Machine learning-assisted single-cell Raman fingerprinting for in situ and nondestructive classification of prokaryotes

iScience, Год журнала: 2021, Номер 24(9), С. 102975 - 102975

Опубликована: Авг. 11, 2021

Язык: Английский

---

Machine learning-assisted quantitative prediction of thermal decomposition temperatures of energetic materials and their thermal stability analysis

Energetic Materials Frontiers, Год журнала: 2023, Номер unknown

Опубликована: Сен. 1, 2023

In this study, machine learning (ML)-assisted regression modeling was conducted to predict the thermal decomposition temperatures and explore factors that correlate with stability of energetic materials (EMs). The performed based on a dataset consisting 885 various compounds using linear nonlinear algorithms. tree-based models established demonstrated acceptable predictive abilities, yielding low mean absolute error (MAE) 31°C. By analyzing through hierarchical classification, study insightfully identified affecting EMs' temperatures, overall accuracy improved targeted modeling. SHapley Additive exPlanations (SHAP) analysis indicated descriptors such as BCUT2D, PEOE_VSA, MolLog_P, TPSA played significant role, demonstrating process is influenced by multiple relating composition, electron distribution, chemical bond properties, substituent type molecules. Additionally, Carbon_contents Oxygen_Balance proposed for characterizing EMs showed strong correlations temperatures. trends their SHAP values most suitable ranges were 0.2–0.35 −65 −55, respectively. Overall, shows potential ML temperature prediction provides insights into characteristics molecular descriptors.

Язык: Английский

---

Probing of Neural Networks as a Bridge from Ab Initio Relevant Characteristics to Differential Scanning Calorimetry Measurements of High‐Energy Compounds

physica status solidi (RRL) - Rapid Research Letters, Год журнала: 2021, Номер 16(3)

Опубликована: Июнь 1, 2021

The relationships between the theoretical values calculated using density functional theory and experimental data derived from differential scanning calorimetry of high‐energy organic compounds are studied. number atoms bonds different types their lengths, minimum eigenfrequencies, atomization energies, ionization potentials, electron affinities, frontier orbital energies. amounts releasing heat (the first peaks higher than 1 kJ g −1 ) corresponding temperatures. Neural networks regression, factor, discriminant, cluster analysis applied to find dependencies data. It is found that amount cannot be predicted in general cases, whereas temperature can with a neural network an accuracy ≈30 °C. Cluster discriminant provides way for classification into three groups. Some these groups require particular rules prediction values.

Язык: Английский

---

Exploring an accurate machine learning model to quickly estimate stability of diverse energetic materials

iScience, Год журнала: 2024, Номер 27(4), С. 109452 - 109452

Опубликована: Март 8, 2024

High energy and low sensitivity have been the focus of developing new energetic materials (EMs). However, there has a lack quick accurate method for evaluating stability diverse EMs. Here, we develop machine learning prediction model with high accuracy bond dissociation (BDE) A reliable representative BDE dataset EMs is constructed by collecting 778 experimental compounds quantum mechanics calculation. To sufficiently characterize EMs, hybrid feature representation proposed coupling local target into global structure characteristics. alleviate limitation dataset, pairwise difference regression utilized as data augmentation advantage reducing systematic errors improving diversity. Benefiting from these improvements, XGBoost achieves best R2 0.98 MAE 8.8 kJ mol−1, significantly outperforming competitive models.

Язык: Английский

---

Estimating the lattice thermal conductivity of AlCoCrNiFe high-entropy alloy using machine learning

Journal of Applied Physics, Год журнала: 2024, Номер 135(13)

Опубликована: Апрель 3, 2024

The lattice thermal conductivity stands as a pivotal thermos-physical parameter of high-entropy alloys; nonetheless, achieving precise predictions the for alloys poses formidable challenge due to their complex composition and structure. In this study, machine learning models were built predict AlCoCrNiFe alloy based on molecular dynamic simulations. Our model shows high accuracy with R2, mean absolute percentage error, root square error test set is 0.91, 0.031, 1.128 W m−1 k−1, respectively. addition, low 2.06 k−1 (Al8Cr30Co19Ni20Fe23) 5.29 (Al0.5Cr28.5Co25Ni25.5Fe20.5) was successfully predicted, which good agreement results from dynamics mechanisms divergence are further explained through phonon density states elastic modulus. established provides powerful tool developing desired properties.

