For a reliable acoustic emission (AE) monitoring of petrochemical storage tanks, collection of AE events for several hours under quiet environment is required. Then the diagnosis is based on the classification of the recorded data. For a good classification, an improved phenomenon identification and separation is needed. Commonly, the methods of classification mainly used are: neural network classification and pattern-recognition classification. However, the questions are: what is the uncertainty rate of each classification and what is their success rate in the phenomenon identification. This work consists of a comparison between a neural network classification and a pattern-recognition classification of AE events collected from laboratory tests. For this, several new and old corrosion tests have been carried out in laboratory. The new corrosion tests have been realized on a low-carbon steel used in crude-oil manufacturing. The old corrosion tests have been carried out on steel specimens cut from a real storage tank. Finally, AE from leaks is simulated on the bottom of a mini-structure, which stores liquid (water and diesel oil), producing AE through turbulent flow.


    Access

    Access via TIB

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Comparison between a neural networks and pattern recognition classification


    Contributors:


    Publication date :

    2008


    Size :

    6 Seiten, 8 Bilder, 1 Tabelle, 9 Quellen



    Type of media :

    Conference paper


    Type of material :

    Print


    Language :

    English




    Learning Flexible Neural Networks for Pattern Recognition

    A. Mirzaaghazadeh / H. Motameni / M. Karshenas et al. | BASE | 2007

    Free access


    Pattern Recognition using Fourier Descriptors and Neural Networks

    Haynes, B. P. / Sanders, D. A. / Decker, P. et al. | British Library Conference Proceedings | 1994



    Lattice Algebra Approach to Neural Networks and Pattern Classification

    Ritter, G. X. / Iancu, L. | British Library Online Contents | 2004