海口经济学院

The omnipresence of global forces and rapid technological advancements is shaping the behavior of SMEs as they increasingly adopt intelligent manufacturing to achieve sustainable goals. Despite their efforts, manufacturing SMEs in China are facing complex challenges while striving to maintain sustainable performance and promote green, ambidextrous innovation. These stumbling blocks create further challenges due to resource paucity and other related barriers. Nevertheless, an academic exploration of the factors that influence green ambidextrous innovation is warranted, mainly through the prism of Chinese manufacturing SMEs. Therefore, the present study is about how SMEs can benefit from green knowledge capabilities in balancing the exploration and exploitation of ambidextrous green innovation. The analysis is further augmented by incorporating green entrepreneurial intention as a mediator and digital financial literacy as a moderator. The data was obtained from 211 manufacturing SMEs in China through an email survey and analyzed using PLS-SEM. Findings reveal that green knowledge capabilities improve the green entrepreneurial intentions of SMEs to pursue ambidextrous green innovation. Results also confirm the role of digital financial literacy in enhancing green entrepreneurial intentions and fostering ambidextrous green innovation. The relationship is stronger for firms with high digital financial literacy and weaker for those with low digital financial literacy. Findings confirm that green entrepreneurial intentions serve as a prerequisite for firms to adopt environmental management plans, reflecting the firm’s commitment to green initiatives, proactiveness, and risk-taking.

Assessing landscape ecological risk (LER) is vital for sustainability in ecologically fragile regions. However, LER in cultural heritage sites remains insufficiently explored. Using Guyi Basin as a case study, this study analyzes LER dynamics under land-use change and its driving factors. The PLUS model was applied to simulate land-use changes from 1990 to 2060 under three scenarios, while GeoDetector was used to identify key drivers of LER. Results show a marked expansion of construction land in urban areas and traditional villages and a continuous decline of cultivated land. LER exhibits an “internally high–externally low” pattern characterized by “high–high” and “low–low” clusters. The ecological protection scenario presents a significantly lower risk Interactions such as DEM∩population density exerted the strongest combined effect on LER, with natural factors showing greater influence than social factors. This study provides methodological insights and practical implications for LER management and sustainable development in cultural heritage sites.

Frequent oil spills pose severe environmental and ecological threats, necessitating the development of materials capable of effective all-weather remediation of high-viscosity crude oil. In this study, a self-assembly strategy was employed to integrate covalent organic framework (COF) and polypyrrole (PPy) on a flexible cotton fabric (CF). Subsequent hydrophobic modification with polydimethylsiloxane (PDMS) produced a multifunctional composite membrane (PDMS@PPy@COF@CF) that integrated photothermal conversion, electrothermal response, and superhydrophobicity. The membrane exhibited excellent superhydrophobicity (WCA = 153.5°), superior oil-water separation efficiency (≥98.5%), and a high separation flux (≥24 010.4 L m–2 h–1). Notably, the construction of micro/nanostructures via π-π stacking and hydrogen bonding between PPy and COF layers endowed the material with outstanding photothermal and electrothermal conversion performances. Specifically, the surface temperature rapidly increased to 73.8 °C within 180 s under 1.0 kW m–2 irradiation and attained 145.9 °C at 10 V. Synergistic utilization of photothermal (0.5 kW m–2) and electrothermal (4 V) effects enabled rapid crude oil adsorption within merely 45 s. Furthermore, the composite membrane maintained stable conductivity and superhydrophobicity under mechanical strain, ensuring its reliability and durability for practical applications. In summary, the multifunctional PDMS@PPy@COF@CF membrane enables energy-efficient all-weather cleanup of high-viscosity crude oil spills with practical oil recovery potential.

Aiming at the problems of uneven illumination, strong interference from suspended solids, easy missed detection of small bleached patches, and insufficient feature extraction of models in underwater coral detection scenarios, this paper proposes a spatial attention enhanced underwater coral image detection method based on YOLOv8. Taking YOLOv8 as the benchmark framework, PSA polarization attention and CA coordinate attention are introduced to construct a dual-path attention enhancement module to strengthen the spatial features and channel correlation features of targets. The Mamba structured state-space model is embedded to improve the modeling ability of long-range feature dependencies and global context perception. Lightweight design is realized through channel pruning and model structure optimization to balance detection accuracy and inference speed. Experimental results on a self-built underwater coral dataset show that the mAP@0.5 of the proposed method reaches 0.763, which is 9.2 percentage points higher than the YOLOv8+Mamba baseline. The detection accuracy and robustness are significantly improved, which can meet the demand for rapid detection of coral health status in complex underwater environments.