Optimizing input data for training an artificial neural network used for evaluating defect depth in infrared thermographic nondestructive testing

Чулков, А. О.; Нестерук, Д. А.; Вавилов, В. П.; Moskovchenko, Alexey; Saeed, Numan; Omar, Mohammed

doi:10.1016/j.infrared.2019.103047

Cited by 24 publications

(7 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The characterization errors were in the range of 6 to 31%, and in the most cases did not exceed 20%l. This is in good agreement with earlier published results [14,15,31,34,35].…”

Section: Data Processing and Nn Trainingsupporting

confidence: 93%

“…For creating and training the NN, the TensorFlow algorithm set was used, and the NN training quality was estimated by applying a loss function. It was shown elsewhere that an increase in the number of NN hidden layers improves the accuracy of evaluating the depth of deeper defects [35]. In this study, the criterion for choosing the number of layers was a compromise between the accuracy of defect characterization and computation time.…”

Section: Data Processing and Nn Trainingmentioning

confidence: 94%

See 1 more Smart Citation

A method and apparatus for inspecting large flat composite parts by using the Infrared LST technique

Вавилов

Чулков

Нестерук

et al. 2022

Frattura Ed Integrità Strutturale

View full text Add to dashboard Cite

The principle of Line Scan Thermography (LST) was used to develop a self-propelled infrared thermographic nondestructive testing device for the inspection of large, relatively flat composite aerospace parts, such as aircraft wings. The design of the unit allowed the suppression of noise from reflected radiation. The new equipment, using the LST method, provided defect detectability similar to that achieved with a classic, static, flash heating procedure, but with a higher rate of testing. Also, the line heating principle ensured more uniform thermal patterns, and the proper choice of scan speed and field of view allows the selection of optimal time delays and the creation of maps of defects at different depths. Defect characterization efficiency was improved by using a trained neural network.

show abstract

“…The characterization errors were in the range of 6 to 31%, and in the most cases did not exceed 20%l. This is in good agreement with earlier published results [14,15,31,34,35].…”

Section: Data Processing and Nn Trainingsupporting

confidence: 93%

Section: Data Processing and Nn Trainingmentioning

confidence: 94%

A method and apparatus for inspecting large flat composite parts by using the Infrared LST technique

Вавилов

Чулков

Нестерук

et al. 2022

Frattura Ed Integrità Strutturale

View full text Add to dashboard Cite

show abstract

“…Previous studies have demonstrated that needless data could restrict the effectiveness of model training (Chulkov et al, 2019;Cova and Pais, 2019;Xu et al, 2019). In contrast, in the classification model, the optimal size (m, n) of the input varied depending on the structure based on whether the CNN structures adopted a skip connection or a parallel CNN layer (He et al, 2016b;Szegedy et al, 2016).…”

Section: Optimal Input Designmentioning

confidence: 99%

Deep Learning for Simulating Harmful Algal Blooms Using Ocean Numerical Model

et al. 2021

View full text Add to dashboard Cite

In several countries, the public health and fishery industries have suffered from harmful algal blooms (HABs) that have escalated to become a global issue. Though computational modeling offers an effective means to understand and mitigate the adverse effects of HABs, it is challenging to design models that adequately reflect the complexity of HAB dynamics. This paper presents a method involving the application of deep learning to an ocean model for simulating blooms of Alexandrium catenella. The classification and regression convolutional neural network (CNN) models are used for simulating the blooms. The classification CNN determines the bloom initiation while the regression CNN estimates the bloom density. GoogleNet and Resnet 101 are identified as the best structures for the classification and regression CNNs, respectively. The corresponding accuracy and root means square error values are determined as 96.8% and 1.20 [log(cells L–1)], respectively. The results obtained in this study reveal the simulated distribution to follow the Alexandrium catenella bloom. Moreover, Grad-CAM identifies that the salinity and temperature contributed to the initiation of the bloom whereas NH4-N influenced the growth of the bloom.

show abstract

“…In contrast, indirect detection requires the adoption of specific devices to measure material deterioration using different nondestructive technologies (e.g., eddy current, infrared, and ultrasonic thickness testing) [10,11]. The main advantage of indirect detection nondestructive testing (NDT) is a higher probability of detection and quantification of corrosion.…”

Section: Introductionmentioning

confidence: 99%

Aircraft Fuselage Corrosion Detection Using Artificial Intelligence

Brandoli

Geus

Souza

et al. 2021

Sensors

View full text Add to dashboard Cite

Corrosion identification and repair is a vital task in aircraft maintenance to ensure continued structural integrity. Regarding fuselage lap joints, typically, visual inspections are followed by non-destructive methodologies, which are time-consuming. The visual inspection of large areas suffers not only from subjectivity but also from the variable probability of corrosion detection, which is aggravated by the multiple layers used in fuselage construction. In this paper, we propose a methodology for automatic image-based corrosion detection of aircraft structures using deep neural networks. For machine learning, we use a dataset that consists of D-Sight Aircraft Inspection System (DAIS) images from different lap joints of Boeing and Airbus aircrafts. We also employ transfer learning to overcome the shortage of aircraft corrosion images. With precision of over 93%, we demonstrate that our approach detects corrosion with a precision comparable to that of trained operators, aiding to reduce the uncertainties related to operator fatigue or inadequate training. Our results indicate that our methodology can support specialists and engineers in corrosion monitoring in the aerospace industry, potentially contributing to the automation of condition-based maintenance protocols.

show abstract

Optimizing input data for training an artificial neural network used for evaluating defect depth in infrared thermographic nondestructive testing

Cited by 24 publications

References 10 publications

A method and apparatus for inspecting large flat composite parts by using the Infrared LST technique

A method and apparatus for inspecting large flat composite parts by using the Infrared LST technique

Deep Learning for Simulating Harmful Algal Blooms Using Ocean Numerical Model

Aircraft Fuselage Corrosion Detection Using Artificial Intelligence

Contact Info

Product

Resources

About