Experimental Selection of Functions and Operators Of Genetic Algorithms of Matlab Environment

The publication was focused on a comparison study of the functions and operators of genetic algorithms available in the Matlab environment. The research was carried out by solving an optimisation task aimed at reconstructing the surface shape of the flaccid membrane of a pneumatic extracorporeal Ventricular Assist Device – VAD. The paper presents the results of the experimental selection of the best set of selection, mutation and crossover functions to achieve the lowest possible reconstruction error by determining the optimal distribution of markers for determining the membrane surface. This task is a typical optimisation task consisting in searching the space of acceptable solutions. The correct distribution of markers is important for the accuracy of modelling a membrane surface shape. Studies have been performed for a convex membrane having a known mathematical description. The comparison criterion for the used sets of functions and operators was the accuracy of mapping the obtained shape of the membrane surface in relation to the shape of the reference surface (the reference). The membrane surface model is determined by recognising positions of markers of real membrane by using visual and interpolation techniques in 3D space, as presented in a research study by Murawski (2015) and Sulej et al (2017).