TWI Industrial Membership Report Summary 645/1998
A M Lank and P Wilkinson*
This report describes work carried out under the Core Research Programme to develop a neural network method capable of classifying ultrasonic signals according to flaw type, which could be incorporated into standard digital flaw detection equipment.
Background
Manual ultrasonic inspection is a highly skilled operation and generally it requires a great deal of experience in order to be confident about predicting flaw types from the signals appearing on a flaw detector screen. It is not unknown for there to be disagreement, even between experienced operators, when diagnosing signals from weldments. It is clear that any means of enabling the ultrasonic test technician to achieve better and more consistent results would be beneficial.
When an operator performing a manual ultrasonic test detects a flaw, he will formulate an opinion about the nature of that flaw by scanning the ultrasonic probe over the surface of the workpiece and observing how the A-scan signals change as the ultrasonic beam impinges on that flaw. The ability of a neural network to deal with a variety of inputs, occurring in unpredictable sequences, mirrors the experience-based approach of the operator to ordering the information generated by an ultrasonic test, in order to arrive at a decision about flaw type.
The aim of this work was to explore whether application of neural network processing offers the potential to provide the technician carrying out manual ultrasonic testing with an objective interpretation of the nature of the flaw being examined. Any technique developed would have to be capable of being implemented in standard manual digital flaw detection equipment, via a new software version, for example. This precluded any dependence on the system knowing precise probe location, for example, as generally, such hardware is only attached to automated systems.
Objective
- To develop a flaw characterisation system for manually collected ultrasonic signals capable of discriminating between different types of welder induced flaw, using a neural network based classifier.