TY - JOUR
AU - Averty,, Rodolphe
AB - Abstract In this study, the thermal monitoring of a bridge deck is carried out over several days thanks to an adapted infrared measurement system. This system does not just operate a single uncooled infrared camera but also other sensors (i.e., a weather station and a global positioning system (GPS). The detection of the inner structure of the deck is achieved by pulse phase thermography and principal component thermography approaches. A first characterization of the inner structure of the deck is proposed thanks to an original thermal modelling approach. The results obtained are discussed and analysed. infrared thermography, wireless measurement system, thermal monitoring, thermal modelling, image processing, in situ measurements 1. Introduction One of the objectives of the ISTIMES (Integrated System for Transport Infrastructures surveillance and Monitoring by Electromagnetic Sensing) project1 1 www.istimes.eu. was to evaluate the potentialities offered by the integration of different electromagnetic techniques able to perform non-invasive diagnostics for the surveillance and monitoring of transport infrastructures. Among the electromagnetic (EM) methods investigated, infrared thermography using an uncooled infrared camera is a promising technique due to its dissemination potential (according to its low cost on the market). Furthermore, active infrared thermography for the detection of defects has been used for many years now for the non-destructive control of materials such as metals or composites, as described in the literature (Maldague 2001). Its application to civil engineering materials like cement concrete, a slightly porous and almost homogeneous material, has been shown too (Maierhofer et al2006). But, to our knowledge, active infrared thermography based on natural solar thermal excitation has not been investigated for transport infrastructure monitoring. Although, some research works are available in the building domain, the majority of the work carried out in the natural environment deals with security applications and vision. In such a context, infrared thermography, when it is used in a quantitative mode (not in laboratory conditions) and not in a qualitative mode (vision applied to survey), needs to process thermal radiative corrections on the raw data acquired in real time, to take into account the influences of the natural environment's evolution with time. However the camera sensor has to be smart enough to apply calibration laws and radiometric corrections in real time in a varying atmosphere. So our applied research work addresses the study and the implementation of an adapted infrared measurement system for the thermal monitoring of transport infrastructures. In particular, the measurement system operates at the same time a low-cost infrared camera available on the market, coupled with other sensors to feed simplified radiative models running, in real time, on the GPU available on a small PC. Experiments are carried out on a real transport infrastructure open to traffic, the ‘Musmeci’ bridge. The detection of the inner structure of the deck is achieved by two image processing techniques (pulse phase thermography (PPT) and principal component thermography (PCT)) initially used in the domain of non-destructive testing (NDT) by active infrared thermography. A first characterization of the inner structure of the deck is proposed thanks to an original thermal modelling approach. The promising results obtained are shown and discussed. Finally, a set of perspectives is proposed. 2. The infrared measurement system implemented The measurement system implemented on the real site is a multisensor one. It uses a fast ethernet camera FLIR A320 (Maldague 2001) coupled with a VAISALA WXT520 (VAISALA 2010) weather station, and a light GPS unit (UBLOX 2009) for positioning and timing. It can be used with other ethernet infrared cameras (i.e., visible ones) but needs to be able to access measured data at the raw level. In the present study, this has been made possible thanks to a specific agreement signed with the FLIR company. Figure 1 (left) shows a schematic representation of the interconnection between all of the measurement systems. Figure 1. Open in new tabDownload slide Schematic representation of the measurement system developed (left) and the principle of interoperability (right). Figure 1. Open in new tabDownload slide Schematic representation of the measurement system