Mean R2 Values Research Articles

Hybrid LSTM-Based Fractional-Order Neural Network for Jeju Island’s Wind Farm Power Forecasting

Efficient integration of wind energy requires accurate wind power forecasting. This prediction is critical in optimising grid operation, energy trading, and effectively harnessing renewable resources. However, the wind’s complex and variable nature poses considerable challenges to achieving accurate forecasts. In this context, the accuracy of wind parameter forecasts, including wind speed and direction, is essential to enhancing the precision of wind power predictions. The presence of missing data in these parameters further complicates the forecasting process. These missing values could result from sensor malfunctions, communication issues, or other technical constraints. Addressing this issue is essential to ensuring the reliability of wind power predictions and the stability of the power grid. This paper proposes a long short-term memory (LSTM) model to forecast missing wind speed and direction data to tackle these issues. A fractional-order neural network (FONN) with a fractional arctan activation function is also developed to enhance generated wind power prediction. The predictive efficacy of the FONN model is demonstrated through two comprehensive case studies. In the first case, wind direction and forecast wind speed data are used, while in the second case, wind speed and forecast wind direction data are used for predicting power. The proposed hybrid neural network model improves wind power forecasting accuracy and addresses data gaps. The model’s performance is measured using mean errors and R2 values.

Influence of sample size, model selection, and land use on prediction accuracy of soil properties

Digital soil mapping (DSM) uses models that integrate field and laboratory data with environmental factors to predict soils and soil properties. The accuracy of predictions depends on the models used, the data collected, and the environmental factors. This study assesses the influence of sampling density and distribution, covariates, and modeling approach on the prediction accuracy of soil organic matter (SOM) and cation exchange capacity (CEC) at three sites in Indiana (ACRE; DPAC; SEPAC) with different management intensity and sampling designs. Ordinary Kriging (OK) and three machine learning models Cubist (CB), Random Forest (RF), and Regression Kriging (RK) were used. The Coefficient of Determination (R2), Root Mean Square Error (RMSE), Mean Square Error (MSE), concordance coefficient (pc), and bias were used for the accuracy assessment.The accuracy of the predictions was influenced by the site, sample density, model type, soil property, and their interactions. Sites were the single largest source of significant variation followed by sampling density and model type for both SOM and CEC. ACRE, with multiple fields and complex management practices, had a higher average RMSE and wider range of RMSE for SOM compared to SEPAC and DPAC with uniform management. At ACRE the RMSE for SOM decreased from 2.75 to 0.85 and from 17.38 to 3.61 for CEC with increasing number of samples from 36 (6 points/ha) to 66 (12points/ha), but did not change with further increases up to 146 samples. At SEPAC and DPAC the RMSE decreased only slightly at sampling densities above 5 points/ha and 1–2 points/ha, respectively (68 and 43 samples, respectively). Based on cross validation, all models performed poorly for SOM with R2 varying from 0.13 to 0.38, while for CEC the model performance varied widely from 0.11 to 0.64. The accuracy predictions for CEC were higher compared to SOM at all sites. Overall, RF performed better while OK performed the worst for both SOM and CEC. The mean R2 values across all sites were 0.35 (SOM) and 0.51 (CEC) for RF and 0.19 (SOM) and 0.17 (CEC) for OK. At ACRE, OK performed worse for both SOM and CEC with only slight differences among the other models, while at SEPAC and DPAC there were only slight differences among all models. Spatial predictions for CB, RF and RK were more detailed and conformed to soil landscape models compared to OK. The spatial differences between sampling densities for predicted SOM and CEC were greater in lower elevation areas compared to higher elevation areas. The results from this study demonstrate that the selection of modeling approach is site-specific, and depends on sampling density, soil properties and their interactions.

