Fujian Petrochemical Group (China)

Abstract In order to adapt to the development of automated dam monitoring systems, artificial intelligence (AI) models based on deep learning are used for dam deformation prediction, effectively overcoming the limitations of traditional statistical analysis methods. These AI models have demonstrated excellent performance in capturing the deformation characteristics and potential dependencies of dams during their long-term service. This study takes a certain rolled concrete gravity dam as an object and combines the dam deformation statistical model with the Informer deep learning model to propose an attention mechanism dam deformation prediction model that considers water pressure, temperature, and aging factors. The comparative analysis of model performance indicates that, compared with commonly used models, such as multiple linear regression, support vector machines, and long short-term memory neural networks, the proposed model can better adapt to the complex nonlinear relationship between dam deformation and influencing factors and has excellent predictive performance and generalization ability, effectively improving the accuracy of dam deformation prediction and providing practical engineering benefits for dam safety monitoring.

This study explores the key factors involved in activating and sustaining student motivation in e-learning environments, viewed through the lens of educational psychology. The research begins by thoroughly analyzing various elements that influence student motivation, including individual differences, learning environment design, and external motivational mechanisms. The paper then presents a series of targeted strategies aimed at igniting students' intrinsic motivation, such as nurturing curiosity, setting achievable learning goals, and providing timely feedback. Additionally, the study emphasizes strategies for maintaining long-term learning motivation, including offering personalized learning support, increasing interactive opportunities, and fostering a positive learning community to enhance the sustainability and engagement of learning. In e-learning environments, student motivation is not only driven by traditional classroom methods but also relies on modern technological tools to increase students' sense of involvement and achievement. By incorporating elements such as gamification, interactive feedback, and social networking, e-learning can effectively spark initial student interest while also sustaining intrinsic motivation throughout the learning process.

Biological foaming is a major problem in activated sludge (AS) wastewater treatment systems. In this study, four wastewater treatment plants (WWTPs) (a total of six AS treatment systems) were investigated. The microscopic examination shows that foaming was mainly caused by gram-positive short branch microorganisms, sludge fragments, and/or other microorganisms, while the long unbranched filamentous was easy to cause bulking. The high throughput sequencing (HTS) and Linear discriminant analysis effect Size (LEfSe) identified the significant discrepancy of bacteria in foams compared to normal AS. Mycobacterium, Mycobacteriaceae, Nocardiaceae, Actinomycetales, Chryseobacterium, Flavobacterium, Ormithobacterium, Flavobacteriaceae, and Portibacter were considered as the dominant foaming-potential bacteria but not the most abundant bacteria in the foams. The excessive growth of foaming bacteria (including Haliscomentbacter, Saprospiraceae, and Tetrasphra) directly led to bulking with a high sludge volume index and was positively correlated with sludge retention time (SRT) and negatively correlated with dissolved oxygen (DO), which means long SRT and low DO may lead bulking instead of foaming. It also found that the foaming bacteria (including Skermania, Comamonadaceae, Cloacibacterium, Flavobacterium, and Chryseobacterium) had significant positive correlations with suspended solids and mixed liquid suspended solids, and negative correlations with temperature and DO concentration.

ABSTRACTTraffic forecasting plays a pivotal role in the advancement of intelligent transportation systems, with significant implications for congestion alleviation and optimal route planning. Existing approaches typically focus on capturing the temporal dynamics of traffic states and the spatial dependencies across road networks to improve prediction accuracy. Nevertheless, two noteworthy limitations persist in these approaches: (1) A lack of consideration for the interaction between spatiotemporal features over varying time scales, which impedes the effective utilization of traffic state information for forecasting future conditions. (2) The inherent stochasticity and distributional imbalances in traffic flow, which introduce uncertainty and contribute to overfitting issues in deep learning models. To address these challenges, we propose a novel method, the heterogeneous‐scale multi‐graph convolution networks based on kernel density estimation (KDE‐HSMGCN). This method integrates two core components: the frequency feature layer and the heterogeneous‐scale spatiotemporal layers. The frequency feature layer employs a mapping network to learn and equalize traffic flow distributions, mitigating the effects of distribution imbalance and overfitting during model training. The heterogeneous‐scale spatiotemporal layers utilize stacked spatiotemporal layers to capture traffic state information across varying time scales. Experimental evaluations on two diverse traffic datasets demonstrate the superior performance of KDE‐HSMGCN in medium and long‐term forecasting scenarios.