A Novel Single Parent Mating Technique in Genetic Algorithm for Discrete – Time System Identification

Abstract

System identification is concerned with the construction of a mathematical model based on given input and output data to represent the dynamical behaviour of a system. As a step-in system identification, model structure selection is a step where a model perceived as adequate system representation is selected. A typical rule is that the model must have a good balance between parsimony and accuracy in estimating a dynamic system. As a popular search method, genetic algorithm (GA) is used for selecting a model structure. However, the optimality of the final model depends much on the optimality of GA. This paper introduces a novel mating technique in GA based on the chromosome structure of the parents such that a single parent is sufficient in achieving mating that demonstrates high exploration capability. In investigating this, four systems of linear and nonlinear classes were simulated to generate discrete-time sets of data i.e. later used for identification. The outcome shows that GA incorporated with the mating technique within 10% - 20% of the population size is able to find optimal models quicker than the traditional GA.