Signal processing algorithms for acoustic emissions

• Grant number: 17/20933-2
• Support type: Scholarships in Brazil - Master
• Effective date (Start): February 01, 2018
• Effective date (End): September 30, 2019
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Vitor Heloiz Nascimento
• Grantee: Carlos Augusto Prete Junior
• Home Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

There is nowadays a deep interest in the development of techniques that allow continuous monitoring of a mechanical structure (structural health monitoring - SHM). The method of acoustic emissions for detection of faults is one of the main techniques used for SHM. The method has great sensitivity, but there is room for improvement of the algorithms for detection of faults, for estimation of their position, and for automatic classification of the type of fault. In an acoustic emissions essay, several sensors are spread around the test object's surface. When the material undergoes an irreversible change in its internal structure (for example, formation or growth of a crack), a signal is propagated along the material, which allows the localization of the fault (through measurement of the time difference of arrival between the sensors), and classification of the kind of fault (crack, rivet breaking, delamination, noise, etc.) For the correct localization and classification of the fault, one must have reliable algorithms for detection and grouping of the signals arriving at at each sensor (i.e., the algorithms must decide which signals received at each sensor correspond to the same event.)We intend to develop efficient algorithms for hit detection, extraction and grouping. The goal is to develop algorithms with low computational cost (adequate for future hardware implementation) and with better performance, as compared to available algorithms. In particular, we propose to use all the information available: the geometry of the specimen, positions of rivets, differences in signal propagation speed, as well as advanced signal processing methods, such as sampling based on sparse estimation.This work has partial support from EMBRAER, furthering a partnership begun in 2014. This will allow the use of real data, provided by EMBRAER, to test the performance of the algorithms. (AU)