Due to the growing emphasis on sustainability and digitalization in today's business context, there are newly grounds to consider the need to address developing Digital Innovation (DI) alongside green practices to improve firm sustainability. The current research seeks to investigate from DI, which consists of include Artificial Intelligence and Internet of Things, as well as green practices which consists of renewable energy solutions and waste reductions, the constructs of corporate sustainability and business performance. To aid in understanding this relationship, a survey of 200 individuals across the diverse industries was collected then analyzed using descriptive statistics, as well as using correlation and regression approaches. Results indicated the mean of DI (2.9800), Green Practices (2.9793), and Sustainability Outcomes (3.0742) all reflected low levels of implementation. Correlation and regression analyses indicated weak development relationships reflecting of statistically insignificant relationships (p > 0.05), indicating that leadership appear to be potentially more influential that other observable factors towards sustainability performance. The research provides insight on how little emerged from the interaction data between digital and green purposes, and puts forth additional implications for future research on how cultural and regulative strategies uniquely influence the change-based nature of sustainable corporate performance.
Coastal environments are increasingly threatened by emerging contaminants; however, integrated assessments of multiple pollutant classes remain scarce. This study provides a simultaneous investigation of microplastics (MPs) and total petroleum hydrocarbons (TPHs) in seawater and sediments along the southwestern coast of the Caspian Sea. Samples were collected from fourteen coastal stations during the dry season. MPs were characterized by type, size, shape, color, and polymer composition using microscopy and Raman spectroscopy, while TPHs were quantified using GC-MS. The results revealed MP abundances ranging from 0.05 to 2.35 particles L-1 in water and from 2.01 to 24.13 particles kg-1 in sediment. Fibers and fragments in the 100-500 & micro;m size class were dominant, with polyethylene (PE) and polyethylene terephthalate (PET) identified as the most prevalent polymers. TPH concentrations varied from not detected to 20.152 & micro;g L-1 in water and from 8.37 to 1835.29 & micro;g kg-1 in sediment. Spatial analysis showed no overlap between the contamination hotspots of MPs and TPHs, indicating distinct sources and transport pathways. This concurrent assessment underscores a complex contamination profile in the region and highlights potential ecological risks from both physical and chemical stressors. The distinct spatial patterns provide critical information for designing targeted pollution control measures.
This study investigates the seasonal variability of meteorological feature importance in forecasting Global Horizontal Irradiance (GHI) using machine learning (ML) and deep learning models. High-resolution solar and meteorological data from NREL's NSRDB (24.25 degrees N, 45.34 degrees E, 740 m) were seasonally partitioned into winter, spring, summer, and autumn. Feature selection was conducted using Pearson correlation (threshold 0.25), followed by dimensionality reduction through Principal Component Analysis (PCA). Six ML models XGBoost, LightGBM, Random Forest, SVR, MLP, and LSTM were trained on the processed datasets, and SHAP analysis was used to interpret feature contributions. The results revealed that clear-sky irradiance parameters (GHI, DNI, DHI) consistently dominated GHI prediction (correlation >0.95; SHAP >10(-)& sup1;), while features like temperature and relative humidity varied across seasons. Wind direction, though weakly correlated, showed increased influence in winter. PCA enhanced model stability in spring and winter but slightly reduced accuracy during periods of high irradiance variability. Overall, XGBoost and Random Forest models provided the most accurate and reliable forecasts across seasons.
As patterns of energy usage become more complicated, smart, data-driven strategies are needed for effective forecasting and management. In order to optimise decision support systems in the energy industry, this study suggests a hybrid data mining framework that combines Extreme Gradient Boosting (XGBoost) with K-Means clustering. The model is intended to increase the precision and interpretability of energy usage forecasts while detecting discrete consumption behaviour clusters by utilising the publicly accessible UCI Individual Household Electric Power Consumption dataset. The suggested XGBoost + K-Means model performs noticeably better than conventional models like Linear Regression, Decision Tree, and Random Forest, according to a comparative analysis. It achieves a high R2 score of 0.91, a mean absolute error (MAE) of 39.7 Wh, and aroot mean square error (RMSE) of 49.6 Wh. Furthermore, evaluation criteria including F1-score, precision, and recall attest to the model's resilience and appropriateness for real-time applications. These results demonstrate how hybrid machine learning techniques can convert energy data into useful insights, which will ultimately help develop more intelligent and sustainable energy management plans.
The rapid growth of the photovoltaic device market is imperative for meeting global sustainability goals, such as reducing chlorofluorocarbon emissions to improve air quality, fulfilling the increasing public energy demands, and ultimately lowering the cost of electricity production. The development and application of advanced energy materials are gaining significant attention within the scientific and industrial communities. In this context, recent research has focused on exploring various photophysical mechanisms that can be integrated into photovoltaic devices to achieve conversion efficiencies theoretically surpassing the Shockley-Queisser limit. This limit, which represents the maximum theoretical efficiency for single-junction solar cells, has long been considered a fundamental constraint. However, innovative strategies such as multi-junction architectures, down-conversion and up-conversion layers, hot carrier extraction, and plasmonic enhancements are being investigated to overcome these limitations. It is noteworthy that approximately 55% of incident photon energy is lost, predominantly due to sub-bandgap losses, where photons have insufficient energy to excite electrons across the bandgap, and thermalization losses, where excess photon energy above the bandgap is dissipated as heat. Addressing these intrinsic loss mechanisms is crucial for the development of next-generation photovoltaic technologies capable of delivering higher efficiencies and supporting a sustainable energy future.
