Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “FEATURE”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 973 records · Page 54

Hidden Features: How Subsurface and Landscape Heterogeneity Govern Hydrologic Connectivity and Stream Chemistry in a Montane Watershed

ABSTRACT Hydrologic connectivity is defined as the connection among stores of water within a watershed and controls the flux of water and solutes from the subsurface to the stream. Hydrologic connectivity is difficult to quantify because it is goverened by heterogeniety in subsurface storage and permeability and responds to seasonal changes in precipitation inputs and subsurface moisture conditions. How interannual climate variability impacts hydrologic connectivity, and thus stream flow generation and chemistry, remains unclear. Using a rare, four‐year synoptic stream chemistry dataset, we evaluated shifts in stream chemistry and stream flow source of Coal Creek, a montane, headwater tributary of the Upper Colorado River. We leveraged compositional principal component analysis and end‐member mixing to evaluate how seasonal and interannual variation in subsurface moisture conditions impacts stream chemistry. Overall, three main findings emerged from this work. First, three geochemically distinct end members were identified that constrained stream flow chemistry: reach inflows, and quick and slow flow groundwater contributions. Reach inflows were impacted by historic base and precious metal mine inputs. Bedrock fractures facilitated much of the transport of quick flow groundwater and higher‐storage subsurface features (e.g., alluvial fans) facilitated the transport of slow flow groundwater. Second, the contributions of different end members to the stream changed over the summer. In early summer, stream flow was composed of all three end members, while in late summer, it was composed predominantly of reach inflows and slow flow groundwater. Finally, we observed minimal differences in proportional composition in stream chemistry across all four years, indicating seasonal variability in subsurface moisture and spatial heterogeneity in landscape and geologic features had a greater influence than interannual climate fluctuation on hydrologic connectivity and stream water chemistry. These findings indicate that mechanisms controlling solute transport (e.g., hydrologic connectivity and flow path activation) may be resilient (i.e., able to rebound after perturbations) to predicted increases in climate variability. By establishing a framework for assessing compositional stream chemistry across variable hydrologic and subsurface moisture conditions, our study offers a method to evaluate watershed biogeochemical resilience to variations in hydrometeorological conditions.

Johnson, Keira [College of Earth, Ocean, and Atmos↗

A distinctive feature of the surface structure of quasicrystals

This paper reviews a feature of atomically-clean quasicrystal surfaces that distinguishes them from surfaces of crystalline materials. That feature is a high degree of heterogeneity among different terraces, and among structurally-identical adsorption sites. The heterogeneity can be both structural and chemical in origin. A large variability is expected even for a surface which is perfectly bulk-terminated, and we call this intrinsic heterogeneity. Additional variability can derive from the surface preparation process, which can yield metastable structures. We call this extrinsic heterogeneity. Experimental evidence is given for both cases. This heterogeneity can be an important factor in understanding and predicting surface phenomena such as chemisorption.

36 MATERIALS SCIENCE↗

Ca X ML: Chemistry‐informed machine learning explains mutual changes between protein conformations and calcium ions in calcium‐binding proteins using structural and topological features

Proteins' flexibility is a feature in communicating changes in cell signaling instigated by binding with secondary messengers, such as calcium ions, associated with the coordination of muscle contraction, neurotransmitter release, and gene expression. When binding with the disordered parts of a protein, calcium ions must balance their charge states with the shape of calcium-binding proteins and their versatile pool of partners depending on the circumstances they transmit. Accurately determining the ionic charges of those ions is essential for understanding their role in such processes. However, it is unclear whether the limited experimental data available can be effectively used to train models to accurately predict the charges of calcium-binding protein variants. Here, we developed a chemistry-informed, machine-learning algorithm that implements a game theoretic approach to explain the output of a machine-learning model without the prerequisite of an excessively large database for high-performance prediction of atomic charges. We used the ab initio electronic structure data representing calcium ions and the structures of the disordered segments of calcium-binding peptides with surrounding water molecules to train several explainable models. Network theory was used to extract the topological features of atomic interactions in the structurally complex data dictated by the coordination chemistry of a calcium ion, a potent indicator of its charge state in protein. Our design created a computational tool of Ca X ML, which provided a framework of explainable machine learning model to annotate ionic charges of calcium ions in calcium-binding proteins in response to the chemical changes in an environment. Our framework will provide new insights into protein design for engineering functionality based on the limited size of scientific data in a genome space.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Characterizing climate pathways using feature importance on echo state networks

