Artificial Intelligence-Based Estimation of Natural Frequencies in Discretely Supported Plates
Description
This dataset contains finite element simulation results for a 3000 × 2000 × 4 mm steel plate with six variable-position bolted connections. The main database comprises 128 configurations (121 geometrically distinct layouts), for which the first three natural frequencies were obtained using ANSYS Workbench. The dataset also includes 10 independent FEM test configurations, not included in the original database, which were used to assess the generalization capability of ANN surrogate models for previously unseen fastening layouts. The dataset contains the fastening configurations and corresponding FEM natural frequencies and supports the reproducibility and further development of research on structural dynamics, finite element modelling, and machine-learning-based natural-frequency prediction.
Files
Steps to reproduce
Create a 3000 × 2000 × 4 mm steel plate model using the material properties specified in the associated publication: Young’s modulus \(E=2\times10^5\) MPa, density \(\rho=7850\) kg/m³, and Poisson’s ratio \(\nu=0.30\). Define the Cartesian coordinate system with the origin at the lower-left corner of the plate. Define the six bolted connections according to the coordinate and numbering convention described in the publication. Generate the 128 fastening configurations according to the three configuration groups and bolt-position rules described in Section 2.1 of the publication. For each configuration, create a 20 × 20 mm imprinted square footprint at each fastening location and apply a fixed support to the corresponding faces. Generate the finite element mesh using a global element size of 20 mm and the quadrilateral-dominant meshing method. Perform a linear, undamped modal analysis in ANSYS Workbench without prestress or bolt preload. Extract the first 20 natural modes and retain the first three natural frequencies for each configuration. Repeat the same procedure for the 10 independent test configurations, which are not included in the original 128-case database. Use the resulting 128-case database as the reference dataset for ANN development and use the 10 additional FEM configurations as an independent test set for evaluating ANN generalization to previously unseen fastening layouts.