Reduction method based structural model updating method via neural networks

摘要

Model updating methods for structural systems have been introduced in various numerical processes. To improve the updating method, the process must require an accurate analyses and minimized experimental uncertainties. In previous study, an identification method has been introduced and verified by measured experiment data of perturbed plate models. Iterated improved reduced system has been employed to the identification and updating method. Perturbed locations and thicknesses were identified accurately through the introduced identification method. To improve the updating method, in this study, we have devised methods to identify more diverse models. In order to detect minute changes in more various models, tetrahedral elements and neural network based updating methods were employed. The previous shell element models were able to identify the thicknesses with relatively thin slenderness. However, the tetrahedral elements can identify the physical properties of the complex models. The accuracy of the updating method for the numerical element models with neural networks was verified through the experimental examples.