Deep Learning-based Thickness Measurement With Pulse Eddy Current Testing
2023 12th International Conference on Control, Automation and …, 2023•ieeexplore.ieee.org
The Pulse Eddy Current (PEC) technique represents an innovative approach within the field
of eddy current testing (ECT), primarily employed to assess the structural integrity of
engineering components. However, this method typically encounters challenges stemming
from variations in lift-off distances and necessitates a manual calibration procedure prior to
operation. In this research paper, we introduce a novel approach for thickness measurement
that leverages deep learning techniques to enable automatic and highly precise thickness …
of eddy current testing (ECT), primarily employed to assess the structural integrity of
engineering components. However, this method typically encounters challenges stemming
from variations in lift-off distances and necessitates a manual calibration procedure prior to
operation. In this research paper, we introduce a novel approach for thickness measurement
that leverages deep learning techniques to enable automatic and highly precise thickness …
The Pulse Eddy Current (PEC) technique represents an innovative approach within the field of eddy current testing (ECT), primarily employed to assess the structural integrity of engineering components. However, this method typically encounters challenges stemming from variations in lift-off distances and necessitates a manual calibration procedure prior to operation. In this research paper, we introduce a novel approach for thickness measurement that leverages deep learning techniques to enable automatic and highly precise thickness assessment. Our proposed solution encompasses an Auto-Encoder model, incorporating convolutional layers and long short-term memory layers, designed for effectively learning raw PEC signals and predicting thickness values across varying lift-off conditions. To validate this approach, we conducted experiments focusing on the measurement of thickness in a Mild Carbon Steel specimen. Our results demonstrate that the PEC system successfully predicts thickness values within the range of 1 mm to 10 mm, exhibiting a mean and standard deviation error of the thickness of 0.026 and 0.038 mm, respectively, even when confronted with lift-off variations ranging from 0.5 mm to 2.0 mm.
ieeexplore.ieee.org
Showing the best result for this search. See all results