Hong Kong University of Science and Technology
Hong Kong University of Science and Technology Institutional RepositoryNot a member yet
162821 research outputs found
Sort by
Symmetrical virtual 48-pulse rectifier with hybrid multimode pulse multiplication for high-current-density hydrogen electrolyzers
No full textThe integration of renewable energy with electrolytic hydrogen production presents a promising pathway for reconciling generation-demand imbalances and reducing carbon-intensive hydrogen costs. However, existing power conversion systems suffer from critical limitations including excessive harmonic distortion, prohibitive output current ripple, and inadequate fault tolerance under high-current operation. This paper proposes a novel virtual 48-pulse rectifier topology incorporating modular passive pulse-multiplication circuits (PPMCs) to address these challenges. The key innovation lies in deploying symmetric PPMC modules at the DC side of a dual 6-pulse rectifier pair, which strategically synthesizes multi-level current waveforms through passive component interactions. This configuration achieves three key merits: i) Unprecedented power quality with input THD <3% and output ripple <1.5%; ii) Cost-effective scalability requiring only few of additional passive components compared to baseline 12-pulse systems; iii) Inherent fault tolerance via parallel redundancy, maintaining functionality during single-module failures. The effectiveness of the proposed rectifier in electrolytic hydrogen production is demonstrated through validation with a 1.2 kW experimental prototype.</p
RLLM-SS: A knowledge-guided simplex search method integrating large language model and reinforcement learning for injection molding quality control
No full textIn injection molding (IM), product quality and process stability are highly dependent on the setting of key process parameters, making efficient parameter tuning essential for achieving reliable and consistent production. However, the tuning process is traditionally guided by expert experience and trial-and-error methods, which often lead to low efficiency and prolonged adjustment cycles. To address this challenge, we propose a knowledge-guided simplex search method that integrates a large language model (LLM) with the Soft Actor–Critic (SAC) reinforcement learning algorithm in a collaborative optimization framework, called RLLM-SS. In RLLM-SS, a quasi-gradient mechanism leverages historical data to dynamically estimate the step size and gradient compensation direction of the simplex search method. These estimated variables, integrated with domain knowledge, are encoded into structured prompts that guide the injection molding quality LLM in dynamically adjusting simplex coefficients through natural language reasoning. This enables the simplex search to overcome fixed-coefficient limitation and avoid local optima to the maximum extent. To mitigate the drawbacks of the LLM, such as its tendency to generate hallucinated outputs and lack of memory of past tuning adjustments, a SAC-based evaluation module is introduced. It assigns rewards based on optimization performance, thereby reinforcing effective strategies and fostering continuous policy improvement when similar conditions recur. Experimental evaluations first verified LLM-SS on standard high-dimensional benchmark functions, confirming its effectiveness in complex search spaces, and were then conducted on an injection molding quality simulation platform built on a neural network trained with practical IM process data. Results show that RLLM-SS outperforms several advanced methods, reducing the average number of iterations by 27.6% and the final Euclidean distance to the target quality curve by 68.3%. It also maintains strong robustness under Gaussian noise perturbations.</p
Effect of statistical uncertainty on kriging interpolation of 2D geospatial data from sparse measurements
No full textGeospatial data are often spatially varying but measured sparsely in a two-dimensional (2D) plane. Therefore, spatial interpolation methods, such as Kriging, are frequently used to estimate values at locations without measurement and quantify the associated uncertainty. However, Kriging generally requires extensive measurements to effectively divide non-stationary geospatial data into a deterministic trend and stationary residuals (i.e., detrending) and to estimate semi-variogram parameters from the detrended residuals (i.e., semi-variogram fitting). When measurements are limited, a scenario often encountered in practice, detrending might be challenging, and the estimated semi-variogram parameters inevitably contain statistical uncertainty. The statistical uncertainty may significantly affect the subsequent Kriging interpolation, but it is often ignored in practical applications of Kriging. This study develops a 2D Kriging method (SR-Kriging) that is featured by a sparse representation of covariance function from a Bayesian perspective and explicitly models both spatial variability and statistical uncertainty for interpolation of 2D geospatial data directly from sparse measurements. The proposed method requires neither detrending nor semi-variogram fitting. Both simulated and real data are used to illustrate and validate the proposed method. Results demonstrate that the proposed method directly interpolates 2D geospatial data from limited measurements, with quantified interpolation uncertainty, and explicitly accounts for statistical uncertainty. Ignorance of statistical uncertainty may lead to an underestimation of Kriging interpolation uncertainty.</p
