Article
Thermodynamics
Jianfeng Yang, Guanyu Suo, Liangchao Chen, Zhan Dou, Yuanhao Hu
Summary: This paper proposes a data-driven method for corrosion prediction based on multi-source data. The method efficiently predicts key corrosion parameters and establishes prediction models. The optimized models show good prediction performance with real data, which can guide equipment safety management and hazard identification.
Article
Environmental Sciences
Xuan Li, Chaofan Wu, Michael E. Meadows, Zhaoyang Zhang, Xingwen Lin, Zhenzhen Zhang, Yonggang Chi, Meili Feng, Enguang Li, Yuhong Hu
Summary: The study analyzed the spatiotemporal variations in key factors influencing PM2.5 in Zhejiang Province, China from 2000 to 2019 using random forest and SHAP algorithms. The results showed that factors influencing PM2.5 varied significantly, with meteorological factors being the most important, followed by socioeconomic factors and topography/land cover factors. While GDP and transportation factors initially increased in importance, their contribution has declined recently, indicating that economic and infrastructural development may not necessarily lead to higher PM2.5 concentrations. Vegetation productivity, as indicated by NDVI changes, has become more essential in improving air quality in the region.
Article
Forestry
Zhentian Ding, Biyong Ji, Hongwen Yao, Xuekun Cheng, Shuhong Yu, Xiaobo Sun, Shuhan Liu, Lin Xu, Yufeng Zhou, Yongjun Shi
Summary: This study utilized data from 773 permanent plots in Zhejiang Province, China, to identify key variables influencing forest mortality and construct mortality prediction models. The findings revealed that soil and stand-related factors had significant effects on mortality rate, while terrain and climate factors were not statistically significant. The Random Forest model showed the best fitting and prediction effect for mortality, using variables such as stand age, tree height, ADBH, crown cover, humus layer thickness, and the biodiversity index.
Article
Computer Science, Interdisciplinary Applications
Leyre Torre-Tojal, Aitor Bastarrika, Ana Boyano, Jose Manuel Lopez-Guede, Manuel Grana
Summary: This article utilizes random forest models to estimate the biomass of Pinus radiata species in a region of the Basque Autonomous Community. By tuning the hyperparameters and conducting cross-validation, two models with high R-2 values were obtained. These models were then applied to the municipality of Orozko, predicting a biomass that is 16-18% higher than the predictions made by the Basque Government.
JOURNAL OF COMPUTATIONAL SCIENCE
(2022)
Article
Biochemistry & Molecular Biology
Zahray Amiri, Mina Nosrati, Payam Sharifan, Sara Saffar Soflaei, Susan Darroudi, Hamideh Ghazizadeh, Maryam Mohammadi Bajgiran, Fahimeh Moafian, Maryam Tayefi, Elahe Hasanzade, Mahdi Rafiee, Gordon A. Ferns, Habibollah Esmaily, Mahnaz Amini, Majid Ghayour-Mobarhan
Summary: Vitamin D supplementation response was evaluated using a random forest (RF) model to analyze 76 potential influencing factors, which revealed anthropometric factors, liver function tests, hematological parameters, and measurement of insulin sensitivity as highly correlated to the serum vitamin D response. The RF accuracy in predicting response was 93%, significantly higher than logistic regression at 40%.
Article
Environmental Sciences
Yihui Ge, Zhenchun Yang, Yan Lin, Philip K. Hopke, Albert A. Presto, Meng Wang, David Q. Rich, Junfeng Zhang
Summary: This study developed and evaluated two approaches to enhance the extrapolating ability of random forest models in areas with sparse monitoring data. By incorporating low-cost sensor data and using the regression-enhanced random forest method, the models showed improved accuracy and predictive capabilities.