The 2022 National Defense Strategy of the United States listed climate change as a serious threat to national security. Climate intervention methods, such as stratospheric aerosol injection, have been proposed as mitigation strategies, but the downstream effects of such actions on a complex climate system are not well understood. The development of algorithmic techniques for quantifying relationships between source and impact variables related to a climate event (i.e., a climate pathway) would help inform policy decisions. Data-driven deep learning models have become powerful tools for modeling highly nonlinear relationships and may provide a route to characterize climate variable relationships. In this paper, we explore the use of an echo state network (ESN) for characterizing climate pathways. ESNs are a computationally efficient neural network variation designed for temporal data, and recent work proposes ESNs as a useful tool for forecasting spatiotemporal climate data. However, ESNs are noninterpretable black-box models along with other neural networks. The lack of model transparency poses a hurdle for understanding variable relationships. We address this issue by developing feature importance methods for ESNs in the context of spatiotemporal data to quantify variable relationships captured by the model. We conduct a simulation study to assess and compare the feature importance techniques, and we demonstrate the approach on reanalysis climate data. In the climate application, we consider a time period that includes the 1991 volcanic eruption of Mount Pinatubo. This event was a significant stratospheric aerosol injection, which acts as a proxy for an anthropogenic stratospheric aerosol injection. Furthermore, we are able to use the proposed approach to characterize relationships between pathway variables associated with this event that agree with relationships previously identified by climate scientists.

black-box models↗

Association of qualitative and quantitative imaging features on multiphasic multidetector CT with tumor grade in clear cell renal cell carcinoma

Purpose: The purpose of the study was to determine if enhancement features and qualitative imaging features on multiphasic multidetector computed tomography (MDCT) were associated with tumor grade in patients with clear cell renal cell carcinoma (ccRCC). Methods: In this retrospective, IRB approved, HIPAA-compliant, institutional review board-approved study with waiver of informed consent, 127 consecutive patients with 89 low grade (LG) and 43 high grade (HG) ccRCCs underwent preoperative four-phase MDCT in unenhanced (UN), corticomedullary (CM), nephrographic (NP), and excretory (EX) phases. Previously published quantitative (absolute peak lesion enhancement, absolute peak lesion enhancement relative to normal enhancing renal cortex, 3D whole lesion enhancement and the wash-in/wash-out of enhancement within the 3D whole lesion ROI) and qualitative (enhancement pattern; presence of necrosis; pattern of; tumor margin; tumor–parenchymal interface, tumor–parenchymal interaction; intratumoral vascularity; collecting system infiltration; renal vein invasion; and calcification) assessments were obtained for each lesion independently by two fellowship-trained genitourinary radiologists. Comparisons between variables included χ{sup 2}, ANOVA, and student t test. p values less than 0.05 were considered to be significant. Inter-reader agreement was obtained with the Gwet agreement coefficient (AC1) and standard error (SE) was reported. Results: No significant differences were observed between the LG and HG ccRCC cohorts with respect to absolute peak lesion enhancement and relative lesion enhancement ratio. There was a significant inverse correlation between low and high grade ccRCC and tumor enhancement the NP (71 HU vs. 54 HU, p < 0.001) and EX (52 HU vs. 39 HU, p < 0.001) phases using the 3D whole lesion ROI method. The percent wash-in of 3D enhancement from the UN to the CM phase was also significantly different between LG and HG ccRCCs (352% vs. 255%, p = 0.003). HG lesions showed significantly more calcification, necrosis, collecting system infiltration and ill-defined tumor margins (p < 0.05). Overall agreement between the two readers had a mean AC1 of 0.8172 (SE 0.0235). Conclusions: Quantitatively, high grade ccRCC had significantly lower whole lesion enhancement in the NP and EX phases on MDCT. Qualitatively, high grade ccRCC were significantly more likely to be associated with calcifications, necrosis, collecting system infiltration, and an ill-defined tumor margin.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Computed tomographic features of abdominal tuberculosis: unmask the impersonator!

