At present, the learning ability satisfaction of university students under the blended learning mode is low, and this has an impact on the effect of teaching and students’ learning experiences. In order to effectively improve learning ability satisfaction, empirical research and a structural equation model were used to explore the influencing factors of university students’ learning ability satisfaction. This study identified the influencing factors of university students’ learning ability satisfaction under the blended learning model, and constructed a structural equation model of these factors based on their relationship with learning ability satisfaction. The model was verified and optimized to determine the weight of factors that affected university students’ learning ability satisfaction. The study showed that the external and internal factors with the greatest effect on university students’ learning ability satisfaction were resource support and learning self-efficacy, respectively. Internal influencing factors had a greater impact on the satisfaction of university students’ learning abilities than external factors. This study will expand the existing research perspective on learning ability satisfaction, provide important references for improving curriculum design and teaching effectiveness, and provide a theoretical basis for solving various practical questions caused by low learning ability satisfaction in teaching practice.

Socially regulated learning (SoRL) is an important way to maintain the sustainable development of collaborative learning (CL). Usually, learners can achieve sustainable and high-quality SoRL with the intervention of teachers. To improve the appropriateness of the intervention, teachers need to clarify the profiles of SoRL to which learners belong, as well as the influence of collaborative learning motivation (CLM) and the relevant background variables. This study used three non-duplicate samples to provide evidence for the psychometric properties of the SoRL and CLM scales through item analysis, exploratory factor analysis (sample 1, n = 531), and confirmatory factor analysis (sample 2, n = 1278). The profiles of SoRL among university students were determined through latent profile analysis (sample 3, n = 909). This study identified three profiles of regulation (strong SoRL, progressive SoRL, and weak SoRL). The analysis of multivariate variance and multiple logistic regression methods further explored the differences in the dimensions of SoRL structures across different profiles and the extent to which CLM and background variables predicted profiles. The results showed that collaborative motivation (CM) and learning motivation (LM) were the predictors of learners’ transformation from a low regulation level to a medium regulation level. CM, LM, altruistic motivation, and major background were the predictors of learners’ transition from the medium regulation level to the high regulation level. Accordingly, teachers can provide learners with an appropriate external intervention to promote the improvement of SoRL. This study contributes to improving learners’ SoRL levels and promoting the sustainable development of education. In the future, the changing characteristics of learners’ SoRL profiles over time will be explored, and the application of learning process data will be strengthened.

Change detection (CD) is essential for Earth observation tasks, as it identifies alterations in specific geographic areas over time. The advancement of deep learning has significantly improved the accuracy of CD. However, encoder–decoder architectures often struggle to effectively capture temporal differences. Encoders may lose critical spatial details, while decoders can introduce bias due to inconsistent receptive fields across layers. To address these limitations, this paper proposes an enhanced symmetric network, termed FDRF (feature difference and receptive field), which incorporates two novel components: the multibranch feature difference extraction (MFDE) module and the adaptive ensemble decision (AED) module. MFDE independently extracts differential features from bitemporal images at each encoder layer, using multiscale fusion to retain image content and improve the quality of feature difference modeling. AED assigns confidence weights to predictions from different decoder layers based on their receptive field sizes and then combines them adaptively to reduce scale-related bias. To validate the effectiveness and robustness of FDRF, experiments were conducted on five public datasets (SYSU, LEVIR-CD, WHU, NJDS, and CLCD), as well as a UAV-based dataset collected from two national coastal nature reserves in Guangxi Beihai, China. The results demonstrate that FDRF consistently outperforms existing methods in accuracy and robustness across diverse scenarios.