developed (left) and the principle of interoperability (right). The measurement system can be remotely control via a WiFi connection. For field application an outdoor WiFi access point was used to improve communication performance. Though the run-time measurement platform is running under Linux, the remote control PC used can be run on a different operating system (from PC or Mac OS up to Android and IOS). In the present study, a VNC (virtual network computing) server was mounted on the measurement platform and a VNC Client was used under a remote PC running an MS Windows operating system. To facilitate interoperability at ground level in the field, a tactile tablet PC was used for control operations and setting the system parameters at the beginning of the experiments. The right-hand side of figure 1 shows a scheme of the wireless remote connection principle with VNC that uses a RFB (remote frame buffer) protocol. The infrared measurement prototype system is implemented on a low-cost small computer that integrates a GPU card to allow real-time parallel computing (NVIDIA 2010). Measurement corrections are carried out inline using the weather station measurements. It takes advantage of the IFSTTAR models compiler (Dumoulin and Averty 2012) that allows one to use its own calibration and measurement correction laws or the ones proposed by the infrared camera provider. The compiled models are compliant for GPU computing. The apparent temperature data conversion from the raw data matrix and measured environmental conditions is operated for each time step at the pixel level using parallel computing. Figure 2 shows a schematic representation of the infrared raw data matrix conversion into an infrared image of the apparent surface temperature using models running on the GPU. Figure 2. Open in new tabDownload slide Schematic representation of the temperature conversion process. Figure 2. Open in new tabDownload slide Schematic representation of the temperature conversion process. Tests made with a mini-PC using a CUDA GPU show that we can gain a performance ratio of 40 times or more versus a standard CPU (computing and displaying thermal images at the same time). In the present study, a classical radiative heat balance (Gaussorgues 1989), as shown in figure 3, was used. Figure 3. Open in new tabDownload slide Schematic representation of the hardware. Figure 3. Open in new tabDownload slide Schematic representation of the hardware. The simplified radiometric (Gaussorgues 1989) heat balance given by equation (1) is calculated at each time step of the measurement using information measured by the weather station coupled with an atmospheric transmission model (for instance, see Shettle and Fenn (1979)): DL 0′=τ atm ɛ0 DL 0+τ atm (1-ɛ0) DL e+(1-τ atm ) DL atm ,1 where τatm is the atmospheric transmission computed, ε0 is the measured object emissivity and DL is the digital level at the detector outlet level. Finally, the whole infrared measurement system was operated through a dedicated software called IrLaW (Infrared thermography through Lan and WiFi), developed at IFSTTAR (Dumoulin and Averty 2012). This interface is illustrated in figure 4. It has the standard functionalities that allow image averaging, fixing the number of frames by the thermal image sequence saved, changing the sampling frequency during an acquisition with no interruption. For field application, scenarios were implemented to automatically generate an average image at a fixed frequency (while the camera was running at a higher frequency) and the thermal image sequences were saved at a fixed time period to avoid a loss of data in case of malfunction. Figure 4. Open in new tabDownload slide Screen copies of IrLaw control windows: main window (left); trig, GPS and weather measurement activation (right). Figure 4. Open in new tabDownload slide Screen copies of IrLaw control windows: main window (left); trig, GPS and weather measurement activation (right). The whole system was tested in laboratory conditions and then deployed at the field level in continuous measurement operating conditions over several days. The next paragraph presents the experiment carried out on the ‘Musmeci’ bridge in the framework of the European project (Shettle and Fenn 1979, Proto et al2010) ISTIMES. 