Purpose: Abdominal tuberculosis (ATB) mimics various infectious, inflammatory, and neoplastic conditions and hence requires a high index of suspicion for accurate diagnosis, especially in low prevalence areas. It is difficult to consistently establish a histopathological diagnosis of ATB which underlines the importance of supportive evidences for institution of prompt empirical therapy to prevent associated morbidity and mortality. Methods: We retrospectively evaluated clinical and imaging features of 105 ATB cases and classified their CT findings based on peritoneal, lymph node, bowel, and solid organ involvement. Concomitant pulmonary and extra-pulmonary involvement was assessed. Results: Abdominal pain (78.1%) followed by fever (42.9%) were the commonest presenting symptoms. Peritoneal TB (77.14%) most commonly presented with a mix of ascites (49.38%), peritoneal (28.40%), and omental involvement (27.16%). Lymphadenopathy (57.1%) most commonly presented as necrotic nodes (81.67%) at mesenteric, peripancreatic, periportal, and upper paraaortic regions. Commonest site of bowel involvement (cumulative of 62.85%) was ileocecal region, with the commonest pattern of involvement being circumferential bowel wall thickening without bowel stratification with mild luminal narrowing. Hepatic (13.33%) and splenic (16.2%) involvement predominantly presented as multiple microabscesses. Adrenal and pancreatic involvement was noted in 4.7% and 1.9% of patients, respectively. 38.1% patients showed concomitant pulmonary and extra-pulmonary TB. Conclusion: ATB has varied radiological features; however, peritoneal involvement in the form of mild ascites, smooth peritoneal thickening, smudgy omentum, multi-focal bowel involvement, necrotic nodes, and multiple visceral microabscesses point towards a diagnosis of ATB in appropriate clinical setting.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

CT and MR imaging features of pancreatic adenosquamous carcinoma and their correlation with prognosis

Purpose: To retrospectively investigate the computed tomography (CT) and magnetic resonance (MR) imaging features of pancreatic adenosquamous carcinoma (PASC) and the association between imaging findings and prognosis. Materials and methods: CT, MR images of 26 patients with PASC were analyzed. Clinical symptoms, tumor markers, and patients’ survival were recorded. Tumor attenuation, enhancement pattern and degree, vessel involvement, adjacent tissue invasion and metastasis were evaluated. The association between imaging features and overall survival (OS) were also assessed using Cox proportional hazards ratio model. Results: Fourteen masses were found in the head of the pancreas and 12 in the body/tail. The mean tumor size was 4.47 ± 1.76 cm. PASC usually showed ill-defined (96.2%), lobulated (76.9%) and predominantly solid mass (92.3%). Ring enhancement in the peripheral area of the tumor was commonly seen (76.9%). Vessel invasion was seen in 17 cases (65.4%), encasement of adjacent arteries in 7 cases (26.9%), upstream main pancreatic duct (MPD) dilatation in 16 cases (61.5%) and double duct sign in 9 cases (34.6%). Multivariate Cox proportional hazards model demonstrated that patients with vessel invasion may predict a poor prognosis (p = 0.037). Conclusion: PASC tends to be an ill-defined solid mass with peripheral ring enhancement, and relatively poor enhancement in the central area. PASC may also show vessel invasion, vessel encasement and upstream MPD dilatation. Vessel invasion may indicate a poor prognosis.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Using In Situ Neutron Diffraction to Isolate Specific Features of Additively Manufactured Microstructures in 304L Stainless Steel and Identify Their Effects on Macroscopic Strength

Additive manufacturing of metal components results in unique microstructures with, necessarily, mechanical properties that are distinct from conventionally produced components. In this paper, four distinct microstructural features associated with directed energy deposition of 304L stainless steels, their stability, and their influences on flow strength were examined. These were (1) high dislocation density comparable with deformed materials, (2) increased ferrite content, (3) local chemical heterogeneity, and (4) tortuous grain morphology. In situ neutron diffraction measurements were used to monitor the evolution of the as-built microstructure during post-build heat treatment and relate the specific microstructural features to the strength behavior of the material following the heat treatment. The increased flow strength of the additively manufactured material relative to wrought counterparts is found to be due primarily to an increased dislocation density in the as-built material. However, the increased dislocation density does not completely account for the increased strength and it is hypothesized that some of the additional strength is related to the unique AM grain structure.