State of Health Estimation Procedure for Lithium-Ion Batteries Using Partial Discharge Data and Support Vector Regression

Battery aging is a complex phenomenon, and precise state of health (SoH) monitoring is essential for effective battery management. This paper presents a data-driven method for SoH estimation based on support vector regression (SVR), utilizing features built from both full and partial discharge capacity curves, as well as battery temperature data. It provides an in-depth discussion of the novel features constructed from different voltage intervals. Moreover, three combinations of features were analyzed, demonstrating how their efficacy changes across different voltage ranges. Successful results were obtained using the full discharge capacity curves, built from the full interval of 2 to 3.4 V and achieving a mean R2 value of 0.962 for the test set, thus showcasing the adequacy of the selected SVR strategy. Finally, the features constructed from the full voltage range were compared with ones built from 10 small voltage ranges. Similar success was observed, evidenced by a mean R2 value ranging between 0.939 and 0.973 across different voltage ranges. This indicates the practical applicability of the developed models in real-world scenarios. The tuning and evaluation of the proposed models were carried out using a substantial dataset created by Toyota, consisting of 124 lithium iron phosphate batteries.

Spatiotemporal Analysis and Prediction of Carbon Emissions from Energy Consumption in China through Nighttime Light Remote Sensing

With burgeoning economic development, a surging influx of greenhouse gases, notably carbon dioxide (CO2), has precipitated global warming, thus accentuating the critical imperatives of monitoring and predicting carbon emissions. Conventional approaches employed in the examination of carbon emissions predominantly rely on energy statistics procured from the National Bureau of Statistics and local statistical bureaus. However, these conventional data sources, often encapsulated in statistical yearbooks, exclusively furnish insights into energy consumption at the national and provincial levels, so the assessment at a more granular scale, such as the municipal and county levels, poses a formidable challenge. This study, using nighttime light data and statistics records spanning from 2000 to 2019, undertook a comparative analysis, scrutinizing various modeling methodologies, encompassing linear, exponential, and logarithmic models, with the aim of assessing carbon emissions across diverse spatial scales. A multifaceted analysis unfolded, delving into the key attributes of China’s carbon emissions, spanning total carbon emissions, per capita carbon emissions, and carbon emission intensity. Spatial considerations were also paramount, encompassing an examination of carbon emissions across provincial, municipal, and county scales, as well as an intricate exploration of spatial patterns, including the displacement of the center of gravity and the application of trend analyses. These multifaceted analyses collectively contributed to the endeavor of predicting China’s future carbon emission trajectory. The findings of the study revealed that at the national scale, total carbon emissions exhibited an annual increment throughout the period spanning 2000 to 2019. Secondly, upon an in-depth evaluation of model fitting, it was evident that the logarithmic model emerged as the most adept in terms of fitting, presenting a mean R2 value of 0.83. Thirdly, the gravity center of carbon emissions in China was situated within Henan Province, and there was a discernible overall shift towards the southwest. In 2025 and 2030, it is anticipated that the average quantum of China’s carbon emissions will reach 7.82 × 102 million and 25.61 × 102 million metric tons, with Shandong Province emerging as the foremost contributor. In summary, this research serves as a robust factual underpinning and an indispensable reference point for advancing the scientific underpinnings of China’s transition to a low-carbon economy and the judicious formulation of policies governing carbon emissions.

Rapid Determination of Wine Grape Maturity Level from pH, Titratable Acidity, and Sugar Content Using Non-Destructive In Situ Infrared Spectroscopy and Multi-Head Attention Convolutional Neural Networks.

