|
|
Application of Arrival Time Profiles to AE Source Location by Neural Networks
Chlada, Milan ; Blaháček, Michal ; Převorovský, Zdeněk
The localization procedures using artificial neural networks (ANN) represent today highly effective, alternative approach to classical triangulation algorithms. Nevertheless, their application possibilities are limited due to several reasons. The main problems are in the collecting of sufficiently extensive training and testing data sets together with the non-portability of particular trained network to any other object. In recent time, a new ANN-based AE source location method using so-called signal arrival time profiles was proposed to overcome both limitations. The new way of signal arrival time characterization provides the ANN training on numerical models and allows the application of learned ANN on real structures of various scales and materials. In the paper, this new method is illustrated on experimental data obtained at complex aircraft structure part testing, and its remarkable advantages concerning the considerable extension of ANN application possibilities are discussed.
|
|
|
Lokalizace zdrojů akustické emise pomocí neuronových sítí na základě časových profilů
Chlada, Milan ; Blaháček, Michal ; Převorovský, Zdeněk
Correct localization of acoustic emission (AE) sources is a basic requirement in AE analysis and consequent evaluation of damage mechanism. The localization procedures using artificial neural networks (ANN) represent today highly effective, alternative approach to classical triangulation algorithms. Nevertheless, their application possibilities are limited due to problematic collecting of sufficiently extensive training and testing data sets together with the non-portability of particular trained network to any other object. A new ANN-based approach, using so-called signal arrival time profiles, is proposed to overcome both limitations. Such approach provides the ANN training on numerical models and allows the application of learned ANN on real structures of various scales and materials. This enables considerable extension of ANN application possibilities. New method is illustrated on experimental data obtained during pen-tests on a steel plate, and its remarkable advantages are discussed.
|
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
Vybrane vlastnosti atributy anotovanych dat
Řimnáč, Martin
The paper compares features of learning and querying process in the situation, when values in the input data set are annotated by attributes or this information is not available. The attribute annotation enables to consider global relationships, which are useful to express the data semantics in a explicit way. It will be shown data can be accessed with no semantic interpretation and then, after the evaluation process, the result can be interpreted.
|
|
| |
|
| |
guest ::