36 MATERIALS SCIENCE↗

Deformational Features and Microstructure Evolution of Copper Fabricated by a Single Pass of the Elliptical Cross-Section Spiral Equal-Channel Extrusion (ECSEE) Process

The objectives of this work are to study the deformational feature, textures, microstructures, and dislocation configurations of ultrafine-grained copper processed by the process of elliptical cross-section spiral equal-channel extrusion (ECSEE). The deformation patterns of simple shear and pure shear in the ECSEE process were evaluated with the analytical method of geometric strain. The influence of the main technical parameters of ECSEE die on the effective strain distribution on the surface of ECSEE-fabricated samples was examined by the finite element simulation. The high friction factor could improve the effective strain accumulation of material deformation. Moreover, the pure copper sample fabricated by ECSEE ion shows a strong rotated cube shear texture. The refining mechanism of the dislocation deformation is dominant in copper processed by a single pass of ECSEE. The inhomogeneity of the micro-hardness distribution on the longitudinal section of the ECSEE-fabricated sample is consistent with the strain and microstructure distribution features.

36 MATERIALS SCIENCE↗

Optimization of stochastic feature properties in laser powder bed fusion

Process parameter selection in laser powder bed fusion (LPBF) controls the as-printed dimensional tolerances, pore formation, surface quality and microstructure of printed metallic structures. Measuring the stochastic mechanical performance for a wide range of process parameters is cumbersome both in time and cost. As such, in this study, we overcome these hurdles by using high-throughput tensile (HTT) testing of over 250 dogbone samples to examine process-driven performance of strut-like small features, ~1 mm 2 in austenitic stainless steel (316 L). The output mechanical properties, porosity, surface roughness and dimensional accuracy were mapped across the printable range of laser powers and scan speeds using a continuous wave laser LPBF machine. Tradeoffs between ductility and strength are shown across the process space and their implications are discussed. While volumetric energy density deposited onto a substrate to create a melt-pool can be a useful metric for determining bulk properties, it was not found to directly correlate with output small feature performance.

316 L stainless steel↗

Reduction of baseplate distortion during directed energy deposition using compliant features

Distortion in additive manufacturing (AM) remains a barrier to its adoption in precision industries. Baseplate warpage is one such issue which compromises the feasibility of post-process precision machining. The restriction to thermal contraction of the deposited part, imposed by the baseplate, generates bending moments, that in turn causes warpage. A novel distortion mitigation strategy using baseplates with integrated compliant features, which enables thermal contraction of the build is presented. Six unique design concepts are evaluated, including a solid reference, using two deposition geometries. A laser, hot-wire, directed energy deposition (DED) process is used to deposit a symmetric cylindrical part (C-part) and a T-shaped asymmetric part (T-part). Flatness deviation of baseplates is measured using structured light 3D scanning. Measurements reveal a reduction in net flatness deviation of 58.8% for the C-part and 40.9% for the T-part, compared to the solid reference. While most designs yielded reductions exceeding 30% and 20% for the C- and T-parts, respectively, one configuration resulted in increased deviation. Finite element (FE) simulations are performed to elucidate the underlying mechanisms affecting distortion of compliant baseplates during DED. Despite variations between predictions and measurements, agreement in the general trend is observed. It also revealed that initial flatness errors in baseplates significantly affect its deviation during deposition. Predictions indicate that compliant features significantly affect the thermal distribution as well as the evolution of flatness deviation in the baseplate during deposition. Notably, one design exhibited a reduction in distortion during cooling, following its initial increase during deposition. FE predictions show a maximum reduction of 60.9% and 38.8% in net flatness deviation for the C- and T-parts, respectively. The performance of compliant baseplates is found to be governed by both its the thermal and mechanical characteristics, which are crucial factors to be considered during design.

Mathews, Ritin [ORNL] (ORCID:0000000301440828)↗

Industrial control system device classification using network traffic features and neural network embeddings

Characterization of modern cyber–physical Industrial Control System (ICS) devices is critical to the evaluation of their security posture and an understanding of the underlying industrial processes with which they interact. In this work, we address two related ICS device identification tasks: (1) separating ICS from non-ICS devices and (2) identifying specific ICS device types. We propose two distinct methods (one based on the existing IP2Vec method, and a novel traffic-features-based method) for achieving the first task. For transferability of the first task between two datasets, the traffic-features-based method performs significantly better (75% overall accuracy) compared to IP2Vec (22.5% overall accuracy). We further propose a novel method called DNP2Vec to address the second task. DNP2Vec is evaluated on two different datasets and achieves perfect multi-class classification accuracy (100%) for both datasets.

