Machine learning based simulation of realistic signals for an enhanced automatic diagnostic in non-destructive testing applications

Model-based solutions for automatic diagnostics in the field of non-destructive testing are currently a topic of great interest in both academic and industrial communities. Their ultimate objective is to provide a qualitative or quantitative evaluation of the inspected material state (sound, flawed, flawed with anomaly dimensions or criticality) in an industrial context like a production line. Such tools, providing inputs for real-time process control, contribute to the general trend in Europe that aims at modernizing industry and services. The CEA LIST Institute is an internationally recognized research institution in non-destructive testing and evaluation (NDT&E). It develops the CIVA software, which offers multi-physics models and is considered a leading product for simulation for NDT&E applications. Accurate models able to reproduce experimental signals prove very helpful in an inversion process aiming at classifying or characterizing flaws. However, as they do not account for disturbances and parameter variability occurring during an experimental acquisition, simulated signals inherently look "perfect" and are, for instance, easily distinguishable from experimental data. This PhD subject aims to improve the match between simulation and experimental data by augmenting the simulation with another contribution generally referred to as "noise". The strategy proposed to obtain such noise contribution is to apply machine-learning techniques to a set of representative experimental data. Alternatively, a deep learning model can be trained to analyze "real" data and distinguish between contents (flaw signals) and style (the rest, which physical models do not simulate). Afterwards, the augmented simulation tool will be able to reproduce closely experimental data, account for specific discrepancies due to a particular environment and reproduce the variability observed experimentally. It will thus enhance the performance of model-based tools developed at CEA LIST for sensitivity analysis, management of uncertainty and diagnostic.