In the pursuit of enhancing the wine production process through the utilization of new technologies in viticulture, this study presents a novel approach for the rapid assessment of wine grape maturity levels using non-destructive, in situ infrared spectroscopy and artificial intelligence techniques. Building upon our previous work focused on estimating sugar content (∘Brix) from the visible and near-infrared (VNIR) and short-wave infrared (SWIR) regions, this research expands its scope to encompass pH and titratable acidity, critical parameters determining the grape maturity degree, and in turn, wine quality, offering a more representative estimation pathway. Data were collected from four grape varieties-Chardonnay, Malagouzia, Sauvignon Blanc, and Syrah-during the 2023 harvest and pre-harvest phenological stages in the vineyards of Ktima Gerovassiliou, northern Greece. A comprehensive spectral library was developed, covering the VNIR-SWIR spectrum (350-2500 nm), with measurements performed in situ. Ground truth data for pH, titratable acidity, and sugar content were obtained using conventional laboratory methods: total soluble solids (TSS) (∘Brix) by refractometry, titratable acidity by titration (expressed as mg tartaric acid per liter of must) and pH by a pH meter, analyzed at different maturation stages in the must samples. The maturity indicators were predicted from the point hyperspectral data by employing machine learning algorithms, including Partial Least Squares regression (PLS), Random Forest regression (RF), Support Vector Regression (SVR), and Convolutional Neural Networks (CNN), in conjunction with various pre-processing techniques. Multi-output models were also considered to simultaneously predict all three indicators to exploit their intercorrelations. A novel multi-input-multi-output CNN model was also proposed, incorporating a multi-head attention mechanism and enabling the identification of the spectral regions it focuses on, and thus having a higher interpretability degree. Our results indicate high accuracy in the estimation of sugar content, pH, and titratable acidity, with the best models yielding mean R2 values of 0.84, 0.76, and 0.79, respectively, across all properties. The multi-output models did not improve the prediction results compared to the best single-output models, and the proposed CNN model was on par with the next best model. The interpretability analysis highlighted that the CNN model focused on spectral regions associated with the presence of sugars (i.e., glucose and fructose) and of the carboxylic acid group. This study underscores the potential of portable spectrometry for real-time, non-destructive assessments of wine grape maturity, thereby providing valuable tools for informed decision making in the wine production industry. By integrating pH and titratable acidity into the analysis, our approach offers a holistic view of grape quality, facilitating more comprehensive and efficient viticultural practices.

Sorption studies of Tetraoxophosphate (v) ions on some agricultural soils in Akwa Ibom State, eastern Niger Delta, Nigeria

The phosphate sorption capacity regulates principally the phosphorus concentration and available phosphorus in any soil solution. This work aims to study the phosphate sorption characteristics of five agricultural soils in Akwa Ibom State, Nigeria. The phosphate adsorption isotherm was obtained by shaking each soil sample with 20, 40, 60, 80, 100 and 120 mgL–1 for 24 h with 20 mL of KH2PO4 in 0.01 M CaCl2 solution. The Langmuir, Freundlich, Van Huay and Temkin isotherms were applied to study the PO43-sorption characteristics of the soil samples. The Freundlich model provided the best choice for phosphate sorption with mean R2 value of 0.9872. Results obtained for the phosphate sorbed (µg g–1soil) with 120 mgL-1 solution, heterogeneity factor of adsorption sites (nF), and the Freundlich constant (KF) (Lg–1) were 101(84.3), 0.515 and 7.355 for Onna soil; 62.8(52.3), 0.641 and 2.464 for Ikot-Amama 2; 60.6(50.5), 0.708 and 3.985 for Ogu-Itumbonuso soil; 59.3(49.5), 0.652 and 2.505 for Ikot-Amama 1 soils; 58.0(48.3), 0.704 and 3.682 for Ikot-Nkon soils, respectively. The equilibrium concentration for phosphates (EC-P) were used as desorption index to determine whether the soils could serve as potential sources for their bioavailability to plants and results ranged from 5.0 to 9.90 mgL–1, indicating high vulnerability to phosphate loss. The relationship between phosphate sorption parameters and soil physicochemical properties showed that the Freundlich adsorption intensity factor, nF, strongly correlated negatively with organic carbon and organic matter (r = -0.953 and -0.947; P<0.01), Fe and Mn (r = -0.911 and -0.812, P<0.05). The EC-P correlated strongly with soil pH(H2O) (r = 0.959, P<0.01), clay (r = 0.913, P<0.05), cation exchange capacity (CEC) (r = -0.824, P<0.05), Mn (r = -0.792, P<0.05) and Fe (r = -0.826, P<0.05). The study illustrated that soil management practices which involves the using organic fertilizers as the basis for PO43- sorption in relation to soil properties can be a useful tool for sustainable crop production and environmental soil-water contamination.