3. System implementation on the ‘Musmeci’ bridge and measurement realized The ‘Musmeci’ bridge, shown in figure 5, is located in Potenza (Italy). It is made up of a reinforced concrete box plank, held each 17.30 m by a reinforced concrete continuous vault equivalent to four arches with wheelbases of 17.30 m × 4 = 69.20 m and a free span amongst the supports of 58.80 m. The entire bridge consists of four spans of 69.20 m (figure 5) with an overall length of 277 m. Figure 5. Open in new tabDownload slide View of the rear face and pillars (left), and aerial picture and scheme of the ‘Musmeci’ bridge (right). Figure 5. Open in new tabDownload slide View of the rear face and pillars (left), and aerial picture and scheme of the ‘Musmeci’ bridge (right). The whole measurement system was implemented on the ‘Musmeci’ bridge in July 2011. No traffic interruption was required during the mounting of our measurement system. 3.1. System implementation on the ‘Musmeci’ bridge Figure 6 shows a schematic representation of the infrared camera implementation on a mast placed on the side of the bridge deck. The camera was mounted on top of the mast at an elevation of 6 m from the surface of the bridge deck. The weather station was mounted on the same mast at 1 m under the camera. A GPS antenna was also fixed at the base of the mast and at a same elevation to the bridge deck surface. Figure 6. Open in new tabDownload slide Schematic representation of the infrared camera on top of ‘Musmeci’ bridge (left), and pictures of the bridge and of the camera mounted on a mast with a weather station (right). Figure 6. Open in new tabDownload slide Schematic representation of the infrared camera on top of ‘Musmeci’ bridge (left), and pictures of the bridge and of the camera mounted on a mast with a weather station (right). Figure 4 (left) presents a picture of the bridge deck with a purple area that matches the field of view of the infrared camera. Figure 4 (right) presents a screen copy of a simulation tools that were used for prototyping the implementation of the infrared camera on the bridge. The characterization of the effective field of view on site, carried out with reflective targets and differential GPS measurements (using an RTK system), has confirmed the predicted field of view. Figure 7. Open in new tabDownload slide Schematic representation of the infrared system field of view evaluated at the bridge deck level (left), and the predicted field of view (right) using IRISSA (InfraRed Image Simulations with Sensor properties Assimilation) software dedicated to a simplified road scene numerical simulation (see Dumoulin et al (2006)). Figure 7. Open in new tabDownload slide Schematic representation of the infrared system field of view evaluated at the bridge deck level (left), and the predicted field of view (right) using IRISSA (InfraRed Image Simulations with Sensor properties Assimilation) software dedicated to a simplified road scene numerical simulation (see Dumoulin et al (2006)). Figure 8 shows a view of the infrared camera with its protective caisson; the computing (steel blue box) and communication (access point for wireless remote control) hardware; and a picture of the bridge deck taken at the infrared camera level with the field of view reported in red. Electrical power was provided at bridge level. Figure 8. Open in new tabDownload slide Detailed views of the prototype system implemented on the bridge under traffic (left and middle); the field of view of the IR camera versus the bridge deck (right). Figure 8. Open in new tabDownload slide Detailed views of the prototype system implemented on the bridge under traffic (left and middle); the field of view of the IR camera versus the bridge deck (right). Deployment of the whole prototype system, including a testing and setting the system before measurements started, took no more than half a working day. 3.2. Measurement realized This trial took place over four days, but our system was left in stand-alone acquisition mode only for three days. This time limitation was due to the time reserved for the assembly and dismantling operations required before and after the experiments in order to leave the bridge as it was before the experiments. Thanks to the infrared measurement system used, thermal images were acquired at a frame rate of 0.1 Hz by averaging 50 thermal images with an initial camera frame rate fixed at 5 Hz. Each hour, a thermal image sequence was stored on the internal hard drive of our system and data were also retrieved, on demand, using a wireless connection with a tablet PC. Figure 9 shows four thermal images of the thermal loading and cooling of the bridge deck under