Amiral Technologies is proud to be one of the winners of the competition: https://cache.media.enseignementsup-recherche.gouv.fr/file/Innovation/41/5/palmares_concours_i-nov-v6_1415415.pdf Waves 5 and 6 of the i-Nov Competition: 131 winners Co-piloted
Crack detection of pipes
Energy infrastructure is critical in maximizing production performance, any crack or defect may reduce performance at best or lead to a significant danger at worse.
Customer uses Non Destructive Testing (NDT) methods based on Eddy Currents to monitor pipelines surfaces. Measurements analysis to classify healthy and non healthy pipelines sections takes huge amount of experts time that could be better used to focus on actual defects/cracks analysis.
DiagFit was used to learn from NDT measurements on a healthy part of the infrastructure (unsupervised learning) to build a predictive model, and then was able to run the model to classify thousands of other measurements as healthy/faulty
100% true positive rate was obtained on historical data with only 17% of false positive rate. The predictive model proved to be even more useful in correcting initial expert labels who finally agreed with Diagfit model diagnosis after re-analysis of some conflictual assessments.