, India For Women, Jamshedpur, Kolhan University, Jharkhand,
3
Dr. B. N. Prasad
3
Principal Singhbhum College Chandil, Jharkhand, India

Abstract:- A neural network for solving fuzzy multiple fuzzy systems. The neural networks, that have efficient
objective linear programming problems is proposed in learning algorithms, had been presented as an alternative to
this paper. The distinguishing features of the proposed automates or to support the development of tuning fuzzy
Neural network are that the primal and dual problems systems. The first studies of the Neuro-fuzzy systems date
can be solved simultaneously, all necessary and sufficient of the beginning of the 90’s decade, with Jang, Lin and Lee
optimality conditions are incorporated, and no penalty in 1991, Berenji in 1992[8,12] and Nauck from
parameter is involved. we prove strictly an important 1993[4,7,11].
theoretical result so that, for an arbitrary initial point,
the trajectory of the proposed network does converge to Fuzzy Neural networks try to make the biological
the set of its equilibrium points, regardless of whether a functions of the human brain. The main advantages of the
multiple objective linear programming problem has neural networks is the fact of these structures could learn
unique or infinitely many optimal solutions. Numerical with examples (training vectors, input and output samples of
simulation results also show that the proposed network the system).The main characteristics of the neural networks
is feasible and efficient. In addition, a general method for are: • learning capacity • generalization capacity •
transforming nonlinear programming problems into robustness in relation to disturbances.
unconstrained problems is also proposed.
II. STATE OF THE ART
Keywords:- Fuzzy Neural Network ,Fuzzy Multiple
objective ,Neuro-fuzzy, Constraint satisfaction Learning Let us consider fuzzy decision problems in the form
Fuzzy constraints, Linear Programming Problems . of

I. INTRODUCTION max (C1x,…,Ckx) (1)

x

Fuzzy logic systems can deal with incomplete

Where Ci =(Ci1,…,Cin) is a vector of fuzzy numbers
knowledge of an environment with a vague nature. This
makes fuzzy logic a powerful tool for reasoning in imprecise and x Rn is the vector of ’crisp decision variables.
knowledge based systems[2]. Due to continuous-valued Suppose that for’ each objects e function of (1) we have two
logic nature, fuzzy logic is much closer to human thinking references fuzzy numbers. denoted be mi and Mi, which
than the conventional two-valued logic. Problems with need represent undesired and desired levels for the i-th objective
to reasoning on imprecise knowledge are called fuzzy ,respectively.
programming problems. Neuro Fuzzy programming
We now can state (1) as follows: find an x  Rn such
*
problems have many utilization, such as in the biological
field, artificial intelligence, operations research field . In this *
that Ci x is as close as possible to the desired point Mi, and
so many researcher and research scholar are emphasized on
it is as far as possible from the undesired point in mi for
these field [4].
each i.
The Fuzzy systems propose a mathematic calculus to
III. FUZZY NEURAL NETWORKS
translate the subjective human knowledge of the real
processes. This is a way to manipulate practical knowledge Suppose that there is a fuzzy neural network with the ser
with some level of uncertainty. The fuzzy sets theory was of input-output pairs {(Ai Bi,), i = 1. .. . ,m},where Ai =
initiated by Lofti Zadeh in 1965.Since the moment that
(Ai1,…,Ain) is a vector’ of fuzzy numbers and the output Bi, is
fuzzy systems become popular in industrial application, the
a fuzzy numbers. In computer applications we usually use
engineers designer perceived that the development of a
discrete versions of the continuous fuzzy sets. The discrete
fuzzy system with good performance is not an easy task.
version of the above system is
Thus, appears the idea of applying learning algorithms to the

IJISRT23APR034 www.ijisrt.com 19
Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

…. …

where (x1….xM)and (y1…..y N) are well-chosen partitions of the input and output spaces.

Other possibility is to input the α -level sets of fuzzy numbers [6]. We need to find weights. such that for all input vectors .Ai all
the computed α -level sets are as close as possible to the α -level sets of the target fuzzy numbers Bi . The number of inputs and outputs
depend on the number of α -level sets considered.

A dicretization of fuzzy input Aij where aij =Ai(xj)

In the architecture [5] input vectors and target outputs where given by fuzzy numbers. Weights and biases,however were ,
given by real numbers as in the case of standard back propagation a1gorithm.

The following extension of back propagation algorithm to train neura1 networks from fuzzy training pattern was proposed
by [5].

Wi

Wij

Ap1 Apn

The structure of Neural fuzzy networks is proposed in [5].

The output of the i-th hidden unit is

IJISRT23APR034 www.ijisrt.com 20
Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

For the output unit

Let us denote the α-level sets of the computed output OP and the target output BP

The cost function to be minimized can be stared as follow s

Where
eP L ( )  ( BP L ( )  OP L ( ))2 / 2
eP R ( )  ( BP R ( )  OP R ( ))2 / 2
The cost function for the training pattern p is

From the cost function ep(α) the following learning rules can be derived

IV. FUZZY LINEAR EQUATIONS

Let us consider the system of fuzzy linear equations

A11x1 +… + A1n xn = B1 …. ….. (2)

where Aij and Bi are fuzzy numbers.

The problem is to find a crisp vector x∈Rn satisfying this system of equations as far as possible. We use the following single layer
fuzzy neural network for finding an approximate solution to (2).

where the training set is {( Ai1,…. , Ain, Bi, ), i = 1, . ., .,m} the coordinates of decision variable x.

It is clear that in the end of the learning we get an x satisfying bestly (in the sense of closeness of α -level sets) the system of
equations (2).

