暨南大学

The separation of multicomponent signals with crossing instantaneous frequency (IF) curves remains a fundamental challenge in time-frequency analysis. Although the synchrosqueezed wavelet-chirplet transform (SWCT) enhances time-frequency readability by introducing a chirprate variable, its effectiveness is constrained by the underlying assumption of local linear chirp. Consequently, this method does not perform well when analyzing signals characterized by strong frequency modulation. This paper extends the SWCT framework by relaxing the linear chirp assumption. We model signal components as having polynomial phase behavior over short intervals and derive compact expressions for high-order IF and chirprate reassignment operators. The proposed high-order synchrosqueezed wavelet-chirplet transform (HSWCT) enables accurate estimation of both IF and chirprate, and supports robust mode retrieval even with intersecting IF curves. Another key contribution is a rigorous mathematical analysis of the approximation errors of arbitrary-order reassignment operators for IF and chirprate estimation. When the chirprate vanishes, HSWCT simplifies to the traditional high-order synchrosqueezed wavelet transform. To the best of our knowledge, no theoretical analysis exists in the literature on the approximation of arbitrary-order SST IF reassignment operators to the IF. As a by-product of this work, our established theorem provides such an analysis, thereby filling a gap in the theoretical framework of high-order SSTs.

Wireless sensor networks (WSNs), as a fundamental component of the Internet of Things, are often exposed to multiple interacting failure mechanisms, including local functional failures of sensors and relays, intentional external attacks causing propagated failures with global effects (PFGE), and relay-induced random isolation effects on dependent sensors. These characteristics pose significant challenges for reliability analysis. Existing studies mainly rely on probabilistic combinatorial approaches that evaluate WSN reliability from a static perspective. For such WSNs, however, a lifetime-based modeling framework that explicitly captures the temporal competition among different failure mechanisms is still lacking. This paper develops a lifetime-based modeling framework for the reliability analysis of relay-assisted WSNs with random isolation effects and PFGE. The system lifetime is characterized as a function of local failure and external attack arrival times, together with random functional dependence states under an arbitrary sensor-level structure. Based on this formulation, explicit reliability expressions are systematically derived in terms of the underlying arrival-time distributions and the dependence probability function. A Monte Carlo simulation algorithm is further provided for large-scale systems. The framework is also specialized to modular WSNs, yielding simplified reliability expressions and computable bounds. Numerical studies on body sensor networks and air quality monitoring systems demonstrate the applicability of the proposed approach and reveal the impacts of functional dependence and propagated failures on system reliability.

Traditional citation- and keyword-based methods struggle to capture science-technology (S-T) topic linkages owing to data sparsity, citation bias, and substantial linguistic differences across publication and patent corpora. To address these limitations, we propose a novel embeddingbased approach that integrates large language model (LLM) embeddings with the neural embedded topic model (ETM). The approach comprises four steps: (1) encoding a union vocabulary from publication and patent corpora into a unified semantic space via LLM embeddings, (2) extracting science and technology (S&T) topics via ETM within this shared space, (3) quantifying S-T topic linkages via cosine similarity between topic embeddings, and (4) comparing against keyword- and citation-based approaches. When applied to 79,997 3D printing publications and 100,786 patents with a lexical overlap of only 26.43%, the method improves topic coherence via Qwen3 embeddings. The resulting S-T linkages (mean = 0.148) are significantly higher than the keyword-based measures (mean = 0.058), correlating positively with both keyword-based (r = 0.425) and citation-based (r = 0.271) linkages. Expert evaluation confirms that the generated S&T topics are interpretable, complete, informative, and with low bias, while the embeddingbased linkages align with real-world knowledge relatedness. These findings demonstrate that the proposed approach transcends the structural sparsity of citations and the lexical limits of keywords, offering a more robust, interpretable, and comprehensive measurement of S-T topic linkages. It also provides evidence-based tools for key stakeholders in innovation systems to monitor translational research impacts and identify innovation opportunities.

In this paper we study a free boundary problem modeling the vertical growth of a triple-layered tumor cultivated on a flat surface under laboratory conditions. Nutrients are supplied externally via the upper boundary to sustain the tumor cells. During its growth, the tumor develops distinct internal structures comprising one, two, or three cell types: proliferating, quiescent, and necrotic cells. We investigate the existence, uniqueness, and asymptotic stability of stationary solutions to this problem under various conditions. The transformation of the tumor structure during growth is determined by the initial condition and the nutrient level prescribed at the upper boundary. Moreover, by establishing a general theorem for a differential system involving a given periodic function, we also find periodic tumor evolution phenomena arising from the periodic nutrient supply.

Biochar is a major product in pyrolysis and enhancing biochar yield is of significance for improving overall economic efficiency of the process. Herein, a pyrolysis-condensation process was proposed for conversion of peach wood to biochar of enhanced yield at 400-700 degrees C via volatile reflux. This aimed to extend residence time of volatiles for promoting their re-condensation to transfer carbon in volatile organics to biochar. The results indicated that the repeated volatile reflux in pyrolysis-condensation was effective for enhancing biochar yields (up to 58.2% at 400 degrees C, compared to 33.5% in conventional fixed-bed pyrolysis) at the expense of bio-oil production. Prolonged residence time of volatiles in pyrolysis-condensation enhanced re-condensation of volatiles at 400 and 550 degrees C and their cracking/gasification at 700 degrees C. The pyrolysis-condensation thus formed oxygen-rich and hydrogen-rich biochar that exhibited lower structural ordering and thermal stability, lower ignition temperature, and improved overall combustion performance. Furthermore, quantitative Life Cycle Assessment (LCA) confirmed that pyrolysis-condensation resulted in significantly lower environmental impacts compared to conventional pyrolysis. Moreover, pyrolysis-condensation remarkably enhanced the overall carbon yield (up to 34.8%) and nearly doubled the overall energy yield (64.6% versus 32.9%) at 550 degrees C. The in-situ IR analysis confirmed formation of cross-linked aliphatic chain segments in carbon matrix of the biochar from recondensation of aldehyde and ketone intermediates, resulting in enhanced adsorption of phenol via a hydrogen bonding mechanism.