Lock-in thermography is a well-established non-destructive technique for detecting defects in composite materials. The qualitative analysis of defects is a challenging task and usually is assessed by an expert operator after the application of suitable algorithms. In this regard, deep learning algorithms are very attractive since they allow to speed up and automatize the identification and characterization of defects. In light of this consideration, the aim of this work is to investigate the influence of lock-in thermography set-up parameters on the capability of a temporal convolutional neural network to characterize defects in a carbon fiber-reinforced polymer specimen. Moreover, to make the lock-in technique suitable for industrial applications, a comprehensive study of reducing both the experimental test time and the processing time has been carried out. The performance of the CNN has been evaluated as a function of some lock-in test parameters such as the number of acquired frames per cycles and the number of excitation cycles. The obtained results have been critically discussed through qualitative and quantitative analyses.

“Flying spot” laser infrared thermography (FST) is a non destructive testing technique able to detect small defects by scanning surfaces with a laser heat source. Defects, such as cracks on metallic parts, are revealed by the disturbance of heat propagation measured by an infrared camera. The association of this examination technique with inspection in the visible spectrum, giving access to surface textures and geometries difficult to observe in the IR spectrum, can increase both robustness and performance of the defect detection. However in a deep learning approach, the acquisition of large amounts of visible-IR pairs can be difficult and time-consuming. The present work proposes to explore visible-FST image pairs generation in the context of surface crack detection for metallic materials, using state-of-the-art deep generative models such as Stable Diffusion. Both accuracy of the generated samples and benefits for multi-spectral deep neural models training will be studied.

The identification and categorization of subsurface damages in thermal images of concrete structures remain an ongoing challenge that demands expert knowledge. Consequently, creating a substantial number of annotated samples for training deep neural networks poses a significant issue. Artificial intelligence (AI) models particularly encounter the problem of false positives arising from thermal patterns on concrete surfaces that do not correspond to subsurface damages. Such false detections would be easily identifiable in visible images, underscoring the advantage of possessing additional information about the sample surface through visible imaging. In light of these challenges, this study proposes an approach that employs a few-shot learning method known as the Siamese Neural Network (SNN), to frame the problem of subsurface delamination detection in concrete structures as a multi-modal similarity region comparison problem. The proposed procedure is evaluated using a dataset comprising 500 registered pairs of infrared and visible images captured in various infrastructure scenarios. Our findings indicate that leveraging prior knowledge regarding the similarity between visible and thermal data can significantly reduce the rate of false positive detection by AI models in thermal images.

Thermal imaging is used to detect moisture inside surfaces such as walls or floors by showing the temperature difference between the moisture and the structure. Surface moisture detection can be critical in quality assurance, healthcare, construction and agriculture. This paper aims to extend the usage of thermal imaging and computer vision to detect the coverage of moisture on the surface using computer vision rather than relying on an end user. This process relies on the thermal properties of the liquid that is sprayed on a surface, which would have a distinct temperature difference compared to the surface it is on. The methodology proposed in this paper is to utilize an infrared thermal image camera to analyze the surface. Then, using computer vision, the output is processed to detect the areas of the largest temperature gradients while filtering the noise. This ensures only areas with a large enough gradient are highlighted, capturing the sprayed surface. These areas are converted to a percentage of the captured area and displayed to the user. Preliminary findings from the experiments show that the system is able to detect liquids that have a temperature difference of at least 5 deg C (9 deg F). As this method only relies on thermal imaging, it is a non-destructive and non-invasive test, where the user does not need to interact with the surface or the liquid directly. The information provided by the technology can contribute to fault detection and quality control when it comes to spray coverage.

