Abstract of Thesis presented at COPPE/UFRJ as a partial fulfillment of the requirements for the degree of Doctor of Science (D.Sc.)

Monitoring of Defects of Rigid Pipes by Analyzing of Features Acousting Emission Using Neural Networks

Carlos Fernando Carlim Pinto

December/2014

Advisors:

Luiz Pereira Calôba Romeu Ricardo da Silva

Department:

Eletrical Engineering

      The real-time pipeline transportation of petroleum products monitoring has become increasingly important, especially when intended for operational security. Acoustic emission tests (AE) are applied to the inspection of various kinds of equipment. This thesis presents a proposal for a study on the use of acoustic emission to detect the propagation of defects in pressurized rigid pipes. The resulting signals were classified as: No Propagation (NP), Stable Propagation (SP) and Unstable Propagation (UP) and used as data in the creation of non-linear classifiers. Classifier performance reached about 100%, proving the efficiency of the method under the conditions tested in this study. From classifiers built are presented the methodologies used for the construction of the curve Possibility of Propagation (PoP) which will make it possible predict when the spread of crack will became a unstable propagation.