Experimental investigation and predictive modeling of compressive strength and electrical resistivity of graphene nanoplatelets modified concrete

Multipurpose cement composites are required for the sustainable development of the construction industry. Incorporation of nanomaterials can produce high-performing and multipurpose cement composites (CC). However, predicting the properties of these composites is challenging due to their complex composite nature and nonlinear behavior. This study investigates the integration of graphene nanoplatelets (GNPs) in concrete to enhance compressive strength and electrical properties. In addition to the experimental study, three machine learning techniques i.e., gene expression programming (GEP), support vector machine (SVM), and random forest regression (RFR), are employed to predict the compressive strength and electrical resistivity of graphene nanoplatelet modified concrete (GrNCC). The experimental results showed that GNP was evenly distributed throughout the matrix, and GrNCC exhibited enhanced compressive strength (CS) and fractional change in resistivity (FCR). GNP concentrations of 0.1% and 0.5% depict strength enhancement of 16.58% and 13.23% respectively. Similarly maximum FCR change of − 13%, and − 12.19% is observed. Moreover, GEP was the most robust ML model with mean R2 values of 0.988 and 0.74 for CS and ER respectively. In addition, SVM has R2 values of 0.93 and 0.71, and RFR shows R2 values of 0.899 and 0.671 for CS and ER respectively. GEP performed 52% and 67.5% better than SVM and RFR for CS, and 34.92% and 35.1% better than SVM and RFR for ER in terms of MAE. Furthermore, sensitivity analysis revealed that graphene content is the most important parameter contributing almost 27.3% and 29.2% for CS and ER respectively, followed by graphene dispersion and water content. Moreover, a graphical user interface (GUI) is made using a model that has been trained to predict the output values based on the input parameters. This simplifies the process and offers a valuable tool for utilizing the model's capabilities in civil engineering.

Excluding quartz content from the estimation of saturated soil thermal conductivity: Combined use of differential effective medium theory and geometric mean method

Saturated soil thermal conductivity (λsat) is the maximum soil thermal conductivity value of a given soil. Although it can be determined accurately with a heat pulse sensor, there are challenges to prepare fully saturated soil samples. Numerous models have been developed to estimate λsat, and among these, the geometric mean method (GMM) generally performs well. The GMM requires soil mineral composition or quartz content information, which is unavailable for most soils. Earlier studies commonly used assumed that quartz content (fquartz) was equal to sand content (fsand) or to 0.5 × fsand, which led to significant λsat estimation errors especially on coarse-textured soils. We derived a novel method to estimate λsat from soil porosity (ϕ) based on a combination of the GMM and differential effective medium theory (DEM). The new DEM-GMM approach has a single parameter, cementation exponent (m). Using a calibration dataset of 43 soils, we determined best fit m values for soils in three groups: 1.66 for Group I (fsand < 0.4), 1.62 for Group II (0.4 <= fsand < 1) and m = -1.34ϕ+1.70 for Group III (fsand = 1). Using best fit m values for different groups, the new model can estimate λsat values from ϕ. Independent validation results on another 46 soils showed that the new model outperformed the GMM method with the assumption that fquartz = fsand or fquartz = 0.5 × fsand. The mean RMSE, Bias and R2 values of the DEM-GMM approach were 0.202 W m−1 K−1, 0.013 W m−1 K−1 and 0.89, respectively, and corresponding values of the GMM with the two assumptions were 0.295 and 0.476 W m−1 K−1, 0.056 and -0.28 W m−1 K−1, 0.80 and 0.82, respectively. The robust performance of the DEM-GMM approach suggests that it can be incorporated into thermal conductivity models to accurately estimate the thermal conductivity of unsaturated soils.

Validity of an automated screening Dixon technique for quantifying hepatic steatosis in living liver donors.

