System Identification Using Neural Network and Computational Mechanics. 2nd Report. Reduction of Necessary Response Measurement Points.

Bibliographic Information

Other Title
  • ニューラルネットワークと計算力学に基づくシステム同定の検討 第2報  計測点の低減法に関する研究
  • ニューラルネットワーク ト ケイサン リキガク ニ モトズク システム ドウテ

Search this article

Description

Computer simulation with finite element model (FEM) plays more and more important role in design stage. For the sake that there are always some differences in physical parameters from those of real structure, it is difficult to make an accurate model actually. This paper contributes the approach to detect the inconsistency between the real structure and the FEM efficiently, only a few necessary measurement locations are required with the sensitivities of eigen frequency and response of structure. Changes in structural parameters, induced by damage, affect the eigensolution matrices and may cause the change of eigen frequency and response of structure. Use of these values to select modes and measurement points for use in damage location or sensor placement studies is proposed and demonstrated by application to a simply supported plate. Learning Vector Quantization (LVQ) Neural Network besed on pattern classifier is used to detect the location of damage. Results show that the LVQ has the superiority to Holographic Neural Network in pattern classification.

Journal

Citations (2)*help

See more

References(21)*help

See more

Related Projects

See more

Details 詳細情報について

Report a problem

Back to top