Abstract:Additive manufacturing (AM) is an emerging method for cost-efficient fabrication of nuclear reactor parts. AM of metallic structures for nuclear energy applications is currently based on laser powder bed fusion (LPBF) process, which can introduce internal material flaws, such as pores and anisotropy. Integrity of AM structures needs to be evaluated nondestructively because material flaws could lead to premature failures due to exposure to high temperature, radiation and corrosive environment in a nuclear reactor. Quality control (QC) requires nondestructive evaluation (NDE) of actual AM structures. Pulsed thermography is a potentially promising QC technique because it is scalable to arbitrary structure size. However, detection sensitivity of this method is limited by noises. We investigate separation of signal from noise in thermography images using several machine learning (ML) methods, including new spatio-temporal blind source separation (STBSS) and spatio-temporal sparse dictionary learning (STSDL) methods. Performance of the ML methods is benchmarked using thermography data obtained from imaging stainless steel 316L and Inconel 718 specimens produced LPBF method with imprinted calibrated porosity defects. The ML methods are ranked by F-score and execution runtime. The ML methods with higher accuracy require longer run time. However, this runtime is sufficiently short to perform QC within a realistic time frame. specimen with the flash lamp. In addition, because detection of flaws is based on heat diffusion, contrast of “hot spots” is reduced due to blurring. In this paper, we investigate separation of signal from noise to enhance material flaw detection in thermography images using several machine learning (ML) methods. Performance of the ML methods is evaluated using the data cube obtained from flash thermography imaging of AM specimens. The specimens used in this study are SS316L and INC718 plates produced with LPBF method, which contain imprinted calibrated porosity defects of various sizes. The ML methods developed and demonstrated in this paper are Spatial-Temporal Blind Source Separation (STBSS) and Spatial-Temporal Sparse Dictionary Learning (STSDL). The STBBS method is developed by combining wavelet-based de-noising with neural learning-based Blind Source Separation (NLBSS) algorithm. The NLBSS method developed recently for ML of thermography images, is based on combining the principal component analysis (PCA) and independent component analysis (ICA), which is implemented as a two-layer neural network. The STSDL method is similarly developed by combining wavelet-based de-noising with the sparse dictionary learning (SDL) method. Performance of the ML methods, as well as the previously developed NLBSS, and established PCA, ICA and SDL methods in detection of flaws in flash thermography data is ranked by F-score and algorithm execution runtime. Results of this paper indicate that STBSS and STSDL outperform existing methods in their capability of defects detection, albeit at the expense of increased runtime. However, the time to perform flash thermography measurements and ML analysis of images is sufficiently short for implementation of these methods in a realistic QC procedure.

Towards the Automation of Non-destructive Fault Recognition: Enhancement of Robustness and Accuracy Through AI Powered Acoustic and Thermal Signal Analysis in Time, Frequency- and Time-Frequency Domains

Two-stream Convolutional Neural Network for Non-Destructive Subsurface Defect Detection Via Similarity Comparison of Lock-in Thermography Signals

A Review of Artificial Intelligence Applications in Thermographic Non-Destructive Testing

Fault detection for sliding bearings using acoustic emission signals and machine learning methods

Towards Interpretable Machine Learning for Automated Damage Detection Based on Ultrasonic Guided Waves

Detection of Defects in Additively Manufactured Metallic Materials with Machine Learning of Pulsed Thermography Images

Spatial and Temporal Deep Learning in Air-Coupled Ultrasonic Testing for Enabling NDE 4.0

Automated Flaw Detection in Multi-channel Phased Array Ultrasonic Data Using Machine Learning

Laser Ultrasonics and Machine Learning for Automatic Defect Detection in Metallic Components

Automated Detection of hidden Damages and Impurities in Aluminum Die Casting Materials and Fibre-Metal Laminates using Low-quality X-ray Radiography, Synthetic X-ray Data Augmentation by Simulation, and Machine Learning

A study of machine learning object detection performance for phased array ultrasonic testing of carbon fibre reinforced plastics

Augmented Ultrasonic Data for Machine Learning

Reinforcement Learning Based Neural Architecture Search for Flaw Detection in Intelligent Ultrasonic Imaging NDE System

Real-Time Hollow Defect Detection in Tiles Using On-Device Tiny Machine Learning

A Reliability Study on Automated Defect Assessment in Optical Pulsed Thermography

An Automated Machine Learning Approach for Real-Time Fault Detection and Diagnosis

A CNN architecture for tool condition monitoring via contactless microphone: regression and classification approaches

Machine Fault Diagnosis through Vibration Analysis: Continuous Wavelet Transform with Complex Morlet Wavelet and Time–Frequency RGB Image Recognition via Convolutional Neural Network

The Effect of Different Flaw Data to Machine Learning Powered Ultrasonic Inspection

Spatial and temporal deep learning for defect detection with lock-in thermography

A Novel Approach for Real-Time Quality Monitoring in Machining of Aerospace Alloy through Acoustic Emission Signal Transformation for DNN