Fuzzy linear programming is one of the linear programming developments which able to accommodate uncertainty in the real world. Genetic algorithm approach in solving linear programming problems with fuzzy constraints has been introduced by Lin (2008) by providing a case which consists of two decision variables and three constraint functions. Other linear programming problem arise with the presence of some coefficients which are fuzzy in linear programming problems, such as the coefficient of the objective function, the coefficient of constraint functions, and right-hand side coefficients constraint functions. In this study, the problem studied is to explain the genetic algorithm approach to solve linear programming problems where the objective function coefficients and righthand sides are fuzzy constraint functions. PT Dakota Furniture study case provides a linear programming formulation with a given objective function coefficients and right-hand side coefficients are fuzzy constraint functions. This study describes the use of genetic algorithm approach to solve the problem of linear programming of PT Dakota to maximize the mean income. The genetic algorithm approach is done by simulate every fuzzy number and each fuzzy numbers by distributing them on certain partition points. Then genetic algorithm is used to evaluate the value for each partition point. As a result, the Final Value represents the coefficient of fuzzy number. Fitness function is done by calculating the value of the objective function of linear programming problems. Empirical results indicated that the genetic algorithm approach can provide a very good solution by giving some limitations on each fuzzy coefficient. Genetic algorithm approach can be extended not only to resolve the case of PT Dakota Furniture, but can also be used to solve other linear programming case with some coefficients in the objective function and constraint functions are fuzzy.
Algoritma genetika (AG) merupakan algoritma pencarian yang didasarkan pada mekanisme seleksi alamiah dan genetika alamiah. Karena didasarkan pada teori-teori dalam ilmu biologi, banyak istilah dan konsep biologi yang digunakan dalam algoritma ini.
Kata Kunci : Algoritma Genetika, Fuzzy Linear Programming, Linear Programming, Metode Simpleks 2 Fase
DESIGN AND DEVELOPMENT OF COMPONENT LIBRARY GENETIC ALGORITHM BY USING OBJECT-ORIENTED DESIGN AND PROGRAMMING
Hadi Suyono, Adharul Muttaqin, Eka Prakarsa Mandyartha
Abstract
DESIGN AND DEVELOPMENT OF COMPONENT LIBRARY GENETIC ALGORITHM BY USING OBJECT-ORIENTED DESIGN AND PROGRAMMING aHadi Suyono, bAdharul Muttaqin, and cEka Prakarsa Mandyartha a,b,cDepartment of Electrical Engineering, Faculty of Engineering, University of Brawijaya E-Mail: hadis@ub.ac.id Abstrak Makalah ini menyajikan desain dan pembuatan komponen library Algoritma Genetik dengan menggunakan pendekatan object-oriented designand programming (OODP) dan Component-based Develepment (CBD). KomponenAlgoritma Genetika (AG) merupakan komponen software enginedibuat sendiri yang digunakan untuk membantu menyelesaikan persoalan optimisasi dengan menggunakan struktur Algoritma Genetika yang disebut dengan Library Algoritma Genetika (LibAGen). Metodologi OODP dan CBD meliputi analisis kebutuhan, diagram use-case, diagram kelas dan diagram sekuensial. Library Algoritma Genetika (LibAGen) ini terdiri dari 22 kelas yang dikelompokkan dalam namespace berdasarkan struktur desain AG yang diperlukan meliputi representasi populasi, fungsi evaluasi, operator genetika (crossover dan mutasi) dan seleksi. Untuk mengukur performansi dari engine LibAGen validasi telah dilakukan dengan menggunakan persamaan fungsi sinusoidal dua parameter. Waktu eksekusi dan nilai optimum parameter dengan beberapa pengujian dengan variasi jumlah generasi (iterasi) juga dilakukan pada makalah ini. Parameter AG yang digunakan adalah probabilitas crossover 25% dan probabilitas mutasi 1%. Hasil uji validasi menunjukkan bahwa nilai fitness terbaik adalah 388,501 dengan nilai parameter x1 = 11,6256 dan x2 = 5,7249. Terdapat perbedaan tidak signifikan antara nilai fitness terbaik dibandingkan dengan hasil Michalewicz (1999) yaitu sebesar 0,08%. Kata kunci:Algoritma Genetika, component library, object-oriented design and programming (OODP) Abstract This paper presents the design and development of Genetic Algorithm (GA) library components by using object-oriented design and programming (OODP) and Componentbased development(CBD). Genetic Algorithm component is an engine software component which is developed by own development for solving the optimization problem by using a structure of Genetic Algorithm (GA) called as Genetic Algorithm Library (LibAGen). OODP and CBD methodologies include requirement analysis, use-case diagrams, and class diagrams. Genetic Algorithm Library (LibAGen) consists of 22 classes which is grouped into namespaces based on GA design structure that include population representation, evaluation function, genetic operators (crossover and mutation) and selection. To measure the performance of the LibAGen engine, a validation has been carried outby using a sinusoidal function with two-parameters. Optimal parameter with some testing through variations of the number generations (iterations) have been performed in this paper. The GA parameters selected are crossover probability of 25% and mutation probability of 5%. Validation test results indicate that the best fitness and parameters are 388,501, x1 = 11,6256 and x2 = 5,7249. There is no significant result in term of the best fitness compared with Michalewicz (1999) i.e. 0.08% Key words:Genetic Algorithm, component library, object-oriented design and programming (OODP)
DOI: https://doi.org/10.21107/kursor.v7i4.1108
Refbacks
- There are currently no refbacks.
Informatika - UTM