ATMOSPHERIC ENVIRONMENT
(2023)
Article
Food Science & Technology
Ozgun Yucel, Fatih Tarlak
Summary: The main aim of this study was to develop machine learning-based regression methods to predict bacterial population on beef, using decision tree regression, generalized additive model regression, and random forest regression. Temperature, salt concentration, water activity, and acidity were used as predictor variables. The random forest regression provided the best prediction capability and can be reliably used to describe the survival and growth behavior of microorganisms in beef.
Article
Chemistry, Multidisciplinary
Swetha Chittam, Balakrishna Gokaraju, Zhigang Xu, Jagannathan Sankar, Kaushik Roy
Summary: There is a high demand for a big data repository for material compositions and analytics in the material science field. This study proposes a big data storage solution, data mining techniques, and machine learning models for material strength prediction. The Random Forest algorithm showed the highest accuracy of 87% on an independent dataset compared to Logistic regression and SVM with 72% and 78%, respectively.
APPLIED SCIENCES-BASEL
(2021)
Article
Environmental Sciences
Patrick Kacic, Frank Thonfeld, Ursula Gessner, Claudia Kuenzer
Summary: Monitoring forest conditions is crucial for preserving biodiversity, protecting carbon sinks, and promoting forest resilience in the face of global climate change. The semi-natural forests in Germany are facing severe challenges, such as insect infestation, due to the impacts of heatwaves and droughts. This study demonstrates the potential of using remote sensing sensors to generate comprehensive forest structure products for Germany, providing valuable information on recent forest conditions and supporting a better understanding of post-disturbance forest structure and resilience.
Article
Engineering, Environmental
Guoxin Huang, Xiahui Wang, Di Chen, Yipeng Wang, Shouxin Zhu, Tao Zhang, Lei Liao, Zi Tian, Nan Wei
Summary: This study proposes a novel hybrid data-driven framework to diagnose the contributing factors of soil heavy metal contamination using data mining methods. By applying a combination of naive Bayes, random forest, and bivariate local Moran's I on a large dataset, the study successfully classifies the enterprises and identifies the contaminating enterprises as contributing factors. The optimized random forest is used to determine the quantitative contributions of the contributing factors to heavy metal concentrations. Spatial clustering maps are generated to visually show the interactions and distributions between the heavy metal concentrations and the contributing factors.
JOURNAL OF HAZARDOUS MATERIALS
(2022)
Article
Engineering, Environmental
Guoxin Huang, Xiahui Wang, Di Chen, Yipeng Wang, Shouxin Zhu, Tao Zhang, Lei Liao, Zi Tian, Nan Wei
Summary: This study proposed a novel hybrid data-driven framework using machine learning algorithms to diagnose the factors influencing soil heavy metal contaminations. It demonstrated the effectiveness of the framework in analyzing contributing factors and obtaining rich information.
JOURNAL OF HAZARDOUS MATERIALS
(2022)
Article
Chemistry, Multidisciplinary
Kakaraparthi Kranthiraja, Akinori Saeki
Summary: This study demonstrates the efficient use of machine learning for the development of organic photovoltaics, achieving high correlation coefficients for polymer and non-fullerene small molecule acceptor OPVs and virtual screening of numerous conjugated polymers. The results show the potential for ML in predicting and optimizing the performance of OPVs with a relatively small number of experimental data points.
ADVANCED FUNCTIONAL MATERIALS
(2021)
Article
Engineering, Chemical
Baolei Liu, Changxuan Li
Summary: This study utilizes data mining techniques to clarify the production characteristics of different reserve types of tight gas reservoirs and establishes production characteristic judgment rules based on reserve size. The findings provide important references for the formulation of development technology policies and assessment of reservoir production potential for tight gas reservoirs.
Article
Engineering, Environmental
Jianfeng Yang, Ru Lia, Liangchao Chen, Yuanhao Hua, Zhan Dou
Summary: Equipment corrosion management is crucial for process safety in the oil refining industry, with corrosion diagnosis playing a significant role in supervision and protection. This paper introduces a data-driven solution for comprehensive evaluation and prediction of corrosion safety state, utilizing algorithms like Borderline-SMOTE and Random Forest (RF) to improve accuracy. The results show promising improvements in critical mechanism identification and corrosion state classification, indicating potential for enhancing equipment safety management.