It is well-known It fuzzy expert systems and neural networks are universal approximators. However, as was pointed out by
Buckley and Hayashi [l], fuzzy neural networks can not approximate all continuous fuzzy functions, which means that the extended
backpropagation algorithm from [5] can not be always used in the process of finding approximate solutions to systems of fuzzy
equations.

IJISRT23APR034 www.ijisrt.com 21
Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

Oi=i
A x ij j B

V. APPLICATION OF FUZZY NEURAL NETWORKS TO FMOLP

Let us Consider the following FMOLP

(3 )

Where Ci is a vector of fuzzy numbers i = 1,……, k.

Suppose that for’ each objective function of (3) we have two references fuzzy numbers. denoted by mi, and Mi. which represents
undesired and desired levels for’ the i-th objective respectively.

Now we should find an x  R . such that Cix* is as close as possible to the desired point Mi. and it is as far’ as possible from the
* n

undesired point mi, for each i.

Let di denote the maximal distance between the α -level sets of mi and Mi and let mi be the fuzzy number obtained by shifting mi by
the value of di , in the direction of Mi.Then we consider mi . as the reference level for’ the biggest acceptable value for’ the i-th objective
function.

M1 m1 C1X*

It is clear’ that good compromise solutions should be searched between Mi and mi we can introduce weights measuring the
importance of ”closeness” and farness”.

C1X* M1

IJISRT23APR034 www.ijisrt.com 22
Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
*
C1 x is close to M1, but not far enough from m1

Let  ∈[0, 1] be the grade of importance of ”closeness” to the desired level and then (1 —  ) denotes the importance of
“farness” from the undesired level.

Then we can use the following training set for Our single layer fuzzy neural network:

Input Output

….. ……

M1 ΩM1+ (1-Ω)m'1 C 1X* m'1

VI. A GOOD COMPROMISE SOLUTION [3.] D. E. Goldberg, Genetic Algorithms in Search,

Optimization, and Machine Learning, Reading,
It is clear that in the end of training we set the optima 1 MA, Addison-Wesley, 1989.
weights (decision variables), for which the values of the fuzzy [4.] E.H.L.Aarts ,J.WESSELS and P.J. Zwietering ,The
objectives are as close as possible to the desired levels and and applicability of neural nets for decision support,In :
far from the undesired levels in the sense of the choosen Proceedings of the First European conference on
importance degrees. Fuzzy and Intelligent Technologies , Aachen ,sep
7-10,1993,Verlag der Augustinus Buchhandlung
VII. CONCLUSION ,Aachen ,1993 379-386.
An effort has been made to solve FMLOP based on [5.] F. M. Ham and I. Kostanic, Principles of Neuro
Neuro- fuzzy theory.This paper presented a approach that computing for science and engineering, New York,
was adopted while figure compromise the solution .The McGraw Hill, 2001.
major benefit of this paper is that the finding approximate [6.] H.Ishibuchi. H.Tanaka and H.Okada, Interpolation
solutions to systems of fuzzy equations.The proposed of fuzzy if –then rules by neural networks ,Int.J.of
approach can be extended to solve fuzzy multi-objective Approximate Reasoning , 10( 1994) 3-27.
linear programming problems FMOLPPs. [7.] H.Ishibuchi. K.Kwon and H.Tanaka,
Implementation of fuzzy if-then rules by fuzzv
ACKNOWLEDGEMENT neural networks with fuzzy weights, In: Proceeding
of First European conference on Fuzzy and
The authors take this opportunity to thank the Kolhan Intelligent Technologies , Aachen ,sep 7-
University P.G. Department of Mathematics and Dr. B.K. 10,1993,Verlag der Augustinus Buchhandlung
Sinha , for providing the facilities and for encouragement to ,Aachen ,1993 209-2015.
carry out this work thoroughly. [8.] H. R. Berenji, “A reinforcement learning—based
architecture for fuzzy logic control,” Int. J. Approx.
REFERENCES Reason., vol. 6, no. 2, pp. 267–292, Feb. 1992.
[9.] J. H. Holland, Adaptation in Natural and Artificial
[1.] A.Castro, V. Miranda, Mapping neural networks Systems, Cambridge, MA, MIT Press. Second
into rule sets and making their hidden knowledge edition, 1992.
explicit application to spatial load forecasting, in: [10.] J.J.Buckley and Y..Havashi, Can fuzzy neural nets
Proceedings of the 14th Power System approximate continuous fuzzy functions?,Fuzzy
Computation Conference, 2002. Sets and Systems, 6(1994) 43-51.
[2.] B. Kosko, Neural networks and fuzzy systems: a [11.] T.Hashiyama, T.Furuhashi and Y.Uchikawa, A
dynamical systems approach, Prentice Hall, study on multi-attribute decision making process
Englewood Ciffs, NJ, 1991. using fuzzy neural networks ,In: Proceedings of the
Fifth IFSA World congress ,1993 810-813.

IJISRT23APR034 www.ijisrt.com 23
Volume 8, Issue 4, April 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[12.] T.Hashiyama, T.Furuhashi and Y.Uchikawa, A
decision making model using fuzzy neural
networks ,In: Proceedings of the 2nd International
conference on Fuzzy Logic &Neural Networks
,Iizuka ,Japan,1992 1057-1060.
[13.] V. S. Kodogiannis, and A. Lolis, "Prediction of
foreign exchange rates by neural network and fuzzy
system based techniques," European Symposium
on Artificial Neural Networks, Bruges, Belgium,
2001.

IJISRT23APR034 www.ijisrt.com 24