This retrospective study aimed to evaluate the validity of an automated screening Dixon (e-DIXON) technique for quantifying hepatic steatosis in living liver-donor patients by comparison with magnetic resonance spectroscopy (MRS) as a reference standard. A total of 285 living liver-donor candidates were examined with the e-DIXON technique and single-voxel MRS to assess hepatic steatosis and iron deposition between January 2014 and February 2019. The sensitivity, specificity, and positive and negative predictive values (PPV and NPV) of the e-DIXON technique for hepatic steatosis were calculated. The mean fat signal fractions obtained in MRS were compared between the donors diagnosed with hepatic steatosis and the normal group. The mean R2 values of donors with or without hepatic siderosis also were compared. The e-DIXON technique diagnosed normal in 133 (47%), fat in 124 (44%), iron in one (0.4%), and a combination of both fat and iron in 27 (10%) donors. The sensitivity, specificity, PPV, and NPV ​​for diagnosing hepatic steatosis were 94%, 70%, 64%, and 96%, respectively. There was a significant difference in the mean fat signal fraction obtained in MRS between the steatosis and normal groups (p < 0.001), but R2 values were not significantly different between siderosis and normal groups (p = 0.11). The e-DIXON technique showed a strong correlation with MRS in fat measurement (r2 = 0.92, p < 0.001). The e-DIXON technique reliably screens for hepatic steatosis but may not accurate for detecting hepatic iron deposition.

Hyperlocal air pollution in an urban environment - measured with low-cost sensors

Air pollution levels can vary significantly over short distances, particularly in urban areas and near emission sources. This study examined the performance of low-cost sensor devices for monitoring levels of NO2, O3, and PM2.5 along two closely spaced (average 8 m) routes in Copenhagen, Denmark. One route was located near a lake (Route 1) and the other near a busy road (Route 2). The routes were walked in tandem for 84 h. The mode of deployment was determined using an accelerometer, gyroscope, and light sensor, achieving a 97.4 % accuracy rate. Field calibration with multivariate linear regression proved the most robust calibration model across pollutants, yielding mean R2-values of 0.64, 0.79, and 0.48 for NO2, O3, and PM2.5, respectively. The sensor intervariability was generally low, with mean R2-values of 0.84–0.94 for PM2.5 measured with optical particle sensors and 0.88–0.90 for NO2 and O3 measured with metal-oxide sensors. Results showed significantly higher NO2 concentrations on Route 2 (21.6 ± 6.6 ppb) compared to Route 1 (10.1 ± 4.0 ppb) during mornings. However, no significant differences in O3 and PM2.5 concentrations were observed. Our findings demonstrate that low-cost sensors can accurately quantify air pollution exposure in urban areas with high spatiotemporal resolution.

Characteristics of secondary inorganic aerosols and contributions to PM2.5 pollution based on machine learning approach in Shandong Province

Primary emissions of particulate matter and gaseous pollutants, such as SO2 and NOx have decreased in China following the implementation of a series of policies by the Chinese government to address air pollution. However, controlling secondary inorganic aerosol pollution requires attention. This study examined the characteristics of the secondary conversion of nitrate (NO3−) and sulfate (SO42−) in three coastal cities of Shandong Province, namely Binzhou (BZ), Dongying (DY), and Weifang (WF), and an inland city, Jinan (JN), during December 2021. Furthermore, the Shapley Additive Explanation (SHAP), an interpretable attribution technique, was adopted to accurately calculate the contributions of secondary formations to PM2.5. The nitrogen oxidation rate exhibited a significant dependence on the concentration of O3. High humidity facilitates sulfur oxidation. Compared to BZ, DY, and WF, the secondary conversion of NO3− and SO42− was more intense in JN. The light-gradient boosting model outperformed the random forest and extreme-gradient boosting models, achieving a mean R2 value of 0.92. PM2.5 pollution events in BZ, DY, and WF were primarily attributable to biomass burning, whereas pollution in Jinan was contributed by the secondary formation of NO3− and vehicle emissions. Machine learning and the SHAP interpretable attribution technique offer a precise analysis of the causes of air pollution, showing high potential for addressing environmental concerns.

Generation of Whole-Body FDG Parametric Ki Images from Static PET Images Using Deep Learning.

