Machine Learning-Based Model for Optimum Design of TMDs by Using Artificial Neural Networks

AbstractIn this study, it was considered that approach where optimization and machine learning tools are utilized together for inducting possible structural damages to minimum via optimum modeling of tuned mass dampers (TMDs) in frequency-domain. For this respect, in the optimization process realized as the first step, the metaheuristic flower pollination algorithm (FPA) was used for the minimization of the biggest amplitude for the transfer function. In the second step, a rapid-effective prediction tool was developed by training some mechanical parameters belonging to optimum TMD designs via a machine learning-based model. Accordingly, they are possible to directly and quickly determine TMD parameters required for new test models thanks to this tool developed by benefiting from artificial neural networks (ANNs) and obtaining transfer function values at minimum level with the help of parameters. Thus, an equivalent solution was found to optimization analyses taking long times with the development of a system, which can predict quickly, effectively, and without additional operations.KeywordsTuned mass dampers (TMDs)Frequency domainMetaheuristic algorithmsFlower pollination algorithmMachine learningArtificial neural networks (ANNs)