The sorting of useful and noise signals is the most labor-intensive and critical stage of expert analysis of diagnostic data. Even a single flaw undetected among millions of magnetic images can lead to gas-pipeline explosion, substantial economic damage, and possible fatalities if erroneous identification makes impossible the timely removal of the flaw. Efforts of many specialists are aimed at automation of the identification of useful signals in the overall body of signals. Although this problem is urgent, an analysis of the scientific literature showed an absence of effective solutions to it. A new method of identification of useful and noise signals produced during magnetic inspection of gas pipelines is reported. Qualitative classification characteristics of the most common types of noise signals have been found. For the characteristics found, quantitative descriptions that are invariant with respect to rotation and scaling transformations were devised and decision rules were formulated. The use of the found characteristics ensures the minimum difference between results of automated and visual interpretation of data. Results of commercial use of this method are reported.
Determination of classification characteristics of magnetic images for sorting of useful and noise signals during flaw detection in gas pipelines
Bestimmung von Klassifikationsmerkmalen magnetischer Abbildungen zur Unterscheidung von nützlichen und von Rauschsignalen beim Fehlernachweis in Gaspipelines
Russian Journal of Nondestructive Testing ; 44 , 11 ; 787-791
2008
5 Seiten, 2 Bilder, 2 Quellen
Article (Journal)
English
Ultrasonic flaw detection during NDE of oil pipelines via a resonance filter
Tema Archive | 2006
|European Patent Office | 2021
|