A DIC Based GA Optimization Scheme Using Six Variable Chromosomes

A Digital Image Correlation using a Genetic Algorithm based optimization technique for measurement of whole field surface displacements and strains has been described. The GA scheme has been implemented using a six variable system which includes direct measurement of strains. Earlier problems reported during cross-over and blending of chromosomes primarily owing to differences in order of magnitude of the strains and displacements have been overcome by using pre-generated separate look-up tables of equal length for most probable values of u and v displacements as well as the gradients. The indices of these tables have then been used as the chromosomes instead of the actual values of these deformation parameters themselves. The chromosomes could then be readily mated and/or mutated. The respective values for the six parameters were required to be retrieved from the corresponding look-up tables at every iteration in order to compute the cost function. Feasibility of adopting this approach to solve problems faced when dealing with other multi-variable GA systems having magnitude mismatch has also been studied and some results presented.