Ryszard Sikora was born in Niechobrz, Poland, on October 1st, 1932. He received the B.Sc. in electrical engineering from the Technical University of Szczecin in 1954, M. Sc. from Silesian Technical University of Gliwice in 1956 and Ph. D. from the Technical University of Gdansk in 1967. Since 1955 to 1962 he was employed in Electrical Design Office in Szczecin.

Prof. R. Sikora started to work at Technical University of Szczecin (now Westpomeranian University of Technology – WUT) in 1957, where he works still. He was organizer and long time chairman of Department of Theoretical Electrotechnics and Informatics. He received professor position in 1971. Currently he is full professor in Electrical Engineering and Informatics at WUT. Additionally he was employed in the Academy of Economy and Informatics in Lodz and in Electrical Institute in Warsaw. Since late sixties of XXth century till now he has been leading many Polish and European scientific projects in electrical engineering and non destructive evaluation.

Prof. R. Sikora was member of two committee Polish Ministry of Science and High Education. For many years he acts as elected member of Electrical Engineering Committee of Polish Academy of Science and as a member of editorial board of two Polish scientific journals: Archives of Electrical Engineering and Electrical Review as well as several local and international Scientific Committees. He is author or co-author of three editions of academic book “Electromagnetic Field Theory”, several student textbooks, several academic monographs and several hundred Polish and international scientific papers. Prof. R. Sikora collaborates with many Polish and foreign universities. In 1999 Prof. R. Sikora was elected member of World Federation of Nondestructive Evaluation Center (Iowa State University, USA).

Prof. R. Sikora received several distinctions of Polish states and professional organizations. In 1999 Prof. R. Sikora was awarded by Volta Silver Medal (University of Pavia, Italy) and in November 2007 by JSAEM Award (Japan Society of Applied Electromagnetics and Mechanics). In April 2008 Prof. R. Sikora was honored with Doctor Honoris Causa academic degree of Technical University of Szczecin.

Abstract

Artificial Neural Networks and Fuzzy Logic in Nondestructive Evaluation

The full paper will be carried a overview of artificial intelligence algorithms applicable to nondestructive testing. It focuses on two methods: artificial neural networks and fuzzy logic. Selected examples of application of these methods in digital radiography and eddy current method will be given.

The reliable detection and classification of defects is one of the most important tasks in nondestructive testing (NDT). Usually, trained interpreters evaluate the achieved results of testing. In many cases, it is time and labor intensive work. Human interpretation is subjective, inconsistent, and in many cases biased. The additional problems are caused by the insufficient quality of utilized signals or images. An incorrect classification may reject a part in good conditions or approve a part with defects exceeding the limit defined by the relevant standards. The progresses in computer performance and artificial intelligence algorithms allow to make the process of evaluation automatic and more reliable.

The process of automatic evaluation can be divided into two basic phases: defect extraction and defect identification.

In first step the real defects have to be identified, but usually a number of false alarms can be observed too. Therefore, an additional step is necessary. The identification (second step) allows to separate the existing defects from the regular structures and to reduce the number of false alarms.

The process of defect identification consists of the following operations: feature extraction and classification.
Various methods have been utilized for automated defect detection and identification:

• artificial neural networks (ANN),
• statistical classifiers,
• fuzzy classifiers,
• fuzzy expert systems,
• data fusion algorithms.

The extended review of published works on this subject will be provided in the full version of the paper.

In the paper, selected applications of artificial neural networks and fuzzy classifiers for radiography will be presented. Next, neural models dedicated for eddy current NDT systems will be described. Finally, a solution suitable for difficult cases, with data fusion algorithms is proposed.

Three different applications of ANN and FL will be presented. The achieved results prove that they are already efficient tools. However, the described algorithms and systems will be still enlarged and developed. It is very important to make sure that application of artificial intelligence will be good substitute or help of the NDT operators’ work.

Acknowledgment

This work is partially supported by the Ministry of Education and Science, Poland, under grant N R01 0037 06 (2009-2012).

Literature

1. European Norm EN ISO 6520-1:1998.
2. R. Sikora, P. Baniukiewicz, T. Chady, W. Ruciński, K. Świadek, M. Caryk, P. Lopato, „Comparison of selected weld defects extraction methods”, QNDE, Review of Progress in Quantitative NDE, 2007.
3. N. Nafa, R. Drai and A. Benchalla, Weld Defect Extraction and Classification in Radiographic Testing Based Artificial Neural Networks, WCNDT 2000
4. Chady T., Enokizono M., Todaka T., Tsuchida Y., Sikora R., Cracks detection and recognition by using of multi-frequency signal processing and neural networks, Electromagnetic Nondestructive Evaluation (III), Studies in Applied Electromagnetic and Mechanics, vol. 15, IOS Press 1999, pp. 98-107
5. Ramuhalli P., Afzal M., Hwang K. T., Udpa S., Udpa L., A Feedback Neural Network Approach for Electromagnetic NDE Signal Inversion, Electromagnetic Nondestructive Evaluation (IV),Studies in Applied Electromagnetic and Mechanics, IOS Press 2000, pp. 1-8
6. Chady T., Enokizono M., Sikora R., Todaka T., Y.Tsuchida, Natural Crack Recognition Using Inverse Neural Model and Multi-Frequency Eddy Current Method, IEEE Transaction on Magnetics, vol. 37, July 2001, pp. 2797-2799
7. Yusa N., Cheng W., Chen Z., Miya K., Generalized neural network approach to eddy current inversion for real cracks, NDT&E International, vol 35, 2002, pp. 609-614

Keywords: Automated Defect Recognition, Digital Radiography, Eddy currents, Neural Networks, Fuzzy Logic