Creating and sharing knowledge for telecommunications

Project: Detection, Classification and Estimation of Defects in Metallic Plates Subject to Manual Inspection

Acronym: CLASSE
Main Objective: Non destructive testing (NDT) is of paramount importance for ensuring the integrity of materials. Eddy currents testing (ECT) is one of several methods like, remote field testing (RFT), flux leakage and Barkhausen noise, that use electromagnetic theory to examine conducting materials. It is a particularly attractive test because it offers both very high detectability and high scanning speeds. Besides direct contact with the test material is not necessary.

Nowadays, with the use of microprocessor based measuring instruments, the potential and user-friendliness of eddy currents testing has been greatly enhanced. The role of the computer has two different components: one related with the possibility to automate the measuring system (data acquisition control, data processing, signal classification and displaying) and other with numerical modeling of the testing process. The computing capabilities are used to quantitatively predict eddy current signals for realistic situations and to evaluate crack profiles from testing signals by solving the inverse problem of the physical phenomenon under test.
At this time the tasks related to the data acquisition, signal processing and graphical representation, are already under way, with some work already published [1-6,8,9].

The point is that the inverse problem of computing the crack profiles from the eddy currents data is ill-posed, i.e., the crack profiles are not univocally determined by the measured eddy currents. A possible approach to mitigate the ill-posedness nature of this inverse problem is to classify the crack profiles according to the different types of material defects. The idea is to exploit the a priori knowledge of the problem and thus select the proper defect model. In order to investigate the assumptions of the technique, particularly, patterns identification, a huge amount of data is essential and experimentation is necessary.
Reference: P247
Funding: IT/LA
Start Date: 01-12-2008
End Date: 01-12-2010
Team: Artur Fernando Delgado Lopes Ribeiro, Helena Maria dos Santos Geirinhas Ramos, Francisco Andre Correa Alegria, Octavian Adrian Postolache, Mário Alexandre Teles de Figueiredo, José Manuel Bioucas Dias
Groups:
Partners:
Local Coordinator: Artur Fernando Delgado Lopes Ribeiro
Links: Internal Page
Associated Publications