natural conditions monitored at different times. Figure 9. Open in new tabDownload slide Example of thermal images at different times of the day–night cycle (under natural thermal solicitation). Figure 9. Open in new tabDownload slide Example of thermal images at different times of the day–night cycle (under natural thermal solicitation). Figure 10 (left graph) shows the external temperature and the wind speed evolution during experiments. On the third day a non-negligible raising of the wind speed (up to 100 km h–1) can be observed, with a quasi-constant external temperature in parallel. Figure 10. Open in new tabDownload slide The measured atmospheric temperature over around 60 h and the wind speed during the same period (left); and the measured apparent temperature evolution by infrared thermography and relative humidity during the same period (right). Figure 10. Open in new tabDownload slide The measured atmospheric temperature over around 60 h and the wind speed during the same period (left); and the measured apparent temperature evolution by infrared thermography and relative humidity during the same period (right). Figure 10 (right graph) shows an example of results for the apparent temperature evolution measured by infrared thermography during experiments. Thermogram was extracted from the thermal image sequence acquired on site. The measured relative humidity shows a burst of humidity during the third day in particular. It matches the sudden wind event but also random sun shadowing by clouds and the external temperature stabilization. A combination of the convective effect and external temperature reduce the thermal relaxation during night time on the beginning of the third day. Figure 11 reports the evolution with time of a longitudinal profile and of a transversal profile extracted from a thermal-image sequence. Figure 11. Open in new tabDownload slide Longitudinal profile evolution with time (left); and transversal profile evolution with time (right). Figure 11. Open in new tabDownload slide Longitudinal profile evolution with time (left); and transversal profile evolution with time (right). For the longitudinal profile, the slab junction recovered by the pavement wearing course in bitumen concrete could be associated with the different evolution of temperature observed a round line 200 on the figure. For the transversal profile, a non-homogeneous spatial distribution of the temperature is observed, which could be linked to heating by the sun, which rises on one side of the bridge and sets on the other side with a perpendicular illumination at noon. 4. Data processing and thermal modelling Two analysis approaches were studied. The first one uses an image processing approach. The second one uses a thermal modelling approach. 4.1. Image processing approach Two image-processing approaches, used for analysis in the present study, are detailed hereafter. 4.1.1. Pulse phase thermography A frequency analysis approach is a way to reduce the number of thermal images to be analysed in a sequence. The Fourier transform (equation (2)) is applied to the temporal evolution of each pixel of the thermal image: Fn=Δt∑k=0N-1T(kΔt)exp(-j2πnk-j2πnkNN)= Re n+ Im n,2 where n designates the frequency increment (n = 0,1,…, N); Δt is the sampling interval; and Re and Im are the real and the imaginary parts of the Fourier transform. In practice, the fast Fourier transform algorithm (Cooley and Tukey 1965) was used to reduce computing time. Then magnitude (An) and phase (Φn) maps were calculated (equation (3)): An= Re n2+ Im n2 and Φn=tan-1 Im n Re n.3 The magnitude and phase maps were analysed to detect the scheme of the inner structure of the bridge deck. 4.1.2. Principal component thermography Investigation of the inner deck structure was also carried out by using a singular value decomposition (SVD) approach, allowing us to extract the spatial and temporal information from a thermal image sequence in a compact or simplified manner. Instead of relying on a basis function (such as in pulsed phase thermography (PPT), which is based on a Fourier transform relying on sinusoidal basis functions), SVD is an eigenvector-based transform that forms an orthonormal space. The SVD of an M × N matrix A (M > N) can be calculated as follows (Rajic 2002, Marinetti et al2004): A=UΣVT4 where U is an M × N orthogonal matrix, Σ is a diagonal N × N matrix (with the singular values of A in the diagonal) and VT is the transpose of an N × N orthogonal matrix (characteristic time). Hence, to apply the SVD to thermographic data, the 3D thermal images sequence matrix representing time and spatial variations has to be reorganized as a 2D M × N matrix A, as depicted in figure 12. This can be done by rearranging the infrared images for every increment of time as columns in A, in such a way that time variations will occur column-wise while spatial variations will occur row-wise. Figure 12. Open in new tabDownload slide Schematic representation of the application of the SVD to the infrared image sequence. Figure 12. Open in new tabDownload slide Schematic representation of the application of the SVD to the infrared image sequence. After rearranging the thermal image for every increment of time as columns in A and applying the SVD, the columns of U represent a set of orthogonal statistical modes known as empirical orthogonal functions (EOF) that describes spatial variations in the data. On the other hand, the principal components, which represent time variations, are arranged row-wise in matrix VT. The first EOF will represent the most characteristic variability of the data; the second EOF will contain the second-most important variability, and so on. Usually, original data can be adequately represented with only a few EOFs (Rajic 2002, Marinetti et al2004). Therefore such a processing method was also applied to detect the scheme of the inner structure of the bridge deck. 4.2. Thermal modelling In the following section, the thermal model is described and the estimation procedure used to estimate the apparent thermal conductivity spatial distribution at the bridge deck level, using measured data, is described. 4.2.1. Thermal modelling of the bridge deck If we focus the analysis on the first two days, the measured data have a periodic behaviour with a status close to the permanent regime. So the complex amplitude method can be used to solve the studied system (Lascoup et al2013, Autrique et al2009). The system to solve is expressed in equation (5). Moreover, considering that the deck has a large thickness, but also considering its nature and the fact that we only look at a few daily periodic thermal perturbations at the deck surface, an adiabatic boundary condition was applied on the rear face of the deck: jωαT˜(ωn)=∂2T˜(ωn)∂xa2ka∂T˜(ωn)∂xax=0=ϕ˜(ωn)ka∂T˜(ωn)∂xax=ea=0.5 Finally, the bridge deck model is expressed using the apparent thermal properties by considering that it is made of a unique layer. In a first approach, this simplification allows us to become independent of the 3D geometry of the deck. Time-dependant variables are defined in the frequency domain thanks to the Fourier integral transform, given in equation (6): T˜(ωn)=12π∫0tpT(t)exp(-jωnt)dt.6 The initial thermal gradient of the bridge deck structure is unknown. To erase its influence on the calculus, ten periods made of a repetition of the initial day–night period considered and were used for computing. The thermal properties estimation procedure is carried out during the seventh period. The model is solved thanks to the thermal quadrupoles method (Maillet et al2000), expressed here for a single-layer material: T˜(ωn)x=0ϕ˜(ωn)x=0=coshjωnαaea1kajωnαasinhjωnαaeakajωnαasinhjωnαieicoshjωnαaea×T˜(ωn)x=ea0=AωnBωnCωnDωnT˜(ωn)x=ea0.7 The return in the time domain is carried out with the Fourier inverse transform computed on the model solution to express the time-dependant temperature: T(x=0,t)-Tm=12π∫0ωfinA(ωn)C(ωn)ϕ˜(ωn)expjωndωn.8 4.2.2. Data resizing: solar and convective heat transfer estimation In order to lower the amount of data and optimize the computation time, the most regular daily period has been chosen and sub-sampled to 0.0083 Hz. Sub-sampled data are presented in figure 14 (left). The global heat flux computation is based on the identification of the convective flux and the solar heat flux at the bridge deck surface. The convective heat transfer due to the wind effect on the bridge surface has to be taken into account. A convective heat transfer coefficient is computed for each wind speed thanks to the McAdams correlation (McAdams 1954). Figure 13 (left) shows its evolution during measurements. Figure 13. Open in new tabDownload slide Sub-sampled measurements at 0.0083 Hz (left); and solar heat flux components (right). Figure 13. Open in new tabDownload slide Sub-sampled measurements at 0.0083 Hz (left); and solar heat flux components (right). The chosen temporal sub-sampling applied to the data preserve, at a