Bridging the Gap: Hijacking the British Boxer Indemnity Fund to Complete the Guangzhou–Hankou Railway
No full textIn 1926, a bilateral committee of overseas-educated Chinese representatives and established British administrators debated the future use of the British Boxer Indemnity. This article argues that Chinese educated abroad played a decisive role in redirecting the funds, initially designated for education, toward infrastructure development, paving the way for the union of the previously separated halves of the Guangzhou–Hankou Railway. Drawing on archival material—including diaries, committee reports, and diplomatic protocols—the article demonstrates how the bilateral consensus reached by the committee transcended simplified notions of political realism and state self-interest. Instead, shared sociopolitical concerns, coupled with mutual respect, “bridged a gap” between British and Chinese commissioners. The unexpected outcome of the negotiations underscores the agency of young, foreign-educated Chinese professionals who engaged in a new form of technodiplomacy and leveraged their specialized expertise gained abroad to frame infrastructure investments as the keystone of solutions to Chinese problems.</p
Insufficient local methane and nitrous oxide production sustaining water column inventories at the Pearl River Estuary
No full textThe increasing anthropogenic input of organic matter and nutrients has rendered estuaries notable sources of methane and nitrous oxide. However, the processes causing the spatial heterogeneity of dissolved methane and nitrous oxide are complex, involving biogeochemical processes as well as anthropogenic influences. Nevertheless, the combined effects regulating spatial distribution differences remain unclear. In this study, we conducted a spatial survey in an economically vital subtropical tidal estuary. Our findings indicate that sedimentary methanogenesis and nitrous oxide production in both sediment and water columns play a minor role in regulating spatial variations of these gases. Contrastingly, waters near highly urbanized areas exhibited a higher potential for methane and nitrous oxide emissions, and this study is the first to report these concentrations in Hong Kong and Macao waters. The investigation in those regions has enhanced our understanding of estuarine greenhouse gas dynamics. The discrepancy between high concentrations and low production rates, along with certain isotope signatures, underscores the significance of external anthropogenic sources in regulating the distribution of greenhouse gases in estuaries. Our findings suggest that transient processes, such as occasional anthropogenic sources, may play an important role in regulating dissolved inventories.</p
Miniaturized Self-Powered Perovskite Spectrometer
No full textMiniaturized reconstructive spectrometers with small footprint, light weight, and low cost have attracted much attention due to their ability to capture spectral information in scientific research and industrial inspection. However, the current state-of-the-art designs face challenges in the ultralow power consumption and high spectral resolution. For example, it is difficult to maintain high spectral resolution while reducing the number of integrated spectral response units. In this work, we construct a miniaturized self-powered polycrystalline perovskite spectral sensing system based on bandgap-tunable perovskite. A representative device exhibits a peak external quantum efficiency (EQE) of 75%, reflecting the high photoelectric conversion capability of the material system. We achieve high spectral resolution comprising only 8 photodetectors by utilizing reconstruction algorithms and dimensionality reduction methodologies. Furthermore, we demonstrate narrow-band spectral sensing in the 680–800 nm wavelength range with spectral resolutions of ∼5 nm and average peak accuracies of ∼0.85 nm under the light intensity below 10 µW cm−2. The photodetectors operate without an external bias, while only the necessary power consumption for readout circuit and algorithm reconstruction. This work greatly paves the way for the development of low-power and high-resolution miniaturized spectrometers and advances the practical application of spectrometers in hyperspectral imaging.</p
360DVO: Deep Visual Odometry for Monocular 360-Degree Camera
