Advanced use of soft computing and eddy current test to evaluate mechanical integrity of metallic plates |
| |
Authors: | Matteo Cacciola Fabio La Foresta Francesco Carlo Morabito Mario Versaci |
| |
Affiliation: | aUniversitá Mediterranea degli Studi di Reggio Calabria, Via Graziella Feo di Vito, 89100 Reggio Calabria, Italy |
| |
Abstract: | The up-to-date structural designing makes by now widely use of high performance numerical codes, mainly in terms of computational powerful, cost and sizing, only available till some time before to limited groups of users. This allowed the experts to focus their attention on a qualifying aspect of the designing, i.e. an use of the materials very close to their limit behavior. Late innovative approaches in material mechanics gave in addition the opportunity to build models very close to the actual behavior but without introducing heavy computational aspects. In this paper, phenomena which relate mechanical stresses with electromagnetic properties of a defined material have been exploited in order to reconstruct electromagnetic maps starting from mechanical quantities by means of support vector regression machines (SVRMs). Purpose of the proposed study is to reconstruct a stress map in strained metallic plates by using electromagnetic measures. Moreover, an heuristic approach is proposed in order to estimate electromagnetic behavior of a stressed plate starting from easily measurable mechanical quantities. It would be very interesting when electrical or mechanical measurements are very hard to realize. The proposed approach could be very useful in such situations as quality controls of civil buildings, without the necessity of applying expensive and time-consuming destructive or non-destructive testing. In this way, it is possible to have a substantially precise idea of mechanical stresses in metallic materials by estimating the local variation of electromagnetic field into the same material using a SVRM-based interpolator. |
| |
Keywords: | Eddy currents Integrity of bi-phase materials Support vector regression machines |
本文献已被 ScienceDirect 等数据库收录! |
|