Automated defect classification in infrared thermography based on a neural network

Duan, Yuxia; Liu, Shicai; Hu, Caiqi; Hu, Junru; Zhang, Hai; Yan, Yiqian; Tao, Ning; Zhang, Cunlin; Maldague, Xavier; Fang, Qiang; Ibarra‐Castanedo, Clemente; Chen, Dapeng; Li, Xiaoli; Meng, Jianqiao

doi:10.1016/j.ndteint.2019.102147

Cited by 64 publications

(27 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Shepard et al [ 21 ] proposed such fitting several years ago; it is known as the TSR (thermographic signal reconstruction) algorithm, with which thermal data from pulsed thermography experiments are fitted using a logarithmic polynomial function [ 22 ]. This method was described well by Duan [ 20 ] for automated defect classification in infrared thermography based on a neural network [ 20 ].

where T 0 is the initial temperature, e is the material thermal effusivity, Q is the energy density absorbed by the surface, and t is the time after excitation.…”

Section: Resultsmentioning

confidence: 99%

“…Thermographic signal reconstruction (TSR) is a useful processing technique in pulsed infrared thermography as it uses surface temperature evolution based on a one-dimensional solution of the Fourier equation for a Dirac delta function in a semi-infinite isotropic solid as shown in Equation (1) [20,21]. The temperature decays can be fitted with a functional relationship.…”

Section: Defect Analysismentioning

confidence: 99%

“…TSR processing commonly amplifies cooling curve differences through a logarithmic operation. TSR provides significant data compression because it only requires saving six polynomial coefficients per pixel [20,21]. Figure 9 is an example of a one-dimensional approximation; this recognizes that heat diffuses mainly in one direction.…”

Section: Defect Analysismentioning

confidence: 99%

“…Therefore, it assumes that lateral diffusion components, more or less, cancel out in a defectfree setting. In the presence of a non-homogeneous subsurface boundary, incident heat flow from the sample surface is impeded [20,21]. The fourth frame in dataset 2.2CooledFront is where the electromagnetic radiation from the flash hit the specimen, and the damage can be seen instantly.…”

Section: Defect Analysismentioning

confidence: 99%

See 3 more Smart Citations

Comparison of Cooled and Uncooled IR Sensors by Means of Signal-to-Noise Ratio for NDT Diagnostics of Aerospace Grade Composites

Deane

Avdelidis

Ibarra-Castanedo

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

This work aims to address the effectiveness and challenges of non-destructive testing (NDT) by active infrared thermography (IRT) for the inspection of aerospace-grade composite samples and seeks to compare uncooled and cooled thermal cameras using the signal-to-noise ratio (SNR) as a performance parameter. It focuses on locating impact damages and optimising the results using several signal processing techniques. The work successfully compares both types of cameras using seven different SNR definitions, to understand if a lower-resolution uncooled IR camera can achieve an acceptable NDT standard. Due to most uncooled cameras being small, lightweight, and cheap, they are more accessible to use on an unmanned aerial vehicle (UAV). The concept of using a UAV for NDT on a composite wing is explored, and the UAV is also tracked using a localisation system to observe the exact movement in millimetres and how it affects the thermal data. It was observed that an NDT UAV can access difficult areas and, therefore, can be suggested for significant reduction of time and cost.

show abstract

where T 0 is the initial temperature, e is the material thermal effusivity, Q is the energy density absorbed by the surface, and t is the time after excitation.…”

Section: Resultsmentioning

confidence: 99%

Section: Defect Analysismentioning

confidence: 99%

Section: Defect Analysismentioning

confidence: 99%

Section: Defect Analysismentioning

confidence: 99%

See 2 more Smart Citations

Comparison of Cooled and Uncooled IR Sensors by Means of Signal-to-Noise Ratio for NDT Diagnostics of Aerospace Grade Composites

Deane

Avdelidis

Ibarra-Castanedo

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recently, neural networks (NN) have prevailed in pattern recognition [14], automatic control [15], signal processing [16] NNs have also advanced the defect classification and depth determination with infrared thermography NDT [17][18][19][20][21]. The featured parameter of the defects can be extracted from NNs trained with data obtained through simulations using finite element method (FEM) or through experiments.…”

Section: Introductionmentioning

confidence: 99%

Defect Depth Determination in Laser Infrared Thermography Based on LSTM-RNN

et al. 2020

View full text Add to dashboard Cite

Carbon fiber reinforced polymer (CFRP) has been increasingly used in aviation industry since it significantly enhances the performance of aircraft. However, imperfections inside the CFRP structures pose a threat to aviation safety. Apart from the defect shape and size, flaw depth is crucial to assess the defect severity. In this work, we utilize a laser infrared thermography (LIT) system to inspect an aviation CFRP sheet and adopt a long-short term memory recurrent neural network (LSTM-RNN) to determine the defect depth. Thermographic sequences obtained by LIT are processed using thermographic signal reconstructions (TSR) method. Raw data and TSR processed data are separately used to train and test the LSTM-RNN. Results show that background noises in the original thermal signals can be effectively reduced by the TSR method, which is helpful for the models to learn the signal characteristics. Compared with two traditional methods, recurrent neural network (RNN) and convolutional neural network (CNN), we find the LSTM-RNN outperforms.

show abstract