Mating in Genetic Algorithm with Application in System Identification

It is well-known by many that the three important genetic operators employed by genetic algorithm (GA) are selection, crossover and mutation. Throughout the decades of its evolution, a variety of these operator variants were introduced. This chapter talks about an ‘operator’ that is new, hence comparatively scarcely discussed by GA researchers—mating. Mating is the process where pairs of chromosomes in a GA population are made before crossover takes place. Attention has started to be given to this operator as studies have shown that the addition of this operator further enhances the effectiveness of GA in escaping local optimum. Some studies on mating are presented followed by a thorough exploration of a type of mating called single parent mating (SPM). The investigation of the effectiveness of incorporating SPM in GA is broken down into a number of aspects: (i) how a suitable proportion of GA population applied with SPM enables the exploration of uncharted search space; (ii) how different types of crossover affect GA performance when used together with SPM; and (iii) how the usage of SPM improves GA performance in overcoming premature convergence. This is carried out by GA application in system identification problem—an optimisation problem where an optimum mathematical model is searched to define the behaviour of a black box system.