Siliwangi University

This study aims to develop and test an integrative model of the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB) to explain farmers' intentions to adopt artificial intelligence (AI) in the agricultural sector. The model positions Perceived Ease of Use (PEOU) as an initial determinant influencing Perceived Usefulness (PU), Attitude Toward AI (ATT), and Behavioral Intention to Use AI (BI), both directly and through mediating mechanisms. The study employed a quantitative explanatory design with a cross-sectional survey approach. Data were collected from 203 respondents, consisting of farmers and MSMEs and small-to-medium-scale agricultural businesses in the Greater Bandung area, from January to March 2024. The research instrument was an online questionnaire with a five-point Likert scale, and data analysis was conducted using covariance-based Structural Equation Modeling (SEM). The results showed that PEOU had a positive and significant effect on PU and ATT. Furthermore, PU, ATT, Perceived Behavioral Control, and Perceived Social Norms were shown to increase the intention to use AI in agricultural activities. Mediation tests confirmed that PU and ATT act as partial mediators in the relationship between ease of use and AI adoption intention. These findings emphasize that AI adoption in the agricultural sector is a gradual process that simultaneously involves cognitive, affective, and social factors. Theoretically, this study strengthens the relevance of TAM-TPB integration in the context of modern agriculture, while practically providing a basis for formulating strategies to accelerate sustainable digital transformation in agriculture. (c) 2026 by the authors; licensee Growing Science, Canada

Greenhouse gas emissions significantly contribute to global warming, highlighting the need for effective and sustainable carbon dioxide (CO2) capture technologies. This study introduces a hybrid carbon capture and storage (CCS) prototype that integrates mist-assisted absorption with coconut fiber adsorption as a low-cost and renewable approach. The system was evaluated in terms of CO2 capture efficiency and energy consumption. The prototype consists of four chambers: mist generation units to enhance gas-liquid contact and coconut fiber packing to adsorb residual CO2. Six experimental modes were tested at two gas flow velocities (0.9 and 1.4 m/s). Real-time CO2 monitoring was performed using NDIR MH-Z16 sensors validated against a reference probe meter, with data acquisition through the NI MyRIO platform. Results showed that capture efficiency increased from 14% to 19% in the baseline condition to a maximum of 65.9% in the hybrid configuration combining six mist modules and coconut fiber. Energy consumption analysis revealed that the energy consumption index (ECI) ranged from 72.95 to 79.85 kWh/kg CO2, with the lowest ECI obtained in the hybrid setup, although mist generation contributed significantly to power demand. Despite higher energy use compared with advanced chemical or vacuum swing adsorption systems, the prototype offers advantages in terms of simplicity, operational practicality, and use of renewable local resources. These findings demonstrate the potential of coconut fiber-based hybrid CCS systems as complementary solutions for decentralized and small-scale applications.

Efficient inspection and maintenance of solar panels is essential to ensure the optimal performance of renewable energy systems. This research proposes a modified Visual Geometry Group 19 (VGG19) deep learning model for inspection of solar panel faults, emphasizing both classification and interpretability. The modifications include the addition of Global Average Pooling two dimensions (2D) and dropout layers to improve generalization when identifying five types of damage across a dataset of 885 solar panel images, complemented with the implementation of Gradient-weighted Class Activation Mapping (Grad-CAM) to strengthen the model's interpretability by visualizing important areas in the images that influence classification decisions. The adapted VGG19 model demonstrates strong performance, maintaining high specificity across all classes (above 96%), indicating robust generalization. Compared to the standard VGG19, the modified model achieves an 8.47% improvement in accuracy and a reduction in loss from 0.81 to 0.45. Grad-CAM visualization results help highlight critical regions associated with each defect, providing more clarity for field technicians to validate classification results. The research conducted not only provides specific classification solutions but also enables visual interpretation that supports increased confidence in the development of more reliable solar panel inspection applications.

This study investigated the performance optimization of photovoltaic system (PV) systems installed on an ellipsoidal rooftop at the Politeknik Balikpapan Building, addressing the unique challenges posed by curved surfaces, such as uneven irradiance distribution and shading effects. This research aims to evaluate and compare the performance of monocrystalline and polycrystalline PV modules using Helioscope software, with a focus on two key metrics: Performance Ratio (PR) and Energy Density Ratio (EDR). Simulations were conducted by segmenting the rooftop into distinct areas based on azimuth angles, allowing a detailed analysis of the shading impact. The results showed that monocrystalline PV modules achieved a PR of 71.5

Developing markets like Indonesia still use natural gas, oil and coal as the main sources of electricity. These traditional energy sources will run out soon. Nowadays, Indonesia needs to pursue renewable energy while evaluating the environmental impact of carbon emissions from electricity generation and the use of oil, gas, and coal. This paper aims to investigate the relationship between the three main energy sources of electricity and renewable electricity to carbon emissions from Indonesia. Data series from 1990 to 2020, compiled by the International Energy Agency (IEA), are identified using the Partial Adjuster Model (PAM) with Robust Least Squares (RLS). Empirical results found that all major energy sources have a positive and significant effect on carbon emissions in both the short and long run. Specifically, electricity from coal has the most dominant impact, while gas contributes the least to carbon emissions. In both the short and long term, renewable electricity has been shown to reduce carbon emissions in Indonesia, albeit relatively modestly. In addition to the main findings, this paper implicitly reveals a strong structural dependence on coal-based electricity in Indonesia, which limits the effectiveness of efforts to reduce emissions despite the development and expansion of renewable energy. The findings indicate that the limited impact of renewable energy is due to limitations in institutional readiness, investment scale, and technological capacity. This paper also contributes to the literature by integrating disaggregated electricity sources (gas, oil, and coal) into the dynamic PAM-RLS framework, thereby offering a more nuanced and comprehensive perspective than conventional aggregate approaches. Notably, it provides policy implications by emphasising the urgency of accelerating renewable energy development to achieve environmental sustainability and inclusive economic growth. Other implications of the current study are discussed in the future.