Vision-based approaches are widely used in steel crack detection. After processing the images taken by the camera, the superficial defects can be detected. Due to the common limitation of the nature of photographic images, internal features of objects cannot be fully discovered. In order to overcome the drawbacks of vision-based methods, this work presents an approach for detecting cracks in infrared thermal imaging steel sheets using Convolutional Neural Networks (CNN). Firstly, a horizontal heat conduction method is researched to thermally excite the surface of the steel sheet, and a rolling electric heating device is developed to be a thermal excitation source. Secondly, we analyze the temperature difference between normal area and different crack depths, and study the influence of temperature change. Thirdly, 3000 infrared thermograms labeled for penetrating cracks, non-penetrating cracks, and surface scratches are fabricated into a databank. Then, the CNN is trained and validated on the data bank. Finally, improving the original Faster Region-based Convolutional Neural Network (Faster R-CNN) network, the feature maps of multiple levels in the feature extraction network are aggregated, and the anchor selection scheme of the Region Proposal Network (RPN) is adjusted from 9 to 25. The robustness of the improved Faster R-CNN is demonstrated by evaluating the detection results on the 125 images outside of the data bank, the accuracy and Mean Average Precision (mAP) are 95.54% and 92.41%, respectively, which outperform the original algorithm by increasing the accuracy of 3.18% and mAP of 1.88%.
Lattice thermal conductivity is important for many applications, but experimental measurements or first principles calculations including three-phonon and four-phonon scattering are expensive or even unaffordable. Machine learning approaches that can achieve similar accuracy have been a long-standing open question. Despite recent progress, machine learning models using structural information as descriptors fall short of experimental or first principles accuracy. This study presents a machine learning approach that predicts phonon scattering rates and thermal conductivity with experimental and first principles accuracy. The success of our approach is enabled by mitigating computational challenges associated with the high skewness of phonon scattering rates and their complex contributions to the total thermal resistance. Transfer learning between different orders of phonon scattering can further improve the model performance. Our surrogates offer up to two orders of magnitude acceleration compared to first principles calculations and would enable large-scale thermal transport informatics.
Evaluating the mechanical response of fiber-reinforced composites can be extremely time-consuming and expensive. Machine learning (ML) techniques offer a means for faster predictions via models trained on existing input–output pairs and have exhibited success in composite research. This paper explores a fully convolutional neural network modified from StressNet, which was originally used for linear elastic materials, and extended here for a non-linear finite element (FE) simulation to predict the stress field in 2D slices of segmented tomography images of a fiber-reinforced polymer specimen. The network was trained and evaluated on data generated from the FE simulations of the exact microstructure. The testing results show that the trained network accurately captures the characteristics of the stress distribution, especially on fibers, solely from the segmented microstructure images. The trained model can make predictions within seconds in a single forward pass on an ordinary laptop, given the input microstructure, compared to 92.5 h to run the full FE simulation on a high-performance computing cluster. These results show promise in using ML techniques to conduct fast structural analysis for fiber-reinforced composites and suggest a corollary that the trained model can be used to identify the location of potential damage sites in fiber-reinforced polymers.
This article develops a deep graph operator network (DeepGraphONet) framework that learns to approximate the dynamics of a complex system (e.g., the power grid or traffic) with an underlying subgraph structure. We build our DeepGraphONet by fusing the ability of graph neural networks to exploit spatially correlated graph information and deep operator networks to approximate the solution operator of dynamical systems. The resulting DeepGraphONet can then predict the dynamics within a given short/medium-term time horizon by observing a finite history of the graph state information. Furthermore, we design our DeepGraphONet to be resolution independent. That is, we do not require the finite history to be collected at the exact/same resolution. In addition, to disseminate the results from a trained DeepGraphONet, we design a zero-shot learning strategy that enables using it on a different subgraph. Finally, empirical results on the transient stability prediction problem of power grids and traffic flow forecasting problem of a vehicular system illustrate the effectiveness of the proposed DeepGraphONet.
Training an effective deep learning model to learn ocean processes involves careful choices of various hyperparameters. We leverage DeepHyper’s advanced search algorithms for multiobjective optimization, streamlining the development of neural networks tailored for ocean modeling. The focus is on optimizing Fourier neural operators (FNOs), a data-driven model capable of simulating complex ocean behaviors. Selecting the correct model and tuning the hyperparameters are challenging tasks, requiring much effort to ensure model accuracy. DeepHyper allows efficient exploration of hyperparameters associated with data preprocessing, FNO architecture-related hyperparameters, and various model training strategies. We aim to obtain an optimal set of hyperparameters leading to the most performant model. Moreover, on top of the commonly used mean squared error for model training, we propose adopting the negative anomaly correlation coefficient as the additional loss term to improve model performance and investigate the potential trade-off between the two terms. The numerical experiments show that the optimal set of hyperparameters enhanced model performance in single timestepping forecasting and greatly exceeded the baseline configuration in the autoregressive rollout for long-horizon forecasting up to 30 days. Utilizing DeepHyper, we demonstrate an approach to enhance the use of FNO in ocean dynamics forecasting, offering a scalable solution with improved precision.