Search

This paper presents a deep learning-aided deterioration modeling approach for highway bridge components (i.e., decks, superstructures, and substructures). The method trains a set of deep learning models for the maximum likelihood estimation of parameters in a Markov chain. The likelihood function is based on observed numbers of transitions between different condition states. The deep learning is leveraged for efficient representation of various factors that influence the deterioration process. Aided by deep learning, the proposed method is suitable for deterioration modeling using large-scale and high-dimensional bridge datasets. The study demonstrates an application of this approach using the historical National Bridge Inventory database from 1993 to 2019. The Convolutional Neural Network is adopted as the deep learning model. The proposed method is validated with ten-fold cross-validation that encompasses a nationwide selection of 88,596, 101,414, and 96,244 bridge decks, superstructures, and substructures, respectively. The validation shows this approach achieved a robust and low prediction error with the maximum mean-squared-error near 0.5 in a 26-years-forecast. The proposed method is also compared with the conventional Artificial Neural Network and four selected Markov-chain based deterioration modeling approaches. The study shows this approach can be a promising data-driven tool for deterioration modeling of highway bridge components. [ABSTRACT FROM AUTHOR]

• A data fusion method is presented, which integrates AE waveform, FFT spectrum, and spectral entropy. • A weighted ensemble regression-based CNN was proposed to estimate the coordinates of damage. • Fusion data, CWT, and STFT coefficients were used as input data. • The proposed model estimates the coordinates of damages with high performance. Nuclear power generation is an essential part of the electrical supply in the United States, and it is an effective way to achieve low carbon power generation. Nuclear power generation produces spent radioactive fuel. If improperly disposed of or improperly stored, spent fuel can affect the environment and human health. Currently, in the United States, spent fuel is typically stored in stainless steel canisters. Some stainless steel canisters deployed in coastal areas are subject to stress corrosion cracks. Long-term monitoring and to provide timely maintenance of the storage canisters is necessary to prevent leakage of spent fuel due to damaged canisters. Acoustic emission (AE) is a structural health monitoring (SHM) technique that can be utilized to monitor large-scale metallic structures because it is extremely sensitive to damage initiation and propagation in materials. However, the challenge in using AE is in deploying a minimal number of AE sensors on a canister due to cost and environmental restrictions while still being able to precisely detect and localize damage formation. The innovation of this paper lies in the development of an automated damage localization method to estimate the coordinates of damage by using a single AE sensor. A data fusion approach was designed to integrate the information from waveforms, fast Fourier transform (FFT) spectrums, and spectrum entropy and then convert the AE signals into three types of images along with short-time Fourier transform (STFT) and continuous wavelet transforms (CWT). A weighted ensemble regression-based convolutional neural network was proposed to analyze the images and compute the coordinates of damage. The proposed method was validated on a large-scale steel plate specimen that simulate the canister, and three-fold cross-validation was conducted to ensure the method was effectively evaluated. The results suggest that the proposed method has a high performance rate for locating damage. [ABSTRACT FROM AUTHOR]

This paper explains the use of a convolutional neural network (CNN) to segment solar panels in a satellite image containing solar arrays, and extract associated metadata from the arrays. A novel unsupervised technique is introduced to estimate the azimuth of each individual solar panel from the predicted mask of the convolutional neural network. This pipeline was developed with the aim of extracting necessary metadata for a solar installation, using only a set of latitude–longitude coordinates. Azimuth prediction results for 669 individual solar installations associated with 387 sites located across the United States are provided. A mean average error and median average error of 21.65 degrees and 1.0 degrees were obtained, respectively, when predicting the azimuth of the solar fleet data set, with about 80% of the results within an error of zero degrees of the ground truth azimuth value and about 85% within an error of 25 degrees. The predicted azimuth was then used to estimate the energy conversion of the solar arrays. Results show a 90.9 and 90.6 R-squared value for estimating alternating current (AC) and direct current (DC) energy, respectively, and a mean absolute percentage error (MAPE) of 1.70% in estimating the alternating current (AC) energy using the fully automated algorithm. [Display omitted] • A novel unsupervised technique is used to estimate the azimuth of solar arrays. • Segmentation of solar panels in the satellite image, using transfer learning and a CNN. • The proposed approach is can be used with high resolution satellite or aerial imagery. [ABSTRACT FROM AUTHOR]

Malignant melanoma accounts for about 1–3% of all malignancies in the West, especially in the United States. More than 9000 people die each year. In general, it is difficult to characterize a skin lesion from a photograph. In this paper, we propose a deep learning-based computer-aided diagnostic algorithm for the classification of malignant melanoma and benign skin tumors from RGB channel skin images. The proposed deep learning model constitutes a tumor lesion segmentation model and a classification model of malignant melanoma. First, U-Net was used to classify skin lesions in dermoscopy images. We implement an algorithm to classify malignant melanoma and benign tumors using skin lesion images and expert labeling results from convolutional neural networks. The U-Net model achieved a dice similarity coefficient of 81.1% compared to the expert labeling results. The classification accuracy of malignant melanoma reached 80.06%. As a result, the proposed AI algorithm is expected to be utilized as a computer-aided diagnostic algorithm to help early detection of malignant melanoma. [ABSTRACT FROM AUTHOR]