NDT of Railway
| METADEX Antwortnummer 1 - © 1998 CSA | |
| Title | |
| Ultrasonic non-destructive control on the vitalization elements of railway locomotive with the use of neural networks. | |
| Author | |
| Jemec, V. (Srednja sola Domzale); Peek, B.L. (Iskraemeco Kranj) | |
| Publication Source | |
| Slovenian Society for Nondestructive Testing. Askerceva 6, Ljubljana, 1000, Slovenia. 1997. 1-8, Graphs, 11 ref. Conference: Application of Contemporary Nondestructive Testing in Engineering, Ljubljana, Slovenia, 24-25 Apr. 1997 ISBN: 961-6238-04-3 | |
| Document Type | |
| Conference Article | |
| Country of Publication | |
| Slovenia | |
| Language | |
| English | |
| Abstract | |
| Ultrasonic control as one of the non-destructive testing methods has lately reached a great progress in development of the instruments for measuring, saving and processing ultrasonic signals. A new intermediate device has been made to assure the safety of railway traffic. With the help of software we record ultrasonic signals in the process of measuring the railway shafts axis and other elements of railway vehicles. The interface can be used for recording and saving ultrasonic signals. It has also the ability of controlling the activity of ultrasonic probes which is provided by standard ultrasonic device. While the first system was able to record 10 signals, the new improved system can record 200 signals in the same time. The interface is smaller and much more practical. The computer program for monitoring of ultrasonic measuring has plenty menus that enable easier work for data processing. Therefore we can say that the program is quite user-friendly. The developed hardware for receiving the ultrasonic signals and software for monitoring and storing of ultrasonic signals of the appropriate form constitute the support for manual control of the testing material quality. The next step is to develop the automatic system for controlling the quality of the testing material. In recent years it has been proved that this task is most efficiently solved by artificial neural networks. This article describes the radial basis function neural network for recognition of measuring conditions. With the help of experts and with support of developed hardware and software it will be later completed for solving the problem of defect recognition. | |
| Accession Number | |
98(1):22-64 | |