convective heat transfer coefficient level, the global evolution of the wind during the trials. Furthermore, the bridge deck is subjected to a variable solar heat flux, dependant on the day and the season. Other parameters, like the slope and the solar angle, have to be considered. In this study, the Duffie and Beckman solar heat flux model (Duffie and Beckman 1991) has been used as it was not measured during the trials. This model allows us to compute the theoretical solar heat flux received at a given time in a region of the world. The total solar heat flux is computed by adding its direct, diffuse and reflected components. Here the system studied is not located on side of a mountain and the slope of the bridge is neglected. To be coherent with our data, the solar heat flux has been computed for 24 h, thanks to information obtained by GPS, as well as the latitude. Figure 13 (right) shows the reconstructed solar heat flux, without cloud, for the first 24 h of measurements. 4.2.3. Identification procedure of apparent thermal parameters The studied identification procedure needs filtered input data. The filtering of raw data reduces the effect of high frequency events which disturb the model with respect to the mean signal studied over 24 h. Thus a low-pass Butterworth filter (Butterworth 1930) has been applied to each piece of input data. The cut-off frequency and filter order are shown in table 1. Table 1. Butterworth filter parameters. . Order . Cut-off frequency in Hz . IR measured temperature 4 0.165 Air temperature 4 0.0165 Wind speed 4 0.0917 . Order . Cut-off frequency in Hz . IR measured temperature 4 0.165 Air temperature 4 0.0165 Wind speed 4 0.0917 Open in new tab Table 1. Butterworth filter parameters. . Order . Cut-off frequency in Hz . IR measured temperature 4 0.165 Air temperature 4 0.0165 Wind speed 4 0.0917 . Order . Cut-off frequency in Hz . IR measured temperature 4 0.165 Air temperature 4 0.0165 Wind speed 4 0.0917 Open in new tab Filtered data (figure 14 (left)) are used to compute the convective heat flux thanks to the determination of the convective heat transfer coefficient h(t). Added to the solar heat flux ϕ0(t), this allows the computation of the global heat flux ϕ(t) on the bridge deck surface: ϕ(t)=h filtered (t)T external _ filtered -T measure _ filtered +ϕ0(t).9 Figure 14 (right), shows the temporal evolution of the global, solar and convective heat flux computed. Figure 14. Open in new tabDownload slide Filtered data (left); and computed heat flux (right). Figure 14. Open in new tabDownload slide Filtered data (left); and computed heat flux (right). As the studied period is repeated, the convective heat transfer coefficient also became periodic. The aim of the implemented estimation procedure is to determine the apparent thermal parameters of the bridge deck. Figure 15 shows a schematic view of the whole procedure studied and an illustration of result obtained. Figure 15. Open in new tabDownload slide Scheme of the implemented estimation procedure. Figure 15. Open in new tabDownload slide Scheme of the implemented estimation procedure. The Levenberg–Marquardt algorithm (Marquardt 1963) was used for the minimization procedure. 5. Measurements data analysis 5.1. PPT analysis A frequency analysis approach (PPT) was applied to the thermal image sequence. The Fourier transform (equation (2)) was applied to the temporal evolution of each pixel of the thermal image. Then magnitude (An) and phase (Φn) maps were calculated (equation (3)) and were analysed. The left-hand image in figure 16 shows an example of a magnitude map computed by FFT calculus where meshing is observed. Figure 16. Open in new tabDownload slide Example of computed magnitude maps (left); the bridge deck cross section (middle); and a 3D image of the bridge deck structure and pillars (right). Figure 16. Open in new tabDownload slide Example of computed magnitude maps (left); the bridge deck cross section (middle); and a 3D image of the bridge deck structure and pillars (right). This grid matches the inner deck structure as presented in the middle and right-hand drawings in figure 16. 