42 ENGINEERING↗

Adipose tissue-deprived stem cells acquire cementoblast features treated with dental follicle cell conditioned medium containing dentin non-collagenous proteins in vitro

Highlights: {yields} In this study we examine the effects of dental follicle cell conditioned medium (DFCCM) containing dentin non-collagenous proteins (dNCPs) on differentiation of ADSCs. {yields} We examined that ADSCs treated with dNCPs/DFCCM underwent morphological changes and significantly lost their proliferative capacity. {yields} dNCPs/DFCCM enhanced the mineralization behaviour and mineralization-related marker expression of ADSCs. {yields} ADSCs acquired cementoblast features in vitro with dNCPs/DFCCM treatment. -- Abstract: Adipose tissue-derived stem cells (ADSCs), which are easily harvested and show excellent pluripotency potential, have generated considerable interest in regenerative medicine. In this study, the differentiation of ADSCs was assessed after treatment with dental follicle cell conditioned medium (DFCCM) containing dentin non-collagenous proteins (dNCPs). ADSCs exhibited a fibroblast-like morphology and high proliferative capacity. However, after treatment with dNCPs/DFCCM, ADSCs changed from a fibroblast-like to cementoblast-like morphology and significantly lost their proliferative capacity. Alkaline phosphatase activity and in vitro mineralization behaviour of ADSCs were significantly enhanced. Mineralization-related markers including cementum attachment protein, bone sialoprotein, osteocalcin, osteopontin and osteonectin were detected at mRNA or protein levels, whereas dentin sialophosphoprotein and dentin sialoprotein were not detected, implying a cementoblast-like phenotype. These results demonstrate that ADSCs acquired cementoblast features in vitro with dNCPs/DFCCM treatment and could be a potential source of cementogenic cells for periodontal regeneration.

60 APPLIED LIFE SCIENCES↗

Down-regulated expression of the protein-tyrosine phosphatase 1B (PTP1B) is associated with aggressive clinicopathologic features and poor prognosis in hepatocellular carcinoma

Highlights: Black-Right-Pointing-Pointer PTP1B protein showed decreased expression in 67.79% of the HCC patients. Black-Right-Pointing-Pointer Low PTP1B expression predicts poor prognosis of HCC. Black-Right-Pointing-Pointer Low PTP1B expression is correlated with expansion of OV6{sup +} tumor-initiating cells. Black-Right-Pointing-Pointer Down-regulation of PTP1B is associated with activation of Wnt/{beta}-Catenin signaling. -- Abstract: The protein-tyrosine phosphatase 1B (PTP1B) is a classical non-transmembrane protein tyrosine phosphatase that plays a key role in metabolic signaling and can exert both tumor suppressing and tumor promoting effects in different cancers depending on the substrate involved and the cellular context. However, the expression level and function of PTP1B in hepatocellular carcinoma (HCC) remain unclear. In this study, PTP1B expression was detected by immunohistochemistry in normal liver tissue (n = 16) and hepatocellular carcinoma (n = 169). The correlations between PTP1B expression level and clinicopathologic features and patient survival were also analyzed. One hundred and eleven of 169 HCC patients (65.7%) had negative or low PTP1B expression in tumorous tissues, whereas normal tissues always expressed strong PTP1B. Decreased PTP1B expression was significantly associated with aggressive clinicopathologic features and poor prognosis. Immunohistochemistry also showed that low PTP1B expression level was correlated with high percentage of OV6{sup +} tumor-initiating cells (T-ICs) and high frequency of nuclear {beta}-Catenin expression in HCC specimens. Our findings demonstrate for the first time that the loss of inhibitory effect of PTP1B may contribute to progression and invasion of HCC through activation of Wnt/{beta}-Catenin signaling and expansion of liver T-ICs. PTP1B may serve as a valuable prognostic biomarker and potential therapeutic target in HCC.

