No abstract

The similarity in the hue of corroded surfaces and coated surfaces, dust, vegetation, etc. leads to visual ambiguity which is challenging to eliminate using existing image classification/segmentation techniques. Furthermore, existing methods lack the ability to identify the source of corrosion, which plays a vital role in framing the corrosion mitigation strategies. The goal of this study to employ hyperspectral imaging (1) to detect corroded surfaces under visually ambiguous scenarios and (2) identify the source of corrosion in such scenarios. To this end, three different corrosive media, namely, (1) 1M hydrochloric acid (HCl), 2) 3.5 wt.% sodium chloride solution (NaCl), and (3) 3 wt.% sodium sulfate solution (Na2SO4), are employed to generate chemically distinctive corroded surfaces. The hyperspectral imaging sensor is employed to obtain the visible and near infrared (VNIR) spectra (397 nm–1004 nm) reflected by the corroded/coated surfaces. The intensity of the reflectance in various spectral bands are considered as the descriptive features in this study, and the training and test datasets were generated consisting of 35,000 and 15,000 data points, respectively. SVM classifier is trained and then its efficacy on the test data is assessed. Furthermore, validation datasets are employed and the generalization ability of the trained SVM classifier is verified. The results from this study revealed that the SVM classifier achieved an overall accuracy of 94% with the misclassifications of 18% and 13% in the case of NaCl and Na2SO4 corrosion, respectively. Reflectance spectra obtained in the VNIR region was found to eliminate the visual ambiguity between the corroded and coated surfaces and, identify the source of corrosion accurately. Further, the range of key wavelengths of the spectra that play an important role in the distinguishability of coating and chemically distinctive corroded surface were identified to be 500–520 nm, 660–680 nm, 760–770 nm, and 830–850 nm.

Application of deep learning (DL) for automatic condition assessment of bridge decks has been on the raise in the last few years. From the published literature, it is evident that lot of research efforts has been done in identifying the surface defects such as cracks, potholes, spalling and so forth using supervised learning methods such as deep learning. However, the health of a reinforced concrete bridge deck is jeopardized substantially due to presence of subsurface defects. Subsurface defects in bridge decks are genearlly detected using non‐destructive evaluation (NDE) methods. Interpertation of NDE data for autonomous deck evaluation requires development of DL models; however, The task of defect detection DL has not received the proper attention for subsurface defect detection in the past. The goal of this paper is to provide a review of existing DL models for analysis of NDE data of bridge decks. The authors reviewed prominent NDE techniques for subsurface defect detection of bridge decks and explored the various DL models proposed to identify these defects. First a brief overview of the working principle of NDE techniques and DL architectures is provided, and then the information about proposed DL models and their efficacy is highlighted. Based on the existing knowledge gaps, various challenges and future prospects associated with application of DL in bridge subsurface inspection are discussed.

Deicing of pavements is essential to ensure safe and timely movement of traffic in geographical locations where snow and ice events are anticipated. State and local municipalities employ brine solution with 23.3 wt% sodium chloride (NaCl) available in the form of rock salt to deice the pavements. Unlike water, the brine solution does not freeze until the temperature falls below −21.0 °C, i.e., the freezing point of water is depressed by −21.0 °C with the addition of 23.3 wt% NaCl. The depressed freezing point of the brine solution plays a key role in deicing pavements. Unfortunately, a further increase in rock salt content does not lower the freezing point of the brine solution. In this study, different combinations of agricultural products such as polyols including sorbitol, maltitol, and mannitol in brine (23.3 wt% of NaCl in water), and NaCl-juice (corn and beet juice) were investigated to achieve freezing point depressions below −21.0 °C for potential deicing applications in extremely cold areas. Different weight fractions of polyols-brine solutions ranging from 7.14% to 27.77% were considered, and corresponding freezing points were determined. While the sorbitol-brine solution exhibited the lowest freezing point of −38.1 °C at a higher concentration, the maltitol-brine solution exhibited a freezing point of −35.6 °C at the same concentration. Based on the °Brix value, beet juice had almost three times more soluble solids and a lower freezing point compared to corn juice. Adding 23.3 wt% of NaCl in 70% corn juice lowered the freezing point up to −23.5 °C.