Imaging and other nondestructive evaluation techniques are commonly used for material characterization and defect recognition in safety critical aerospace applications, with data fusion providing the framework for uncertainty quantification in these contexts. Most commonly, forward physics-based modeling predicts the response conditioned on material properties and defect assumptions, and probabilistic methods are used to infer the hidden state of the subject of the inspection from a combination of prior information, likelihoods, and inspection data. In this paper Bayesian methods are used to estimate bond thickness in lap joints comprised of aluminum adherends using a combination of infrared thermography and ultrasound. The concept of the conflation of probability distributions is applied to combine the posterior distributions derived from thermography and ultrasound and the quality of the fused estimates are compared against the individual estimates against synthetic data that was created to mimic the inspection of a lap joint comprised of aluminum adherends.

The increasing use of unmanned aerial vehicles (drones) and their availability also provide new opportunities for the development of thermal imaging applications. This presentation presents the current applications of some research and development units in Finland and also reflects on the requirements that aerial thermal imaging places on the operation of thermal imagers. In European Union project called S34I, VTT and Smaps Oy together with project consortium partners develop novel in-situ data acquisition methods at both land and shallow waters for mining industry purposes. The goal is to build for example short-wavelength infrared range spectral libraries to be used as ground truth or calibration of Earth observation related methods.

A drone-based inspection system that can fly, hover, and navigate around structures to perform the inspection in an efficient/fast manner can considerably reduce inspection time. Active thermography is a well-known non-destructive testing method for inspection. However, using it on a drone is challenging due to the drone needing to carry an appropriate heat source, batteries or tethering system to power the heat source and to provide adequate flight time. This complicates the inspection process and can restrict the amount of thermal energy that can be applied to the inspected structure. Another challenge with drone-based active infrared thermography (DBAIT) is that, unlike traditional active thermography inspection in which, the source is either stationary or moving in a precisely controlled manner, the drone and the heat source are subjected to undesired dynamic motion. This paper presents the results of experiments performed to compare potential heat sources that can be retrofitted onboard a drone to conduct active thermographic inspection.

A hyperspectral thermal imager developed inhouse is used for segregation of commercial plastics. The imager combines an uncooled microbolometer equipped thermal camera and a low-order scanning Fabry-Pérot interferometer placed in front of its collecting optics. The distribution of transmitted wavelengths is determined by the distance between the two interferometer mirrors. Data cubes are constructed by capturing images at different mirror separations and the recorded interferograms are subject to subsequent analysis. Twelve types of plastics have been heated to 60°C and imaged before being identified based on a nine-component principal component model and k-nearest neighbors.

Singapore produced more than 982,000 tonnes of plastic waste in 2021. Plastic waste is among the top 4 generated wastes in volume. Astonishingly, plastic waste has one of the lowest recycling rates of just 6% compared to the other 3 highly generated wastes (99% for ferrous metal, 39% for paper, 99% for construction waste). Critically, the lack of effective plastic waste sorting technologies is one key factor that inhibits recycling rate.

Existing plastic sorting relies on manual checking of the printed RIC on plastic wastes. As the printed RIC codes could be small in size, printed at different locations on the plastic objects, and potentially contaminated with dirt, mud, filth, etc., manual plastic sorting is slow, labour intensive, error-prone and poses health risks to facility workers. Overall, existing plastic sorting is ineffective and is a critical barrier in plastic recycling.

In this presentation, we report our work on the development of novel AI/ML-assisted multispectral and hyperspectral imaging technologies and integrate that into a robotic platform for automatic plastic waste sorting and recycling. The outcome is a noticeable increase in plastic waste recycling rate.

We present measurements of controlled Li-ion battery explosions using high-speed infrared thermography to elucidate the effects of this phenomenon. In one study, commercial Li-ion batteries were perturbed by slow heating and by rapid puncture at various states of charge (SOC). The results indicate that the SOC has a significant impact on the magnitude of the battery explosion, regardless of the way the battery is perturbed. Another study tested varied heating rates in thermal abuse tests and showed that faster heating also leads to more violent thermal runaway. Within those measurements, the plumes emanating from the safety vents on the batteries were clearly observed. This work focuses on the propagation of the explosion immediately after the battery detonation event and the implications of the results for designing safer, more reliable Li-ion battery systems.