Язык: Английский

---

Enhancing structural analysis and electromagnetic shielding in carbon foam composites with applications in concrete integrating XGBoost machine learning, carbon nanotubes, and montmorillonite

Synthetic Metals, Год журнала: 2024, Номер 307, С. 117656 - 117656

Опубликована: Май 31, 2024

Язык: Английский

---

Application of Machine Learning to the Design of Energetic Materials: Preliminary Experience and Comparison with Alternative Techniques

Propellants Explosives Pyrotechnics, Год журнала: 2022, Номер 48(4)

Опубликована: Дек. 10, 2022

Abstract The last few years have seen a steep rise in the use of data‐driven methods different scientific fields historically relying on theoretical or empirical approaches. Chemistry is at forefront this paradigm shift due to longstanding computational tools involved calculation molecular structures and properties. In paper, we showcase examples from literature as well work progress our lab order give brief overview how these can benefit energetic materials community. A deep learning approach compared “traditional” QSPR semi‐empirical approaches for property prediction, specificities inherent are discussed. Deep generative models design new also presented. We conclude by giving view most promising strategies future silico generation satisfying performance/sensitivity trade‐off.

Язык: Английский

---

Machine learning-based prediction and interpretation of decomposition temperatures of energetic materials

Energetic Materials Frontiers, Год журнала: 2023, Номер 4(4), С. 254 - 261

Опубликована: Сен. 4, 2023

Exploring the application of machine learning (ML) in energetic materials (EMs) has been a hot research topic. Accordingly, prediction detonation properties EMs using ML methods attracted much attention. However, predictive models for thermal decomposition temperatures (Td) have scarcely reported. Furthermore, small datasets used these reports lead to weak generalization ability models. This study created dataset containing 1022 molecules with Td values 38–425 °C and determined an optimal model through training. The gradient boost regression (GBR) yielded coefficient determination (R2) 0.65 mean absolute error (MAE) 27.7 test set. further explored critical features, determining that accuracy was significantly influenced by descriptors representing molecular bond stability (i.e., BCUT metrics) atomic composition Molecular ID). Finally, analysis outlier structure indicated can be improved incorporating features related interactions. results this help gain deep understanding EM properties, particularly construction feature selection.

Язык: Английский

---

EM Database v1.0: A benchmark informatics platform for data-driven discovery of energetic materials

Energetic Materials Frontiers, Год журнала: 2023, Номер unknown

Опубликована: Сен. 1, 2023

Large-scale data demonstrates great significance for the discovery of novel energetic materials (EMs). However, open-source databases EMs are not readily available. In pursuit high-performance before synthetic attempts in laboratory, theoretically predicted properties and experimental results that can be easily accessed desired. Herein, a benchmark informatics platform EMs, namely EM Database, has been developed purpose storage sharing. Database v1.0 currently contains approximately 100000 unique compounds obtained through quantum chemistry (QC) calculations about 10000 extracted from literature. The QC database were via ground-state density functional using B3LYP/6-31G(d,p) method. These include geometrical conformation, electronic structures, (i.e., crystal density, enthalpy sublimation, molar heat formation, detonation pressure, velocity, heat, volume) models quantitative structure-property relationships. manually collected literature then doubly curated by our project team members. physicochemical, thermal, combustion, detonation, spectra, sensitivity properties. this paper, we also discuss techniques constructing present fundamental features database. is expected to serve as an effective forthcoming research on EMs.

Язык: Английский

---

Nanosculptured tungsten oxide: High-efficiency SERS sensor for explosives tracing

Journal of Hazardous Materials, Год журнала: 2024, Номер 476, С. 135171 - 135171

Опубликована: Июль 10, 2024

The accurate and rapid identification of explosives their toxic by-products is an important aspect safety protocols, forensic investigations pollution studies. Herein, surface-enhanced Raman scattering (SERS) used to detect different explosive molecules using improved substrate design by controllable oxidation the tungsten surface deposition Au layers. resulting furrow-like morphology formed at intersection Wulff facets increases nanoroughness improves SERS response over 300 % compared untreated surface. showed excellent reproducibility with a relative standard deviation less than 15 signal recovery 95 after ultrafast Ar/O2 plasma cleanings. detection limit for "dried on surface" measurement case was better 10−8 M moving scanning regime acquisition time 10 s, while "water droplets scenario LoD 10−7, which up 2 orders magnitude UV-Vis spectroscopy method. substrates were successfully classify molecular fingerprints HMX, Tetryl, TNB TNT, demonstrating efficiency sensor label-free screening in practice monitoring traces water medium.

Язык: Английский

---