P.P. van’t Veen, J. Jarmulak, “Automated Interpretation of Heterogeneous NDT Measurements”, in: 7th ECNDT Proceedings, Vol. 3, 1998, pp. 2355-2362. (pdf)
In order to tackle the problem of efficient and reliable interpretation of heterogeneous NDT data an approach has been developed based on a combination of rule-based and case-based reasoning. One of the problems tackled by the approach is automated detection and classification of a large number of different events originating from defects, construction elements, and noise sources. A very important aspect of the approach is that only limited a-priori information of the construction and the expected defect indications is necessary. Furthermore, fluctuations in the inspection conditions can be handled. Also, differences between calibration and real defect indications are tackled. In situations where damage rates are high and interference of indications occurs the approach is capable to analyse the signal in considerable detail and provide suggestions to the operator for the classification of the event and further action.
The effectiveness of the approach is demonstrated on two representative NDT techniques: interpretation of data acquired with an ultrasonic rail inspection system and interpretation of eddycurrent data from heat exchangers in (petro-)chemical industry. The results show that it is possible to provide a high level of automation in combination with efficient operator support for highly variable NDT measurements where up to now use of automated interpretation was only limited.