FDG parametric Ki images show great advantage over static SUV images, due to the higher contrast and better accuracy in tracer uptake rate estimation. In this study, we explored the feasibility of generating synthetic Ki images from static SUV ratio (SUVR) images using three configurations of U-Nets with different sets of input and output image patches, which were the U-Nets with single input and single output (SISO), multiple inputs and single output (MISO), and single input and multiple outputs (SIMO). SUVR images were generated by averaging three 5-min dynamic SUV frames starting at 60 minutes post-injection, and then normalized by the mean SUV values in the blood pool. The corresponding ground truth Ki images were derived using Patlak graphical analysis with input functions from measurement of arterial blood samples. Even though the synthetic Ki values were not quantitatively accurate compared with ground truth, the linear regression analysis of joint histograms in the voxels of body regions showed that the mean R2 values were higher between U-Net prediction and ground truth (0.596, 0.580, 0.576 in SISO, MISO and SIMO), than that between SUVR and ground truth Ki (0.571). In terms of similarity metrics, the synthetic Ki images were closer to the ground truth Ki images (mean SSIM = 0.729, 0.704, 0.704 in SISO, MISO and MISO) than the input SUVR images (mean SSIM = 0.691). Therefore, it is feasible to use deep learning networks to estimate surrogate map of parametric Ki images from static SUVR images.

Prediction of HIV-1 protease resistance using genotypic, phenotypic, and molecular information with artificial neural networks.

Drug resistance is a primary barrier to effective treatments of HIV/AIDS. Calculating quantitative relations between genotype and phenotype observations for each inhibitor with cell-based assays requires time and money-consuming experiments. Machine learning models are good options for tackling these problems by generalizing the available data with suitable linear or nonlinear mappings. The main aim of this study is to construct drug isolate fold (DIF) change-based artificial neural network (ANN) models for estimating the resistance potential of molecules inhibiting the HIV-1 protease (PR) enzyme. Throughout the study, seven of eight protease inhibitors (PIs) have been included in the training set and the remaining ones in the test set. We have obtained 11,803 genotype-phenotype data points for eight PIs from Stanford HIV drug resistance database. Using the leave-one-out (LVO) procedure, eight ANN models have been produced to measure the learning capacity of models from the descriptors of the inhibitors. Mean R2 value of eight ANN models for unseen inhibitors is 0.716, and the 95% confidence interval (CI) is [0.592-0.840]. Predicting the fold change resistance for hundreds of isolates allowed a robust comparison of drug pairs. These eight models have predicted the drug resistance tendencies of each inhibitor pair with the mean 2D correlation coefficient of 0.933 and 95% CI [0.930-0.938]. A classification problem has been created to predict the ordered relationship of the PIs, and the mean accuracy, sensitivity, specificity, and Matthews correlation coefficient (MCC) values are calculated as 0.954, 0.791, 0.791, and 0.688, respectively. Furthermore, we have created an external test dataset consisting of 51 unique known HIV-1 PR inhibitors and 87 genotype-phenotype relations. Our developed ANN model has accuracy and area under the curve (AUC) values of 0.749 and 0.818 to predict the ordered relationships of molecules on the same strain for the external dataset. The currently derived ANN models can accurately predict the drug resistance tendencies of PI pairs. This observation could help test new inhibitors with various isolates.

Satellite-Derived Bathymetry with Sediment Classification Using ICESat-2 and Multispectral Imagery: Case Studies in the South China Sea and Australia

Achieving coastal and shallow-water bathymetry is essential for understanding the marine environment and for coastal management. Bathymetric data in shallow sea areas can currently be obtained using SDB (satellite-derived bathymetry) with multispectral satellites based on depth inversion models. In situ bathymetric data are crucial for validating empirical models but are currently limited in remote and unapproachable areas. In this paper, instead of using the measured water depth data, ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2) ATL03 bathymetric points at different acquisition dates and multispectral imagery from Sentinel-2/GeoEye-1 were used to train and evaluate water depth inversion empirical models in two study regions: Shanhu Island in the South China Sea, and Heron Island in the Great Barrier Reef (GBR) in Australia. However, different sediment types also influenced the SDB results. Therefore, three types of sediments (sand, reef, and coral/algae) were analyzed for Heron Island, and four types of sediments (sand, reef, rubble and coral/algae) were analyzed for Shanhu Island. The results show that accuracy generally improved when sediment classification information was considered in both study areas. For Heron Island, the sand sediments showed the best performance in both models compared to the other sediments, with mean R2 and RMSE values of 0.90 and 1.52 m, respectively, representing a 5.6% improvement of the latter metric. For Shanhu Island, the rubble sediments showed the best accuracy in both models, and the average R2 and RMSE values were 0.97 and 0.65 m, respectively, indicating an RMSE improvement of 15.5%. Finally, bathymetric maps were generated in two regions based on the sediment classification results.