PROCESS SAFETY AND ENVIRONMENTAL PROTECTION
(2022)
Article
Environmental Sciences
Reyhaneh Masoudi, Seyed Roohollah Mousavi, Pouyan Dehghan Rahimabadi, Mehdi Panahi, Asghar Rahmani
Summary: This study compares three popular machine learning algorithms (random forest, boosting regression tree, and multinomial logistic regression) for spatial prediction and mapping of groundwater quality classes and salinity hazard. Groundwater samples and hydro-chemical parameters were collected from an agriculturally intensive area in Fars Province, Iran. The performance of the machine learning models was evaluated using statistical indices, showing that the random forest model was the most accurate. The study also identified the most important parameters for explaining groundwater quality classes.
ENVIRONMENTAL MONITORING AND ASSESSMENT
(2023)
Article
Materials Science, Ceramics
Youxing He, Xiaobo Wang, Tao Guo, Kewei Gao, Xiaolu Pang
Summary: The oxidation resistance and tribological behavior of CrAlN/SiNx multilayer films with different interface structures were investigated. The multilayer film exhibited better oxidation resistance at 900 degrees C, but showed a reversed trend at 1000 degrees C. The stacking faults in the multilayer film were induced by the oxidation of SiNx. The multilayer film with a crystalline interface showed lower coefficient of friction (COF) and better tribological properties compared to the CrAlN film and the multilayer film with an amorphous interface.
CERAMICS INTERNATIONAL
(2023)
Article
Materials Science, Multidisciplinary
Huili Sun, Wenting Lv, Yu Yang, Dongdong Li, Luchun Yan, Xiaolu Pang, Yang He, Kewei Gao
Summary: By altering the structures of Cu-rich precipitates, the hydrogen embrittlement susceptibility of high-strength steels can be significantly reduced while maintaining their ultimate tensile strength. The 9R-structured Cu-rich precipitates show the strongest binding and trapping capacity with hydrogen.
Article
Engineering, Geological
Kewei Gao, Hai Lin, Muhannad T. Suleiman, Pierre Bick, Tomas Babuska, Xiwei Li, Jeffrey Helm, Derick G. Brown, Nabil Zouari
Summary: The paper investigates the shear and tensile strength of microbial-induced calcite precipitation (MICP) treatment at the particle scale. Three different test setups were developed to measure the strength of MICP-treated CaCO3 bonds between glass beads. The results indicate that the improved test setup provides more accurate measurements of the bond strength.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING
(2023)
Article
Materials Science, Coatings & Films
Qi Li, Tao Guo, Lin Liu, Xiaobo Wang, Kewei Gao, Xiaolu Pang
Summary: Adding soft metal phases into transition metal nitrides can effectively adjust the mechanical properties of films. TiN-Cu nanocomposite films with different orientations were prepared on MgO (100) and (110) substrates; the film with (110) orientation had a hardness of 40.4 GPa, while the (111) oriented film had a hardness of 28.1 GPa. The difference in hardness was attributed to the effect of orientation on Cu segregation at grain boundaries, resulting in different thicknesses of amorphous layers. This study provides a new approach for tuning the mechanical properties of ceramic thin films.
SURFACE & COATINGS TECHNOLOGY
(2023)
Article
Metallurgy & Metallurgical Engineering
Yanqi Tu, Saiyu Liu, Rongjian Shi, Shani Yang, Kewei Gao, Xiaolu Pang
Summary: The effects of cementite morphology on hydrogen trapping behavior in low-alloy pipeline steel were investigated in this study. A combination of microstructural observations, electrochemical hydrogen permeation experiments, and thermal desorption spectroscopy (TDS) analyses was used to quantitatively study hydrogen trapping behavior. The results showed that the presence of lamellar cementite in P-2 steel led to better irreversible hydrogen trapping ability compared to the granular cementite in P-1 steel.