No full textMonocular omnidirectional visual odometry (OVO) systems leverage 360-degree cameras to overcome field-of-view limitations of perspective VO systems. However, existing methods, reliant on handcrafted features or photometric objectives, often lack robustness in challenging scenarios, such as aggressive motion and varying illumination. To address this, we present 360DVO, the first deep learning-based OVO framework. Our approach introduces a distortion-aware spherical feature extractor (DAS-Feat) that adaptively learns distortion-resistant features from 360-degree images. These sparse feature patches are then used to establish constraints for effective pose estimation within a novel omnidirectional differentiable bundle adjustment (ODBA) module. To facilitate evaluation in realistic settings, we also contribute a new real-world OVO benchmark. Extensive experiments on this benchmark and public synthetic datasets (TartanAir V2 and 360VO) demonstrate that 360DVO surpasses state-of-the-art baselines (including 360VO and OpenVSLAM), improving robustness by 50% and accuracy by 37.5%.</p
Spatially resolved molecular signatures of Lewy body dementia
No full textLewy body dementia (LBD), encompassing dementia with Lewy bodies and Parkinson’s disease dementia, is neuropathologically defined by neuronal accumulation of α-synuclein encoded by the SNCA gene. Genetic risk factors strongly influence LBD susceptibility, including SNCA multiplication, particularly triplication, and the apolipoprotein E ε4 allele (APOE4), the strongest common genetic risk factor for LBD. While SNCA is predominantly expressed in neurons and APOE primarily in glial cells, how these genetic factors converge to impact neuronal vulnerability and regional pathology in the human brain remains poorly understood. Here, we applied spatial transcriptomics to postmortem temporal cortex tissue from LBD cases with SNCA triplication or different APOE genotypes, alongside age- and sex-matched controls, to map gene expression within intact cortical architecture. We identified layer 5 of the gray matter as a particularly vulnerable region, characterized by elevated SNCA expression, pronounced synaptic and metabolic dysregulation, and exacerbation of these alterations in APOE4 carriers. Reelin signaling emerged as a core Lewy body-associated pathway disrupted across cortical layers, validated in independent postmortem cohorts and human-induced pluripotent stem cell (iPSC)-derived cortical organoids. In contrast, white matter exhibited distinct molecular alterations, including disrupted myelination pathways, with APOE4 carriers showing increased myelin debris and glial responses compared with non-carriers. Cell-type deconvolution informed by single-nucleus RNA sequencing further revealed APOE4-associated impairments in neuronal vulnerability and intercellular communication. Together, these findings define spatially and cell-type-specific mechanisms through which SNCA dosage and APOE4 genotype impact LBD pathology, providing insight into regionally distinct disease processes and potential targets for genetically stratified therapeutic interventions.</p
Supramolecular Zwitterionic Polymers: Dynamic Traits Imparted by Ionic Interactions
No full textSupramolecular ionic polymers (SIPs) exhibit distinctive dynamic properties, which arise from the synergistic combination of substantial strength and pronounced reversibility of ionic interactions. However, conventional SIPs formed by coassembly using anionic and cationic separated double monomers suffer from complicated synthesis and heterogeneous structures, which obscure in-depth investigation of their dynamic behaviors. In response, we developed a zwitterionic monomer, namely TPE-2N2S, that incorporates both charged motifs into one fluorescent tetraphenylethylene (TPE) skeleton. This zwitterionic and single-component monomer design strategy enables the efficient formation of supramolecular zwitterionic polymers (SZIPs) with a uniform and orderly structure, and it can further enhance the understanding of structure–property relationships. In this perspective, we examine the dynamic features of ionic interactions in SZIPs, emphasizing their structural advantages, controllable assembly, and ionic interaction characteristics. We further investigate the dynamic behavior driven by ionic interactions at the intramolecular, intermolecular, and supramolecular architecture levels. Finally, we discuss key challenges and future opportunities in this emerging field, aiming to advance the design and application of SZIPs.</p
Oxidative Potential of Atmospheric Particulate Matter: A Review of the Role of Metal−Organic Interactions, Mechanistic Insights, and Key Determinants
No full textOxidative stress, resulting from antioxidant depletion or excessive reactive oxygen species (ROS) production, is a key mechanism linking ambient particulate matter (PM) exposure to adverse health effects. The oxidative potential (OP) of PM, a measure of inhaled PM’s capacity to deplete antioxidants or generate ROS, is largely driven by transition metals (TMs) such as iron and copper. However, coexisting organic matter also modulates OP, both directly through its own redox activity and indirectly via interactions with TMs that alter redox cycling. This review synthesizes current understanding of TM−organic interactions and their influence on the OP of PM, as assessed by acellular assays. We discuss mechanistic insights, key determinants, and the complexity of these interactions. The importance of considering TM−organic interactions in evaluating aggregate OP from individual components and apportioning OP to specific chemical species is highlighted, with implications for mechanistic studies and health risk assessment.</p