Flexibility for extended simulation realism: Application to the simulation of mechanical behavior of beams

  • Authors:
  • Keny Ordaz-Hernandez;Xavier Fischer;Fouad Bennis

  • Affiliations:
  • ECN IRCCyN - UMR CNRS 6597, Ecole Centrale de Nantes, 1, rue de la Noë/ - BP 92101, 44321 Nantes Cedex 03 and ESTIA LIPSI/ Technopô/le Izarbel, 64210 Bidart, France;(Correspd. E-mail: x.fischer@estia.fr) ESTIA LIPSI/ Technopô/le Izarbel, 64210 Bidart and TREFLE - UMR CNRS 8508 ENSAM, Esplanade des Arts et Mé/tiers, 33405 Talence Cedex, France;ECN IRCCyN - UMR CNRS 6597, Ecole Centrale de Nantes, 1, rue de la Noë/ - BP 92101, 44321 Nantes Cedex 03, France

  • Venue:
  • Integrated Computer-Aided Engineering
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This document presents a technique developed to switch between different behavioral models in the interactive simulation of a mechanical object. When conducting interactive simulations several behavioral models may be required, due to limitations of the models' validity domain. In other cases, alternative models are developed to improve the accuracy or the precision of the estimated response. Smaller time of response is also another feature that fosters model creation. In this work, an all-available model approach is proposed to perform interactive simulations where using only one model during the simulation may not be desirable, since its validity domain is finite and its advantages may appear only in a specific range of behavior. This paper proposes coupling several available models to cover a reasonable validity domain, and in particular, to improve the overall time of response. A selector, based on the Takagi-Sugeno-Kang method of inference, is constructed to dynamically switch between models, accordingly to the environment's status. This flexible model, created from a collection of sub-models, may also adapt itself to manage the accuracy/precision-time relation. The concept has been implemented in the simulation of the mechanical response of a simple structure, and performed better than the sub-models that it integrates.