Inductive thermography is a non-destructive inspection technique. The sample is heated with a short heating pulse and an IR camera records the surface temperature, which is then evaluated to a phase image by Fourier transform. The technique can be well applied for detecting cracks in metals. Additionally, it has also the advantage of providing information about the depth of the crack. Larger contrast is an indication of deeper cracks, while small contrasts refer to shallow cracks. Therefore, the phase contrast can be used to make an estimation of the considered crack. In order to investigate these capabilities, short cracks (length =0.3-3mm) were created in Inconel 718 welded samples by a Varestraint test machine. The samples were then inspected with inductive thermography, computer tomography (CT) and by fluorescent penetrant test (FPT). The crack lengths obtained by all the three methods are compared. The dependency of the phase contrast on the crack depth and length is then analyzed in comparison to the CT results. Finally, additional finite element simulations were carried out and compared to the experimental results.

Due to beneficial mechanical properties, cast manganese (Mn13) steel is used for premium grade railway turnout frogs worldwide. However, its coarse-grain structure makes common non-destructive testing (NDT) methods for defect detection used in this industry very difficult to apply. Inductive thermography is a NDT method well suited for this problem. Scanning inductive thermography is used to localise surface defects on the running surfaces of turnout frogs. Once localised, we propose additional static measurements to characterise the detected surface defects with respect to crack length, depth and penetration angle. Simulations with ANSYS Multiphysics are conducted to study the influence of different crack geometries as well as the influence of different excitation parameters. Validation measurements on samples with defined crack geometries are conducted. The results of both, simulation and measurements on samples, are used to characterize surface defects on actual manganese turnout frogs.

This work presents a preliminary investigation into the quality assessment of Probeless Friction Stir Spot Welding (PFSSW) joints in dissimilar aluminum alloys using laser thermography. This innovative process based on friction stir spot welding, demonstred good capabilities for aluminum joints. To evaluate the quality of these innovative joints in a nondestructive way a long-pulsed laser thermography preliminary procedure has been proposed. The study compares thermographic assessments with conventional testing on two specimens: deemed acceptable and not. Results indicate that preliminary procedure based on laser thermography can effectively assess weld quality, estimating the size of welded zones with an error margin of 3% relative to ultrasonic test UT.

The present work constitutes a preliminary step toward the ultimate goal of developing a thermographic procedure for the non-destructive evaluation of the mechanical strength of these joints using laser thermography, with a simplification of the equipment.

Ceramic thermal barrier coatings (TBCs) play a crucial role in protecting components exposed to combustion gases or high temperature. However, TBCs can degrade over time due to thermal/mechanical fatigue and chemical processes, highlighting the importance of monitoring sintering kinetics. Changes in thermal diffusivity are indicative of sintering progression, necessitating precise and efficient evaluation methods. To this aim, pulsed and lock-in thermography are established techniques, but face limitations such as SNR constraints and the need for multiple tests at single frequencies. To address these challenges, a novel approach called Multi-frequency (MF) lock-in thermography is here introduced. While further research is required for achieving results comparable to state-of-the-art techniques, this study show promises for MF thermography in characterizing TBC properties.

Diffusion reversal techniques are integrated into enhanced truncation-correlation photothermal coherence tomography (eTC-PCT) to restore blurred infrared thermophotonic images to their original, undistorted optical resolution. Diffusion reversal imaging shows promise as a general non-invasive biomedical imaging and non-destructive testing tool, with potentially wide range of applications from biothermophotonic imaging of disease and subcutaneous tissues to manufactured product quality control inspection, including green stages of automotive component production. This represents a significant advancement in this field, as it successfully overcomes the traditional limitations posed by lateral and axial heat diffusion.

Fused filament fabrication (FFF) is the most widely used additive manufacturing (AM) technique to produce fibre-reinforced polymer matrix composites, due to their low wastage, geometric flexibility and ease of use. However, composites manufactured in this way are highly susceptible to defects such as high void content and poor bond quality at the fibre and matrix interfaces. In the present work, a combination method of Infrared Thermography, Acoustic Emission and micro-computerised tomography was developed for the monitoring of the FFF AM process. Both pure plastic and fibre-reinforced composites were manufactured, and the detection and development of defects created during the printing process were monitored. This combination of techniques allows for detection of defects such as porosity, voids and poor fibre-matrix bonding during printing and the verification of their presence after the printing without the need for destructive testing.