Eco-friendly de-lignified and raw Celosia argentea waste solid biofuel: Comparative studies and machine learning modelling

This study produced and compared the combustion properties of alkaline treated Celosia argentea (TCAB) and untreated Celosia argentea (UCAB) waste bio-briquettes using starch as binder. Heat energy of TCAB and UCAB were determined. Potential toxic elements, surface modification and bond orientation of briquettes were monitored using EDXRF, SEM and FTIR, respectively. Levenberg-Marquardt Back-Propagation based Artificial Neural Network (LMBP-ANN) and Fuzzy C-means Clustered Adaptive Neuro-Fuzzy Inference System (FCM-ANFIS) models were adopted for the bio-briquettes. Calorific values of 11.38 ± 0.20 MJ/Kg (UCAB) and 12.79 ± 0.25 MJ/Kg (TCAB) at p < 0.05 were recorded. EDXRF showed percentage reduction in Pb content of TCAB (1.425 %) as against 3.253 % in UCAB. FTIR recorded a CO stretch difference of 4 cm−1, while SEM shows morphological restructuring. FCM-ANFIS outperformed LMBP-ANN with root mean square error (RMSE), mean absolute percentage error (MAPE), mean absolute deviation (MAD) and R2-values of 0.0748, 2.4756, 0.0687 and 0.9323, respectively at the testing phase for heat content, suggesting strong agreement between the experimental data and predicted values. The study shows that alkaline pretreatment enhanced the combustionproperties and eco-friendliness of the solid-biofuel.

Early design stage evaluation of architectural factors in fire emergency evacuation of the buildings using Pix2Pix and explainable XGBoost model

ABSTRACT As a critical factor in architectural design, emergency evacuation is influenced by numerous parameters. Designers utilize modelling software to evaluate their sketches after completion ofbasic design. However, no various alternatives of early design stages could be assessed via simulations, since it is a time-consuming procedure. In this study, deep-learning algorithms have been adapted for the assessment of the evacuation process at early design stages. The main methods applied include an image-to-image translation with a conditional GAN (Pix2Pix) and Extreme Gradient Boosting (XGBoost). The developed Pix2Pix model generates the heat maps of possible route congestions with a Structural Similarity Index (SSIM) of 0.89. Besides, the XGBoost model predicts the evacuation time with the mean absolute error (MAE) and R2 values of 36 s and 0.94, respectively. This method generates the results of intended analyses at high speed and is a reliable alternative for time-consuming evacuation simulations in early design stages.

Universal correlation for falling film evaporation on a horizontal plain tube

Falling film evaporators are widely used in many industrial sectors, such as desalination, refrigeration & air-conditioning, and chemical processes. The heat transfer correlations suitable for wider operating conditions and various refrigerants are useful for designing falling film evaporators. Although many researchers developed the correlation for their own experimental data, there is a limited universal correlation for falling film evaporators. In addition, the dry-out effects were not considered in the available universal correlations. Therefore, this study proposes a new correlation for falling film evaporation on a single horizontal plain tube, which includes nucleate boiling, film evaporation, and dry-out effects. The proposed correlation covers a wider range of operating conditions (Heat flux is of 2.5–168 kWm−2, and film Reynolds number is up to 3500), various refrigerants (R134a, R245fa, R1234ze(E), R123, R32, R410A, R290, and R600a), and the large number of experimental data (1101 data in total). The proposed correlation is validated against the available standard correlation with regard to mean absolute deviation (MAD) and R2 values. The MAD of the new correlation is about 10%, respectively, which is the lowest among other correlations. Similarly, the R2 showed a value of 0.92, which is the highest compared to the existing correlations. A simplified correlation is also proposed.

