Neural network AE source location apart from structure size and material

0349882 - ÚT 2011 RIV AT eng C - Konferenční příspěvek (zahraniční konf.)
Chlada, Milan - Převorovský, Zdeněk - Blaháček, Michal
Neural network AE source location apart from structure size and material.
29th Europen Conference on Acoustic Emission Testing. Vienna: TÚV Austria, 2010, s. 1-8. ISBN 978-3-200-01956-0.
[European Conference on Acoustic Emission Testing 2010 /29./. Vídeň (AT), 08.09.2010-10.09.2010]
Grant CEP: GA MPO(CZ) FR-TI1/274; GA ČR(CZ) GAP104/10/1430
Výzkumný záměr: CEZ:AV0Z20760514
Klíčová slova: AE source location * artificial neural networks * arrival time profiles
Kód oboru RIV: BI - Akustika a kmity
http://www.2010.ewgae.eu/

AE localization procedures using artificial neural networks (ANN) represent extremely effective alternative to classical triangulation methods. Nevertheless, their application always requires full-scale, time consuming ANN training on each specific structure. Disadvantage of particularly trained ANN algorithm is in its non-transferability to any other object. A new ANN-based AE source location approach is proposed in this paper to overcome such limitation. The method replaces standard arrival time differences at the ANN inputs by so called signal arrival time profiles, independent on material and scale changes. The ANN training can be also performed theoretically on geometrical models and learned ANN is then applied on real structures with different dimensions and materials. Such approach enables considerable extension of ANN application possibilities. The use of new AE source location method is illustrated on experimental data obtained during aircraft structure part testing.
Trvalý link: http://hdl.handle.net/11104/0190010