The use of composite materials in aircraft manufacturing is increasing, driven by the need for reduced weight and improved fuel efficiency. This trend extends to military aircraft, as defence manufacturers and suppliers seek to minimise operating costs. However, these composites are susceptible to various defects, which require the use of advanced Non-Destructive Testing (NDT) techniques for their detection and evaluation. Phased Array Ultrasonic Testing (PAUT) is such an NDT technique that provides precise defect characterisation, including delaminations, cracks, voids, and porosity. Active Thermography (AT) is an alternative NDT technique that is emerging as a non-contact and full-field technique for identifying such defects. PAUT and AT are complementary techniques, and a synergistic approach that combines their capabilities promises enhanced aircraft safety and composite structure reliability. This paper presents research findings using PAUT equipment and active infrared thermography for identifying and evaluating defects in composites. The presented combined approach is expected to improve aircraft safety and composite structure reliability in aerospace applications.

Certification of additive manufactured metal parts requires nondestructive evaluation (NDE) to ensure build quality. NDE can be performed during the build process or post build. For large parts with complex geometries, post build NDE can be challenging. In-situ NDE potentially provides a way to perform the inspection layer by layer. This work explores the use of a high speed near infrared (NIR) camera that is focused in-line with a laser to obtain high spatial and temporal resolution thermal imagery of the melt pool and associated cooling areas. The thermal data is obtained during a laser melting process using a Ti-6Al-4V plate and of particular interest is the detection of keyhole porosity. Keyhole porosity can result from non-optimal build conditions, such as excessive laser power at a given laser scanning speed, that creates an entrapped bubble. The NIR measured melt pool and cooling areas are processed to detect keyhole porosity. The results are compared to X-ray computed tomography (CT) for validation. Keyhole pores buried deep were not detectable with this technique, however, some larger subsurface elongated pores and some open surface pores did show some promise for detectability.

The latest version of ShipIR/NTCS (v4.3) includes a more generalized 4-point (quad) element type that not only reduces the total number of surface elements in the ShipIR model, but also delivers a higher-quality coarse wall boundary mesh on which to construct the coupled CFD wall boundary and volume meshes for use in CFD analysis. The objective of the current paper is to explore the impact of these improvements on the coupled ShipIR / ANSYS Fluent CFD model solutions previously discussed for both naval ships (Vaitekunas et al, 2011) and aircraft (Vaitekunas 2022). In the case of the naval ship, methods and inputs used to characterize the exhaust gas plume trajectory and associated risk of plume impingement on specific areas of the superstructure are described and applied to a coupled ShipIR / ANSYS Fluent CFD model of the CFAV Quest. These techniques are used during the detailed design phase of a new warship to help further reduce the risk of combat system equipment failure and/or elevated thermal IR signatures associated with exhaust gas impingement heating.

Head checks and transversal cracks are very common defects on rails due to rolling contact fatigue (RFC) phenomena. Despite not reaching the nominal mechanical strength threshold during operation, these cracks tend to propagate and expand, potentially culminating in abrupt component failure during service. This underscores the crucial role of assessing the remaining service life of these components. Therefore, the development of non-destructive testing methodologies is mandatory. Nowadays, railway rails are usually inspected through ultrasound, visual inspection, and eddy current testing, with some limits due to the contact with the component and the number and type of detectable defects. In the last few years induction and laser thermography have been proposed as innovative non-destructive techniques for rail inspection, with interesting results, considering low speeds and laboratory setups.
In this work, we show some results from inspecting rail pieces after induction and line laser thermographic tests, considering the inspected component moving at speeds up to about 12 km/h.
With an image reconstruction algorithm and some filter operations unique images of rail piece have been obtained, showing the information related to surface defects and their dimensions and position.
Additionally, different coil geometries have been developed and proposed to optimize induction inspection.
To align with on-site applications, the experimental tests have been performed considering a setup configuration with mirrors. Following an investigation into the effects of test parameters on the detectability of the relevant cracks, such as scan speed, frame rate and type of excitation, a preliminary procedure for detecting real cracks is introduced, tailored for on-site applications, and validated through experimental results.

