High Throughput Multi-Solution Genetic Algorithm (HTMGA) for identification of alternate solutions in a structural model updating context

Model updating techniques seek to improve numerical models of existing structure by integrating experimental data from the structure. Many techniques use an error function that describes the differences between the numerical model and the experimental data. For example, if dynamic data is available, the error between experimental and numerical modal parameters can be used. Optimization techniques are used to identify the structural parameters that minimize this function. Most model updating techniques are interested in finding the global minima of the error function. However, one could argue that due to a number of factors, such as modeling errors and low sensor density, a local minimum could provide a more physically meaningful solution with a slightly reduction on the performance of the error function. This research presents a modified genetic algorithm that is able to identify global and local extremes (maxima or minima) within a high throughput computing environment. High Throughput Genetic Algorithm is accomplished by adding several operators to traditional genetic algorithms. The capabilities of the algorithm are explored using analytical functions and experimental data of a bench-scaled structural system. Results indicate that the proposed technique is able to determine the local minima or maxima within the context of model updating of structural systems.