Computational intelligence approaches for data analysis, the next step of innovation for advanced UT techniques in NDT

Gonzalez Rodriguez, Alicia; TWI Ltd; United Kingdom

Gonzalez Rodriguez, A.; TWI Ltd; Mexico
Gosselin, A.; Ondia; Canada
Rhéaume, R.; Ondia; Canada
Harrap, N.; TWI Ltd; United Kingdom

ID: ECNDT-0628-2018
Download: PDF

There is no doubt in how the advanced ultrasonic testing (UT) techniques such as Phased Array Ultrasound (PAUT) and Time of Flight Diffraction (ToFD) have taken over the non-destructive testing industry (NDT). The benefits of using these techniques are countless. Worldwide, there has been an increase in companies from different sectors taking advantage of the available technology for these methodologies. As a result, the equipment is becoming more affordable, data processing for real time and automated applications is more powerful and the amount of data being produced is substantially high. Regarding this last subject, nowadays, the data acquired relies mainly on humans for its analysis, assessment and decision-making which are critical tasks that humans are more than capable to execute. However, when the data available for PAUT and ToFD goes from one simple data set (scan) to hundreds of data sets for just a single project, critical errors can occur.

For most NDT specialists, interpretation of signals as well as analysis of data in large amounts can lead to mistakes such as missing defects, incorrect sizing or false identification of a defect. Furthermore, the amount of time operators dedicate in data analysis is increasing for these advanced UT techniques. The NDT industry relies on consistent and reliable data analysis which, with the technologies available now, is incompatible with the amount of data to be analysed. For these reasons, Ondia and TWI have gathered a group of specialists in NDT, software development and artificial intelligence to implement advanced Computational Intelligence (CI) techniques and develop algorithms to better harness the data available in order to enhance the quality of data analysis. This innovation will reduce the time invested in a single analysis, whilst addressing the challenges that large data analysis and processing bring about for advanced NDT methodologies.