5.2. PCT analysis Investigation of the inner deck structure was also done by using SVD. Principal component thermography (PCT) analysis was performed on the complete thermal-image sequence but also after the extraction of a night period in the full sequence (close to linear thermal relaxation). Figure 17 shows the EOF map obtained. Figure 17. Open in new tabDownload slide Full sequence EOF map (left); reduced sequence second EOF map (middle); and third EOF map (right). Figure 17. Open in new tabDownload slide Full sequence EOF map (left); reduced sequence second EOF map (middle); and third EOF map (right). For the full sequence the weight of information concerning the inner structure requires us to look further at the EOF maps. For the reduced sequence, the EOF 2 map gives a refined sketch of information obtained by the full sequence analysis. Finally, the EOF map, although it is affected by the traffic signature, delivers accurate information on the inner structure of the bridge deck (caisson mapping). 5.3. Thermal analysis Thermal analysis was first applied to two thermograms and then extended to the whole scene. 5.3.1. Model applied at two points of the bridge deck Before applying the model to the whole scene, two independent points were studied: the first one above a concrete caisson and the other one above a sound area. The apparent thermal conductivity estimated for the first one was about 1.75 W m-1 °C and for the second one was 1.8 W m-1 °C. So the apparent conductivity estimated above the concrete caisson has the same magnitude as that for an equivalent conductivity computed with known parameters (NVIDIA 2010). The second apparent conductivity is far from this value. This can be explained by the fact that a wood frame could have been used for building the bridge. Figure 18 (left) shows the comparison between the measured temperature and the estimated one after identification. Figure 18. Open in new tabDownload slide Comparison between measured and estimated temperature (left); and the apparent conductivity map (right). Figure 18. Open in new tabDownload slide Comparison between measured and estimated temperature (left); and the apparent conductivity map (right). 5.3.2. Implementation of the thermal model and comparison with an FFT analysis The infrared camera used had a 320 × 240 px resolution. In the scene, an area of 240 × 200 px was isolated for the analysis and a spatial averaging of 4 pixels was done in order to reduce computation time. Thus the thermal sequence used in computation had a size of 120 × 100 px. Figure 18 (right) show the estimation procedure result applied to the first period of the measurement. The apparent conductivity map shows that the model used is able to detect and characterize any variation in the properties of the deck structure; indeed the first caisson can be seen. However the surface condition of the structure, and the shadow and spatial distortion due to the field of view of the camera, affect the model results. 6. Conclusion and perspectives A dedicated infrared measurement system has been used to follow with time the thermal evolution of the ‘Musmeci’ bridge deck surface. During trials, a wireless communication mode was used and the data were stored on the platform left on site. The experimental data were retrieved on demand with a tablet PC. The experiments did not require the stopping of or reduction of traffic at the bridge level. The analysis of the results obtained shows that the inner structure of the bridge deck was retrieved by FFT analysis, and with more refinement by using SVD analysis. The extraction of the thermal relaxation period was investigated to enhance the inner structure detection. The thermal model was also implemented in the frequency domain to analyse the infrared image sequence coupled with environmental parameter follow-up. Finally a map of apparent thermal conductivity spatial distribution was calculated. This allowed us to retrieve the scheme of the inner structure of the deck with a unique map, though the computing time requested was higher than for the image processing approaches. Nevertheless, the apparent conductivity map is a first step towards the long-term monitoring of a possible state indicator directly linked to the aging of materials subject to climate aggression and traffic. Future works will address the complementary correction of infrared measurement by introducing environmental radiative data in the model issued from additional sensors (pyrgeometer and pyranometer). Furthermore, weather measurements will be analysed inline to allow tagging in the thermal image sequence to introduce adapted processing to the events observed. Finally, experiments will be deployed on real civil engineering elements of transport infrastructure for at least 1 year to feed a database and also evaluate different