
西交利物浦大学(Xi’an Jiaotong-liverpool University)位于江苏省苏州市,是经中国教育部批准、由西安交通大学和英国利物浦大学合作创立的,拥有中华人民共和国学士学位和英国利物浦大学学位授予权的中外合作大学,承担国家“863 计划”科研课题的成员单位。 2017年,学校成为江苏省省级硕士立项建设单位。 2004年9月,西安交通大学与利物浦大学签订协议合作成立西交利物浦(国际)大学;2006年学校正式成立。2010年获准授予利物浦大学研究生学位,2012年教育部同意学校实施英国利物浦大学硕士和博士学位教育。 据2016年9月学校官网显示,学校分为南北两个校区,共610亩,建筑面积40.7万平方米,拥有50余万册图书;截至2020年5月,西交利物浦大学约有17000多名本科生和研究生(其中约3500名在利物浦大学学习)、来自50多个国家和地区的920余名教研人员 。据2020年7月官网显示,学校开设43个本科专业,44个硕士专业 和17个博士专业。
Existing studies on battery thermal management mainly focus on steady operating conditions, whereas dynamic thermal regulation under varying loads remains insufficiently explored. In this study, a hybrid battery thermal management system combining phase change material and liquidcooling channels is developed to enhance thermal buffering and convective heat dissipation. A surrogate-assisted multi-objective optimization framework is established to optimize the composite cold plate by considering maximum temperature, temperature difference, and pressure drop. The optimized design reduces maximum temperature by 1.22 degrees C, temperature difference by 0.45 degrees C, and pressure drop by 36.38%. To support real-time regulation under dynamic conditions, a multi-task deep learning model integrating a convolutional neural network, a long short-term memory network, and an attention mechanism is constructed to predict transient temperature responses under different discharge loads and cooling conditions. Using historical electro-thermal state information, the model captures temperature evolution and achieves an average coefficient of determination (R2) of 0.967 and a root mean square error of 0.173 degrees C. Based on the predicted thermal states, a deep-learning-informed rule-based control strategy is developed to regulate cooling. Compared with constant-flow-rate cooling, it reduces pumping energy consumption by 10.76% and decreases battery-pack temperature difference by 0.3 degrees C. Overall, the proposed framework integrates cold-plate optimization, transient temperature prediction, and predictive thermal control for intelligent hybrid battery thermal management.
Despite the success of Multimodal Sentiment Analysis in ideal settings, real-world data corruption often leads to “variance explosion” in continuous latent spaces and “semantic drift” during teacher–student alignment. To overcome these challenges, we introduce Quantized and Distilled Robust Fusion (QDRF), a framework that replaces fragile continuous imputation with a Finite Scalar Quantization bottleneck to improve robustness against latent-space perturbations through a discrete semantic pivot. We further stabilize the system using an asymmetric Exponential Moving Average distillation framework that enforces dual alignment at both the feature and weight levels. Evaluated on CMU-MOSI, CMU-MOSEI, and CH-SIMS, QDRF achieves state-of-the-art or highly competitive performance on CMU-MOSEI and CH-SIMS, demonstrating that linguistic-prior-guided discrete representation learning is effective for robust sentiment prediction in incomplete data environments.
Community gardens are increasingly recognised as important instruments of urban micro-regeneration, yet their long-term sustainability remains fragile under project-based governance. This study traces the five-year evolution (2021–2025) of maintenance barriers in a government-initiated community garden in Suzhou, China. Using a longitudinal mixed-methods design, this paper tracks the community garden across three stages (inception, growth and struggle) through three phases of Likert-scale surveys and semi-structured interviews, supplemented by Participatory Action Research (PAR). The findings reveal a marked divergence in barrier trajectories. Individual technical constraints declined as residents accumulated horticultural knowledge, with the mean score for ‘lack of gardening knowledge and skills’ falling from 4.00 in 2021 to 1.83 in 2025. By contrast, team and institutional barriers intensified: ‘lack of an internal leader’ rose from 4.07 to 4.83, ‘unclear garden tasks’ from 3.20 to 4.33, and ‘Inadequate support from local authorities’ from 3.53 to 4.83, reaching the ceiling value of 5.0 for several individual respondents in 2025. The withdrawal of external facilitators exposed a governance vacuum that residents themselves labelled the ‘Three-Ignores’ (san-bu-guan) condition, in which fragmented responsibilities among Residents’ Committees, Property Management Companies and Homeowners’ Committees, together with unfulfilled budgetary commitments, produced systemic deterioration. At the same time, a significant demographic shift towards a volunteer team composed exclusively of elderly residents weakened the garden’s collective resilience. Sustaining community gardens requires a shift in urban management from temporary ‘project logic’ to more durable ‘governance logic’, including formal exit strategies, clearer responsibility arrangements and the incorporation of community assets into established administrative and budgetary systems.
Solar-driven photocatalytic synthesis of hydrogen peroxide (H2O2) represents a highly promising green chemistry route, yet its efficiency is severely constrained by the rapid recombination of photogenerated carriers in traditional photocatalysts. In this study, a Cu doped CdSe-diethylenetriamine/benzoxazinebased 3-aminophenol-formaldehyde (Cu-CdSe-D/APF) photocatalyst integrating strain modulation effects and S-scheme heterojunction characteristics was successfully constructed via a Cu2+ modification strategy. Photocatalytic performance tests revealed that the as-synthesized 0.7%Cu-CdSe-D/APF exhibited superior H2O2 evolution activity under visible light irradiation (lambda >= 420 nm), achieving a yield of up to 5324.2 mu mol g(-1) h(-1), which is 5.3 times higher than that of pristine CdSe-D. Combined analyses using geometric phase analysis, high-resolution transmission electron microscopy, and density functional theory calculations confirmed that the introduction of Cu2+ induces a non-uniform compressive strain field within the 0.7%Cu-CdSe-D/APF. This strain field acts synergistically with the built-in electric field at the S-scheme heterojunction interface, which not only significantly suppresses the recombination of photo-generated electron-hole pairs but also accelerates the directional migration of carriers and the kinetics of surface catalytic reactions. This study provides a new paradigm for designing highly efficient photocatalysts for H2O2 synthesis through a "strain engineering-heterojunction construction" synergistic strategy. (c) 2026 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
The manipulation of articulated objects for part-level motion is crucial due to their prevalence in real-world applications. Although current manipulation methods have improved interaction quality, they share a common issue: neglecting the completeness of motion trajectories. For example, when we use a front-loading drum washing machine, the expected action is to manipulate the door from fully closed to fully open. However, these methods might only result in it being half-open. To tackle this limitation, we introduce a novel framework for optimizing motion trajectories based on multimodal fusion. Specifically, we explicitly model trajectory completeness and propose a motion trajectory construction paradigm (MTCP). This paradigm is applied to a large-scale dataset containing a wide range of articulated objects, generating high-quality motion trajectories for multimodal fusion. Furthermore, to handle trajectory homogeneity, we propose a trajectory enhancement policy (TEP) that enriches the trajectory set by capturing the multimodal distribution of feasible trajectories. Subsequently, to enhance the learning efficiency and task adaptability of Multimodal Large Language Models (MLLMs), we propose a learning strategy for 3D perception inspired by 2D perception (3PI2P), complemented by a progressive reasoning approach. This strategy integrates a dual-branch input design using RGB images and depth maps, combined with six forms of visual question answering tasks, to achieve collaborative reasoning and deep fusion of 2D semantics and 3D geometric information. The robustness and generalizability of the framework are demonstrated through evaluations in both simulation and real-world environments.