Mapping the distribution, trends, and drivers of soil organic carbon in China from 1982 to 2019

The evaluation of soil organic carbon (SOC) dynamics and its driving factors is important for developing sustainable soil C sequestration practices. A key challenge is the lack of a time series of spatially explicit SOC estimates; most previous SOC estimates are static. In this study, we applied a digital soil mapping framework to model the relationship between SOC and environmental factors based on a combined dataset of 4695 soil samples collected in China from the early 1980 s and during the 2000–2014 period. We predicted the annual SOC distribution at a depth of 0–100 cm at a 1 km resolution for China between 1982 and 2019. We mapped trends of SOC change and identified change hotspots. To obtain the pixel-level association between SOC change and environmental factors, we used a linear function to quantify the contribution of individual predictors to SOC change. The results show that the accuracy of SOC prediction is acceptable, with a mean R2 value of 0.58 for validation. SOC storage in China was estimated to range from 89.03 to 97.41 Pg C and showed a slight increase (slope = 0.024 Pg C/yr) from 1982 to 2019. Specifically, two distinct phases of SOC were identified as follows: (1) from 1982 to 2004, a considerable decline occurred, which was mainly driven by increased temperature; (2) from 2004 to 2019, SOC increased slowly, resulting from the contribution of greening to sustaining C sequestration in soils under warming conditions. Approximately 17 % of all areas have been characterized by a significant decrease in SOC since 1982, predominantly in central and western China. Environmental controls exhibited large spatial variation, where approximately 69 % of all areas were predominantly controlled by climate factors. For grasslands, the SOC losses in northern China were largely driven by temperature and precipitation, while the SOC losses were primarily attributed to decreased organic inputs driven by grassland degradation on the Tibetan Plateau. Policy-driven land use change has led to an increase in SOC in some cropland areas (e.g., northeast and east China). The implementation of ecological restoration has shown a positive influence on SOC sequestration. This study highlights the diverse spatial patterns and heterogeneity of SOC changes and their controls. Given this spatial differentiation, spatially explicit assessments of soil C dynamics are important for the formulation of regional soil-based mitigation strategies and guidance for effective restoration interventions based on geographic location.

Electrochemical investigation of edible oils: Experimentation, electrical signatures, and a supervised learning–case study of adulterated peanut oils

Traditional approaches to characterize edible oils such as chemical, chromatographic and light absorption techniques are laborious, expensive, and bulky to implement. This paper presents the electrochemical impedance spectroscopy of 13 types of edible oils, a rapid robust approach to characterizing the electrical behavior of oils without sample preparation. This is achieved through probing the oils via oscillating electric fields to capture oil-specific electrical behaviors. The principal component analysis discriminates the oil types well and establishes repetitive behavioral trends, perceived as electrical signatures. This data is applied in a case study of adulterated peanut oils to quantify adulteration via supervised machine learning with batch-wise leave-one-out implementation. The mean absolute errors and R2 values measure 2.18–3.27 and 0.975–0.991 respectively across 4 test batches. This work provides an exemplar for the electrochemical study of edible oils, with potential for portable proof-of-value device configurations for rapid in situ analysis of edible oils and adulterated oils.

Application of machine learning for inter turn fault detection in pumping system

Pump fault diagnosis is essential for the maintenance and safety of the device as it is an important appliance used in various major sectors. Fault diagnosis at the proper time can reduce maintenance costs and save energy. This article uses a Simulink model based on mathematical equations to analyze the effects of parameter estimation of three-phase induction motor-based centrifugal pumps in inter-turn fault conditions. The inter-turn fault causes a massive in, a massive increase in current, which severely affects the parameters of both motor and pump. These have been analyzed by simulation through the Matlab Simulink model. Later, the results are verified by a hardware in loop (HIL) based simulator. In this paper, machine learning (ML) based artificial neural network (ANN) and ANFIS (ANN and Fuzzy) models have been applied for fault detection. ANN and ANFIS-based models provide a satisfactory level of accuracy. These models provide accurate training and testing results. Based on root mean square error (RMSE), R2, prediction accuracy, and mean validation value, these models are compared to find out which is more suitable for this experiment. Various supervised algorithms are compared with ANN, ANFIS, and lastly, found which is the most suitable for this experiment.