This work deals with the characterization of biochar deposited on a thin metallic sheet, used to enhance the evaporation in heat exchangers. The effectiveness of such deposit is done by means of the heated thin foil thermographic technique. The thermal diffusivity of the foil is the main objective of this study. In the proposed method, a pulse of 2 ms duration is produced by a laser, and periodically projected on the surface of an opaque sample of which the thermal parameters have been determined by classical measurements. The spatial distribution of the laser light pattern is random, after passing through a mask like a QR code. Several masks with different spatial features and distribution were prepared by sputtering thin layer (100 nm) of gold on a piece of glass covered by a pattern. Using the masks, samples were photothermally excited by impulsive laser light patterns. The resulting dynamic temperature field evolution at the sample surface was observed by a fast IR camera in the LW, and the thermal diffusion process was recorded by a sequence of IR images. In this contribution, a theoretical model is described and utilized to analyze the spatiotemporal dependence of the temperature field.

Since its introduction in the 1980s, the field of Flash Thermography for Nondestructive Testing (FTNDT) has benefited from enormous advances in the underlying IR camera and computer technologies that enable it, while the flash excitation component has remained essentially unchanged. A typical FTNDT setup comprises a pair of helical or U-shaped xenon flashlamps designed for use in commercial photography, each powered by a bank of capacitors under computer control. A single flashlamp-power supply unit is often described as 4-6 kJ of energy with duration 2-3 msec, the Full Width Half Maximum (FWHM) duration of the flash. However, saturation of the IR camera detector may persist considerably longer due to the extended tail of the flash pulse. In high thermal diffusivity materials, the presence of saturation may mask features of interest, and limit access to early onset signals normally used for depth measurement of subsurface features. Saturation may be mitigated to some degree, but not entirely, by reducing flash energy (power supply voltage). However, more effective elimination of saturation is accomplished using a dedicated hardware device to truncate the duration of the flash pulse. In this paper, we compare the effect of varying flash energy by adjustment of power supply voltage and flash duration on detection of near surface features in an aluminum plate.

We apply lock-in infrared thermography with homogeneous illumination to characterize semi-infinite delaminations (infinite in one direction but with finite length in the perpendicular direction), i.e., we determine the delamination depth, length and thickness. We present two independent calculations of the surface temperature amplitude and phase: a semi-analytical solution by applying the cosine Fourier transform and the quadrupoles method, and a numerical calculation using finite element modelling. The sensitivity analysis indicates that both amplitude and phase data are required to obtain the three parameters (depth, length and width of the delamination) simultaneously. By fitting experimental data taken on AISI-304 stainless steel samples containing calibrated delaminations of different lengths and widths and located at different depths, we prove that it is possible to characterize semi-infinite delaminations with lock-in IRT.

Aircraft wheels and brakes are safety critical systems which play a critical role in supporting aircraft ground operations, specifically during the aircraft deceleration phase. The energy absorbed by the wheels and brakes has a significant impact on the nitrogen pressure within the tyre, which can potentially impact the tyre performance. This could lead to: unscheduled maintenance on wheels and brakes, and/or fuel inefficiency, and/or increase in overall maintenance costs, and/or accidents during ground operation. The existing single aisle commercial aircraft have insufficient instrumentation on wheels and brakes and therefore provide limited technical data to support the understanding of wheel and brake temperatures and their interdependencies with the external environment. In this paper an innovative approach has been proposed, and demonstrated, to capture the temperature signatures of various critical locations on the aircraft wheels and brakes based on the use of an infrared thermographic camera. To support the overall implementation of this research study, the Cranfield University National Flying Laboratory Centre (NFLC) Saab 340B (Registration number G-NFLB) aircraft has been investigated. The wheel and brake temperature signatures have been acquired corresponding to two different flight profiles. The acquired results suggest that the employed thermographic camera can consistently capture temperature trends at all target locations across both flight profiles. Furthermore, trends detected at each location provide engineering insight into the cooling pattern corresponding to each location, including the influence of the external environment. The results therefore pave the scientific foundation to further develop the engineering understanding of the wheel and brake temperature data set that can be utilised in supporting the implementation of a condition monitoring solution for accurate prediction of tyre pressure.