60 APPLIED LIFE SCIENCES↗

Upregulated long non-coding RNA LOC90784 promotes cell proliferation and invasion and is associated with poor clinical features in HCC

A growing amount of literature has indicated that long non-coding RNAs (lncRNAs) are important factors in hepatocellular carcinoma (HCC) progression. However, the significance of lncRNAs in the progression and prognosis of liver cancer is largely unknown. In the present study, upregulated lncRNA LOC90784 was identified through integrative analysis of (GSE58043) and (GSE55191). Furthermore, associations between LOC90784 expression and the clinicopathological characteristics of patients were analyzed with a validated cohort 1 and the Cancer Genome Atlas (TCGA) cohort 2. We investigated the mechanisms by which this highly expressed lncRNA promotes HCC proliferation, invasion and migration via qRT-PCR, fluorescence in situ hybridization (FISH) staining, siRNA transfection, cell proliferation assays, Transwell and colony formation assays, flow cytometry analysis and Western blot. The results showed that LOC90784 expression levels were significantly higher in HCC cell lines and tissues and mainly localized in the cytoplasm. Knockdown of lncRNA LOC90784 expression inhibited proliferation and induced apoptosis and cell cycle arrest by promoting Bax and repressing CDK4 and Cyclin D1 protein expression; it also inhibited invasion and migration by repressing MMP2 and MMP9 expression in HCC cells. LOC90784 overexpression was associated with poor clinical features in the two cohorts and poor overall survival rates in HCC patients with clear resection margins (R0) in cohort 2. These results indicated that LOC90784 upregulation may be a critical oncogene and potential new biomarker in HCC. - Highlights: • LncRNA LOC90784 was overexpressed in HCC tissues and cell lines. • Highly expressed lncRNA LOC90784 was associated with poor clinical features in HCC. • Knockdown of lncRNA LOC90784 expression inhibited proliferation and induced apoptosis and cell cycle arrest. • Knockdown of lncRNA LOC90784 expression inhibited invasion and migration.

60 APPLIED LIFE SCIENCES↗

On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks

Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. Here in this work we investigate this limitation through the lens of Neural Tangent Kernel (NTK) theory and elucidate how PINNs are biased towards learning functions along the dominant eigen-directions of their limiting NTK. Using this observation, we construct novel architectures that employ spatio-temporal and multi-scale random Fourier features, and justify how such coordinate embedding layers can lead to robust and accurate PINN models. Numerical examples are presented for several challenging cases where conventional PINN models fail, including wave propagation and reaction–diffusion dynamics, illustrating how the proposed methods can be used to effectively tackle both forward and inverse problems involving partial differential equations with multi-scale behavior.

42 ENGINEERING↗

Multiresolution classification of turbulence features in image data through machine learning

During large-scale simulations, intermediate data products such as image databases have become popular due to their low relative storage cost and fast in-situ analysis. Serving as a form of data reduction, these image databases have become more acceptable to perform data analysis on. In this work, we present an image-space detection and classification system for extracting vortices at multiple scales through wavelet-based filtering. A custom image-space descriptor is used to encode a large variety of vortex-types and a machine learning system is trained for fast classification of vortex regions. By combining a radial-based histogram descriptor, a bag of visual words feature descriptor, and a support vector machine, our results show that we are able to detect and classify vortex features at various sizes at multiple scales. Once trained, our framework enables the fast extraction of vortices on new, unknown image datasets for flow analysis.

97 MATHEMATICS AND COMPUTING↗

Metal hydride composition-derived parameters as machine learning features for material design and H 2 storage

Though hydrogen is a promising energy carrier for a green future, many challenges persist. One is the difficulty in engineering storage solutions, with metal hydrides being a leading contender among solid-state strategies. To facilitate efficient searching of candidate materials, ridge regression, simple decision trees, random forest ensembles, and gradient boosting ensembles were employed to predict the energy of formation, with the random forest ensemble resulting in the lowest test set error. First, two public databases, Materials Project and HydPark, were searched for metal hydrides. Feature engineering was performed before the models were developed, resulting in electronegativity, density, atomic density, d-character, f-character, band gap, hydrogen weight fraction, magnetization, temperature, and pressure being retained. The models were then benchmarked by the lowest test error before a random forest ensemble was used to populate entries missing energy of formation. Furthermore, all were then scored by hydrogen storage capacity and energy of formation suitability. Readily available features including several derived from only the chemical formula which were found to be highly predictive. and so are promising for high-throughput screening of arbitrary novel hydride formulations and blends for thermodynamic feasibility.

25 ENERGY STORAGE↗