[2A5] AI and edge computing assisted manual phased array inspection
O Jessen-Juhler, T Koskinen and I Virkkunen
Trueflaw Ltd, Finland
Phased array inspections have become more common in ultrasonic inspection. Especially in high-risk inspections, saving the scanned data has become the de-facto method of conducting inspections. This is valuable since it allows more in-depth data analysis as well as gathering more channels and angles with the phased array set-up.
Unfortunately, the usage is commonly restricted to simple geometries and easily accessible locations due to the requirement of mechanisation. It cannot fully take advantage of ample data the technique could offer, since the inspector cannot inspect multiple channels of data at once.
This study demonstrates how artificial intelligence (AI) can use ultrasonic data as an encoder, thus allowing manual recording of data. As the manual movement by hand has varying speed and high speed, the data processing needs to be done with an edge computing unit to allow fast data recording, analysis of the ultrasonic encoding data and adding AI for the data analysis on the fly.
This enables the inspection to be conducted by a human with fewer qualifications than normally required, increases the inspection speed as the set-up time for mechanisation is no longer required and increases the inspection reliability by increasing the amount of data from the inspection.
Unfortunately, the usage is commonly restricted to simple geometries and easily accessible locations due to the requirement of mechanisation. It cannot fully take advantage of ample data the technique could offer, since the inspector cannot inspect multiple channels of data at once.
This study demonstrates how artificial intelligence (AI) can use ultrasonic data as an encoder, thus allowing manual recording of data. As the manual movement by hand has varying speed and high speed, the data processing needs to be done with an edge computing unit to allow fast data recording, analysis of the ultrasonic encoding data and adding AI for the data analysis on the fly.
This enables the inspection to be conducted by a human with fewer qualifications than normally required, increases the inspection speed as the set-up time for mechanisation is no longer required and increases the inspection reliability by increasing the amount of data from the inspection.