Thermal inspections of a structure typically utilize a flash or quartz lamp heat source located on the same side of an infrared camera. The heat source provides light energy for heating while the infrared camera measures the surface transient temperature response. The inspection can be difficult for low emissivity surfaces for several reasons. First, the high intensity light can reflect off the surface and cause “burn-in” to the camera’s detector. The “burn-in” can take time for the sensors to recover and potentially damage the detector. Secondly, the heat source after pulsing has a transient cool down component. The cool down component can be reflected and therefore superimposed over the structure’s thermal response, which can cause an error (false defect indications) in the inspection. Lastly, the heat source is spectrally broad and therefore while heating, infrared components of the heat source can produce non-uniformity in the measured temperature field. Typically for the inspection of low emissivity surfaces, paint or other emissivity enhancing coatings are applied before inspection. In this paper, a pulsed light emitting diodes (PLED) heat source is used. The PLED heat source is spectrally narrow, contained within the visible band, and therefore not detectable by the infrared camera. The PLED heat source is configured to reduce any transient cool down components that could produce false defect indications. The PLED thermal inspections are compared to flash thermography inspections on unpainted aluminum samples with simulated corrosion and an additively manufactured Ti-6AL-4V metal specimens.

This study reports the results of active thermography non-destructive inspection of an ancient artwork. The artwork was decorated with metal ornaments, realized by hands from specialized craftsmen. However, it was seriously damaged by a catastrophic event based on repeated vibrations. The condition of the ancient artwork was evaluated from thermal data recorded during and after a long pulse heating. In this study, the evaluations were performed using not only the raw thermograms but also the results of three post-processing algorithms, namely: pulsed phase thermography (PPT), thermographic signal reconstruction (TSR), and principal component thermography (PCT). Comparison of the results based on signal-to-noise ratio (SNR) calculations showed that TSR was the most effective technique for detecting defects, and PCT was also effective for the detection of crack-shaped defects. In contrast, PPT seems not suitable for inspection of objects with a complex shape.

Fused filament fabrication (FFF) is the most widely used additive manufacturing (AM) technique to produce fibre-reinforced polymer matrix composites, due to their low wastage, geometric flexibility and ease of use. Composite materials generally have superior properties such as being stiffer and more robust than conventional materials at a reduced weight leading to their application in a wide variety of sectors (aerospace, automotive etc). However, composites manufactured in this way are highly susceptible to defects such as high void content and poor bond quality at the fibre and matrix interfaces. These defects stop fibre-reinforced composite materials manufactured this way meeting industry standards and being used for structural applications. In the present work, a combination methodology of acoustic emission (AE) alongside tensile testing has been developed to investigate the structural integrity and mechanical performance of AM fibre-reinforced composites. Pure polymer samples and short carbon fibre reinforced composites were manufactured, and their mechanical properties were observed.

The accurate segmentation and detection of defects in infrared and visible images are critical for non-destructive testing applications, however those steps are often excluded by limited annotated training data. This paper presents an innovative approach for the segmentation and detection tasks into a unified framework. The proposed method introduces and tests a novel framework tailored to the domain of infrared and visible imaging. This framework eliminates the need for annotated defect data during training, enabling models to adapt to real-world scenarios where annotations are scarce. To enhance the accuracy of segmentation and detection, it employs super-pixel segmentation, following by texture analysis.