A2-4 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
IMPROVED GENETIC ALGORITHM FOR DETERMINATE OF LOCAL AREA OF COMPROMISE BY MULTI - CRITERION OPTIMIZATION Boriana Vachova 1 , Svetla Vasileva 2 1 Institute of Computer and Communication Systems – BAS, Akad. G.Bonchev str., Bl.2, 1113 Sofia, Bulgaria boriana@hsi.iccs.bas.bg 2 Institute of Control and Systems Research – BAS, Akad. G.Bonchev str., Bl.2, P.O.Box 79, 1113 Sofia, Bulgaria vasileva@icsr.bas.bg Abstract In this work has been proposed a common idea of genetic algorithms (GA) for multi criteria optimization, when it is supposed the presence of only one domain of compromise defining from a priori knowledge’s. There are possibly two following domains: Domain of agreement and Domain of compromise. Key word – GA, Pseudo-Gradient Method, Fitness Function 1. INTRODUCTION Optimization of the processes in multi inputs objects in real conditions, demanding and e.g. is multi criteria and it is related to series of iteration procedures which could decrease its efficiency [2]. The criteria have been divided in two multitudes: criteria that will be maximized and criteria that will be minimized. For the first one multitude are related criteria such as productivity, property, homogeneity and for the second one - prime cost, expenses, wastes, heterogeneity and e.g. [2]. This kind of multi criteria optimization has not been correct task (in mathematical meaning), because the extremes for the different criteria have been achieved by different values of arguments. It is possibly to transfer multi criteria optimization in to correct task if the additional conditions are involved concerning admissible deviations for local extremes of separated criteria. Concerning the common conceptions of multi criteria optimization, the main task is to find the domains of compromise. In this work has been proposed a common idea of genetic algorithms (GA) for multi criteria optimization, when it is supposed the presence of only one domain of compromise defining from a priori knowledge. 2. THE ESSENCE OF THE METHOD According to common ideas of GA [1], [2] the starting point of optimization procedure has been selected using arbitrary choice procedure. The values of GA fitness functions have been expressed the values for adaptative criteria of two multitudes. In this work has been selected three up to four new values of arguments and has been specified experimentally fitness functions values, using common probabilities for selection, mutation and crossover operators. The relations between variations of fitness functions and variation of arguments have been defined at a place of the optimization procedure. There are possibly two following domains: Domain of agreement - in this domain maximize criteria receives increasing of its fitness functions and minimize criteria receives decreasing of its fitness functions. It is necessity “variation” for values of arguments for the next iteration to correspond with direction of this agreement. Domain of compromise - in this domain where
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 02/18/2010 for the course ITK ETF113L07 taught by Professor Popovskiborislav during the Spring '10 term at Pacific.

Page1 / 4

A2-4 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online