ANTI-CORROSION METHODS AND MATERIALS
(2023)
Article
Materials Science, Multidisciplinary
Youpeng Song, Luchun Yan, Xiaolu Pang, Yanjing Su, Lijie Qiao, Kewei Gao
Summary: In this study, a series of nanocrystalline AlxCuy(FeCrNiCo)100_x-y (x = 6.44-50.51 at%, y = 8.49-40.31 at%) high entropy alloys (HEAs) were synthesized, and the effects of co-alloying Al and Cu on the corrosion resistance of FeCrNiCo HEAs were systematically investigated in a two-dimensional composition space. The microstructure and mechanical properties of typical alloy compositions were examined. The results showed that nanocrystalline Al16.29Cu13.04(FeCrNiCo)70.67 and Al19.59Cu17.08(FeCrNiCo)63.33 alloys exhibited balanced corrosion resistance, hardness, and plasticity. The underlying mechanisms for the effects of co-alloying Al and Cu on the corrosion resistance and mechanical properties were analyzed.
Article
Engineering, Mechanical
Ying Zheng, Huili Sun, Luchun Yan, Xiaolu Pang, Kewei Gao
Summary: The very high cycle fatigue behavior of precipitation hardening stainless steel (PHSS) with different aging times was investigated. The PHSS aged for 1 hour at 480 degrees C exhibited the highest fatigue strength, followed by 4 hours and 0.5 hours of aging. Both subsurface or internal inclusion-induced crack initiation (IICI) and non-inclusion-induced crack initiation (NIICI) were observed. The NIICI phenomenon was attributed to the nucleation and connection of microvoids within the delta-ferrite. Aging treatment had a significant influence on the dominant crack initiation mechanism by altering the interaction of dislocations and copper precipitates, which ultimately affected the fatigue performance.
INTERNATIONAL JOURNAL OF FATIGUE
(2023)
Article
Nanoscience & Nanotechnology
Rongjian Shi, Yanlin Wang, Supeng Lu, Saiyu Liu, Yanqi Tu, Shani Yang, Kewei Gao, Xu-Sheng Yang, Xiaolu Pang
Summary: The high-strength spring steel demonstrates superior hydrogen embrittlement resistance and satisfactory work hardening capacity after hydrogen pre-charging with dispersed multiple precipitates via multi-microalloying of 1.04 wt% Cr and 0.14 wt% V. The multiple precipitates include isolated (Mn, Cr, V)-enriched cementite, isolated V-enriched VC, and co-precipitation of cementite and VC, which act as effective hydrogen traps and reduce HE susceptibility by 23%, making this strategy beneficial for designing high-strength and HE-resistant automotive steels.
MATERIALS SCIENCE AND ENGINEERING A-STRUCTURAL MATERIALS PROPERTIES MICROSTRUCTURE AND PROCESSING
(2023)
Article
Nanoscience & Nanotechnology
Linhao Tan, Dongdong Li, Luchun Yan, Xiaolu Pang, Kewei Gao
Summary: This study fully elucidates the advantages of the novel precipitation-hardened martensitic stainless steel (novel steel) in terms of an excellent strength-ductility balance, using various methods and tests. The novel steel significantly increases yield and ultimate tensile strength, and improves elongation compared to conventional PH13-8Mo steel. It is mainly strengthened and toughened by fine grain strengthening, precipitation strengthening of multiple precipitates, and the TRIP effect. The addition of Cu leads to the formation of metastable austenite and contributes to the TRIP effect.
MATERIALS SCIENCE AND ENGINEERING A-STRUCTURAL MATERIALS PROPERTIES MICROSTRUCTURE AND PROCESSING
(2023)
Article
Metallurgy & Metallurgical Engineering
Rongjian Shi, Yanqi Tu, Liang Yang, Saiyu Liu, Shani Yang, Kewei Gao, Xu-Sheng Yang, Xiaolu Pang
Summary: This study explored the influence of pre-strain and microstructures on the hydrogen trapping behaviors in 1-GPa high-strength martensitic steel. The results showed that the trapped reversible and trapped irreversible hydrogen contents significantly increased after a pre-strain of 5%. The microstructural evolution revealed that the presence of concomitant dislocation cell-twin duplex microstructure and tangled dislocations contributed to the enhanced hydrogen trapping.
