چكيده به لاتين
Abstract
Many methods are developed , up to now , to conform the results of designer’s work , means new product , to customer’s need and desire , like Quality Function Deployment (QFD) or methods for reducing total cost like Value Engineering (VE).
QFD starts with a gallery of customer desires and needs and finally determines some features and engineering quantities based on weighting goals , and it is supposed that the designed product is accordant to customer desire .
But is it true that the new product is the best one ? while simultaneously optimization of several goals were intended.
In this research , in the field of shape design for new products , it is tried to propose a procedure to find the best fitted design to customer’s desire using genetic algorithm as one of the evolutionary algorithms . Important difference between this research and other “Genetic Algorithm” applications , is due to simultaneously optimization of a qualitative variable (like customer satisfaction ) and quantitative physical variable (like geometric dimensions ) as optimization goal , while the independent variable ( here , shape ) is qualitative.
The method used is as this : each design or shape is divided into many rectangular parcels and then many shapes are randomly created with these parcels . then after , these designs are compared with each other in tournaments , with the criteria of similarity to “ good shapes “ and dissimilarity to “ bad shapes” .
Good and bad shapes are determined through gathering customer’s coments. Finally , the best designs are proposed to designer as the output of written software.
Keywords : Genetic Algorithm, evolutionary algorithms , multi-objective optimization design