processing approaches for data reduction and analysis. Acknowledgment The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 225663. References Autrique L , et al. , 2009 On the use of periodic photothermal methods for materials diagnosis , Sensors Actuators B , vol. 135 (pg. 478 - 487 ) 10.1016/j.snb.2008.09.032 Google Scholar Crossref Search ADS WorldCat Crossref Butterworth S . , 1930 On the theory of filter amplifiers , Wirel. Eng. , vol. 7 (pg. 536 - 541 ) OpenURL Placeholder Text WorldCat Cooley J W , Tukey J W . , 1965 An algorithm for the machine calculation of complex Fourier series , Math. Comput. , vol. 19 (pg. 297 - 301 ) 10.1090/S0025-5718-1965-0178586-1 Google Scholar Crossref Search ADS WorldCat Crossref Duffie J A , Beckman W A . , 1991 , Solar Engineering of Thermal Processes New York Wiley Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Dumoulin J , Averty R . , 2012 Development of an infrared system coupled with a weather station for real time atmospheric corrections using GPU computing: application to bridge monitoring Proc. Quant. Infrared Thermography QIRT, Naples, Italy Dumoulin J , Delannay F , Lecamus G . , 2006 Enhancement of infrared vision for drivers by acting on infrastructure Proc. Transportation Res. Arena TRA, Gothenburg, Sweden FLIR systems . , 2007 Thermovision A320 et Thermovision A320G: , User's Manual www.flir.com Gaussorgues G . , 1989 , La Thermographie Infrarouge: Principes, Technologies, Applications 3rd edn Paris Tec Doc—Lavoisier Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lascoup B , Perez L , Autrique L , Crinière A . , 2013 On the feasibility of defect detection in composite material based on thermal periodic excitation , Composites B , vol. 45 (pg. 1023 - 1030 ) 10.1016/j.compositesb.2012.07.020 Google Scholar Crossref Search ADS WorldCat Crossref Maierhofer C , Arndt R , Röllig M , Rieck C , Walther A , Scheel H , Hillemeier B . , 2006 Application of impulse-thermography for non-destructive assessment of concrete structures , J. Cem. Concr. Compos. , vol. 28 (pg. 393 - 401 ) 10.1016/j.cemconcomp.2006.02.011 Google Scholar Crossref Search ADS WorldCat Crossref Maillet D , et al. , 2000 , Thermal Quadrupoles Solving the Heat Equation Through Integral Transforms New York Wiley Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Maldague X P V . , 2001 , Theory and Practice of Infrared Technology for Non-destructive Testing New York Wiley Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Marinetti S , Grinzato E , Bison P G , Bozzi E , Chimenti M , Pieri G , Salvetti O . , 2004 Statistical analysis of IR thermographic sequences by PCA , Infrared Phys. Technol. , vol. 46 (pg. 85 - 91 ) 10.1016/j.infrared.2004.03.012 Google Scholar Crossref Search ADS WorldCat Crossref Marquardt D . , 1963 An algorithm for least-squares estimation of nonlinear parameters , SIAM J. Appl. Math. , vol. 11 (pg. 431 - 441 ) 10.1137/0111030 Google Scholar Crossref Search ADS WorldCat Crossref McAdams W H . , 1954 , Heat Transmission New York McGraw-Hill Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC NVIDIA . , 2010 CUDA Toolkit Reference Manual www.nvidia.com Proto M , et al. , 2010 Transport infrastructure surveillance and monitoring by electromagnetic sensing: the ISTIMES project , Sensors , vol. 10 (pg. 10620 - 10639 ) 10.3390/s101210620 Google Scholar Crossref Search ADS PubMed WorldCat Crossref Rajic N . , 2002 Principal component thermography for flaw contrast enhancement and flaw depth characterization in composite structures , Compos. Struct. , vol. 58 (pg. 521 - 528 ) 10.1016/S0263-8223(02)00161-7 Google Scholar Crossref Search ADS WorldCat Crossref Shettle P , Fenn R W . , 1979 Models for the aerosols of the lower atmosphere and the effects of humidity variations on their optical properties , Air Force Geophysics Laboratory Report pg. 79-0214 OpenURL Placeholder Text WorldCat UBLOX . , 2009 u-blox 5 evaluation kits EVK-5 Technical Notewww.u-blox.com VAISALA . , 2010 Transmetteur météorologique WXT520 , User’s Guide www.vaisala.com © 2013 Sinopec Geophysical Research Institute
TI - The detection and thermal characterization of the inner structure of the ‘Musmeci’ bridge deck by infrared thermography monitoring
JF - Journal of Geophysics and Engineering
DO - 10.1088/1742-2132/10/6/064003
DA - 2013-12-01
UR - https://www.deepdyve.com/lp/oxford-university-press/the-detection-and-thermal-characterization-of-the-inner-structure-of-5DATl7yyL1
VL - 10
IS - 6
DP - DeepDyve
ER -