ACTA METALLURGICA SINICA-ENGLISH LETTERS
(2023)
Article
Chemistry, Physical
Yihong Zhao, Mingwei Ren, Xiangdong Zhu, Zhangyu Ren, Yaofang Hu, Huhu Zhao, Weiheng Wang, Yunbo Chen, Kewei Gao, Yujing Zhou
Summary: This study demonstrates a strategy for simultaneously optimizing the rolling resistance, wet skid, and cut resistance of reinforced rubber using a supramolecular filler. The morphology of the supramolecular filler in the rubber matrix was studied using DSC and TEM. The vulcanizates containing the supramolecular filler showed improved cut resistance and different mechanical properties at different temperatures.
Article
Materials Science, Multidisciplinary
Xiaoge Wang, Luchun Yan, Kewei Gao, Pengcheng Li, Jiujiu Hao
Summary: Zinc-aluminum layered double hydroxides (ZnAl-LDHs) film was prepared on magnesium alloys with different surface roughness using metallographic preparation combined with the hydrothermal method. The results showed that ZnAl-LDHs film grew most intensely when the surface roughness was at a minimum of 0.094 μm, reaching a thickness of 3.8 μm with a static contact angle of 84.34 degrees and a minimum corrosion current density of 1.12 x 10(-4) A/cm(2). The study also proposed the possible growth and corrosion prevention mechanisms of LDHs films.
Article
Engineering, Mechanical
Ying Zheng, Huili Sun, Luchun Yan, Xiaolu Pang, Kewei Gao
Summary: The very high cycle fatigue properties of 17-4 PH stainless steel aged for 1 h and 4 h after hydrogen charging were investigated. The research found that hydrogen reduces fatigue strength and changes the crack initiation mode primarily induced by inclusions. The presence of intergranular facets on the fracture surface suggests that hydrogen accelerates crack initiation and early crack growth. The hydrogen-granular bright facet (GBF) mechanism can be explained as lath martensite fracture and dislocation cells formation, involving hydrogen-enhanced localized plasticity (HELP) and hydrogen-enhanced decohesion (HEDE) mechanisms. The hydrogen trapping effect of e-copper precipitates suppresses the degradation of hydrogen on fatigue properties.
INTERNATIONAL JOURNAL OF FATIGUE
(2023)
Article
Engineering, Geological
Kewei Gao, Pierre Bick, Muhannad T. Suleiman, Xiwei Li, Jeffrey Helm, Derick G. G. Brown, Nabil Zouari
Summary: The wind erosion resistance of MICP-treated soil was investigated using wind tunnel experiments. The erosion modes and calcium carbonate content were analyzed to evaluate the effect of various factors on wind erosion mitigation. A MICP treatment protocol using bacteria medium followed by cementation medium was determined as the optimal treatment for increasing wind load resistance. The minimum calcium carbonate content necessary to mitigate wind erosion was also determined.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING
(2023)
Article
Metallurgy & Metallurgical Engineering
Ji Xiumei, Hou Meiling, Wang Long, Liu Jie, Gao Kewei
Summary: Based on the actual production data of the plate mill of Xingcheng Special Steel, two machine learning methods, including an extreme learning machine (ELM) combined with a traditional mathematical model and a deep learning framework based on TensorFlow, were proposed to predict deformation resistance and improve accuracy. The first method improved the structural form of the original model and optimized the neural network parameters, while the second method built deep neural networks with different structures and utilized optimization algorithms to enhance the model's generalizability and stability. The results showed significant improvement in prediction accuracy and application efficacy.
ACTA METALLURGICA SINICA
(2023)
