Engineering topics
Lu, Hongbing
Publications and source records attributed to Lu, Hongbing.
Machine learning based inverse modeling of full-field strain distribution for mechanical characterization of a linear elastic and heterogeneous membrane
Heterogeneous membranes or films are thin and soft structures with spatial variations in material property and thickness. Mechanical behavior of heterogeneous membranes is not well understood, mainly due to the difficulty in obtaining accurate and reliable material property data. To understand the mechanical behavior of these materials, accurate and efficient characterization methods for heterogeneous membranes are needed. Here, in this paper, an inverse method based on machine learning is developed to efficiently extract mechanical properties from full-field strain distributions. This approach is demonstrated on a flat heterogeneous membrane with uniform thickness formed by up to four linear elastic synthetic materials in a grid arrangement, and deforming in a moderate strain range (true strain ~10%). The results show that the machine learning method achieves accuracy comparable to the traditional inverse finite element method, and is 6 orders of magnitude faster in the demonstrated case studies.
Metamaterial-like aerogels for broadband vibration mitigation
We present metamaterial-like aerogels that provide broadband low-frequency vibration mitigation through their self-assembled locally resonant microstructures. Simulations confirm the presence of a negative dynamic stiffness for the aerogels.
Structural response of 3D-printed rubber lattice structures under compressive fatigue
In this work, a new direct UV curing, polyurethane-like, highly stretchable rubber, Elastic ToughRubber 90™ (ETR 90), is demonstrated for 3D printing via photocuring with Texas Instruments DLP® optical systems. An array of test specimens, comprising a sandwich structure with a body-centered cubic lattice core, are 3D printed from ETR 90. The compressive behavior of these specimens is characterized up to 10 k and up to 100 k fatigue-loading cycles. The internal, visually obstructed, structures under loads are visualized using in situ X-ray tomography. All lattice structures remain intact and full recovery is observed, indicating the cyclic durability of the 3D-printed rubber.
Thermal and Electrical Properties of Isocyanate Derived Organic Aerogels for Cryogenic Insulation Applications
The use and storage of cryogens such as liquefied nitrogen, helium, hydrogen among others requires reliable and efficient thermal insulation systems. Passive insulation from high performance materials that are well-known for their inherent low thermal conductivity would reduce the overall costs involved in design, manufacture and maintenance of such systems. One such class of materials are referred to as aerogels. These materials are known for their low density, high mesoporosity, high surface areas, low thermal conductivity and high acoustic impedance. Aerogels were invented by S.S. Kistler in 1931 and the most common type are those made of silica. However, the inherent fragility of silica aerogels makes them hard to mass produce, and therefore applications have been limited. A major breakthrough was introduced by our team almost 20 years ago with the invention of polymer crosslinked silica aerogels. Those materials shifted attention to all-polymer aerogels that have overcome all fragility issues associated with their inorganic counterparts. This study focuses on such polymeric aerogels that can be mass produced as large monoliths while maintaining the low thermal conductivity of traditional silica aerogels over a wide temperature range. Manufacturing flexibility of polymeric aerogels allows fabrication of blocks and sheets that can be applied in various configurations to insulate cryogenic and superconducting devices. The thermal conductivity with 80 K and room temperature boundary are reported as well as other properties (electrical), that need to be considered when designing devices for cryogenic applications.
Feature‐shared adaptive‐boost deep learning for invasiveness classification of pulmonary subsolid nodules in CT images
Purpose In clinical practice, invasiveness is an important reference indicator for differentiating the malignant degree of subsolid pulmonary nodules. These nodules can be classified as atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma (IAC). The automatic determination of a nodule's invasiveness based on chest CT scans can guide treatment planning. However, it is challenging, owing to the insufficiency of training data and their interclass similarity and intraclass variation. To address these challenges, we propose a two‐stage deep learning strategy for this task: prior‐feature learning followed by adaptive‐boost deep learning. Methods The adaptive‐boost deep learning is proposed to train a strong classifier for invasiveness classification of subsolid nodules in chest CT images, using multiple 3D convolutional neural network (CNN)‐based weak classifiers. Because ensembles of multiple deep 3D CNN models have a huge number of parameters and require large computing resources along with more training and testing time, the prior‐feature learning is proposed to reduce the computations by sharing the CNN layers between all weak classifiers. Using this strategy, all weak classifiers can be integrated into a single network. Results Tenfold cross validation of binary classification was conducted on a total of 1357 nodules, including 765 noninvasive (AAH and AIS) and 592 invasive nodules (MIA and IAC). Ablation experimental results indicated that the proposed binary classifier achieved an accuracy of with an AUC of 81.3 . These results are superior compared to those achieved by three experienced chest imaging specialists who achieved an accuracy of , , and , respectively. About 200 additional nodules were also collected. These nodules covered 50 cases for each category (AAH, AIS, MIA, and IAC, respectively). Both binary and multiple classifications were performed on these data and the results demonstrated that the proposed method definitely achieves better performance than the performance achieved by nonensemble deep learning methods. Conclusions It can be concluded that the proposed adaptive‐boost deep learning can significantly improve the performance of invasiveness classification of pulmonary subsolid nodules in CT images, while the prior‐feature learning can significantly reduce the total size of deep models. The promising results on clinical data show that the trained models can be used as an effective lung cancer screening tool in hospitals. Moreover, the proposed strategy can be easily extended to other similar classification tasks in 3D medical images.
Chemical, Physical, and Mechanical Characterization of Isocyanate Cross-linked Amine-Modified Silica Aerogels
We describe a new mechanically strong lightweight porous composite material obtained by encapsulating the skeletal framework of amine-modified silica aerogels with polyurea. The conformal polymer coating preserves the mesoporous structure of the underlying silica framework and the thermal conductivity remains low at 0.041 plus or minus 0.001 W m(sup -1 K(sup -1). The potential of the new cross-linked silica aerogels for load-carrying applications was determined through characterization of their mechanical behavior under compression, three-point bending, and dynamic mechanical analysis (DMA). A primary glass transition temperature of 130 C was identified through DMA. At room temperature, results indicate a hyperfoam behavior where in compression cross-linked aerogels are linearly elastic under small strains (less than 4%) and then exhibit yield behavior (until 40% strain), followed by densification and inelastic hardening. At room temperature the compressive Young's modulus and the Poisson's ratio were determined to be 129 plus or minus 8 MPa and 0.18, respectively, while the strain at ultimate failure is 77% and the average specific compressive stress at ultimate failure is 3.89 x 10(exp 5) N m kg(sup -1). The specific flexural strength is 2.16 x 10(exp 4) N m kg(sup -1). Effects on the compressive behavior of strain rate and low temperature were also evaluated.