High fidelity analysis of nonlinear flutter in supersonic panels with complex configurations generally requires high-dimensional nonlinear dynamical models, whose direct simulation is prohibitively expensive. To reduce computational cost, this paper develops an equation-driven model reduction framework tailored to nonlinear panel flutter. Because the proposed reduction directly operates on the system matrices and nonlinear coefficient tensors of the full-order models (FOMs), a numerically stable finite element discretization is essential. To this end, a nonlinear Hermite-interpolation-based Bogner–Fox–Schmit element (HBE) is developed. By constructing explicit shape functions, the HBE inherently eliminates the numerical instability that traditional bicubic-interpolation-based Bogner–Fox–Schmit elements typically suffer during mesh refinement. Based on the spectral submanifold (SSM) theory, a master spectral subspace selection strategy incorporating both unstable and coupled stable modes is then formulated. The dimensionless dynamic pressure is embedded into an augmented system as a parameter coordinate, yielding a parameter-dependent mixed-mode SSM. The SSM and its reduced dynamics are constructed directly from the finite element governing equations without using FOM trajectories as training data. This framework reduces finite element models with thousands of degrees of freedom (DOFs) to reduced-order models (ROMs) with only a few coordinates while retaining the dominant nonlinear dynamics. Furthermore, since the dynamic pressure parameter is explicitly embedded in the reduced dynamics, highly efficient simulations under varying dynamic pressure conditions can be achieved via a single offline reduction. Finally, numerical examples demonstrate the computational accuracy and efficiency of the proposed framework, and the effects of various stiffening schemes on the critical dynamic pressure and limit cycle oscillation (LCO) amplitude are investigated.
Continuous high-speed visualization of key intermediate species in turbulent flames is highly desirable yet remains challenging, with such diagnostics historically limited to a few species like OH and CH, primarily in hydrocarbon flames. With society's transition toward a zero-carbon energy future, combustion of carbon-free fuels such as am monia (NH3) and hydrogen (H2) has garnered significant attention. However, robust diagnostic tools for resolving the structures and dynamics of turbulent NH3/H2 flames have been lacking. This work explores the potential for kHz-rate planar laser-induced fluorescence (PLIF) imaging of the NH radical. We present an excitation-detection scheme within the strong NH A3 Pi-X3 Sigma-(0, 0) band near 334 nm, realizing 10 kHz NH imaging, for the first time, with an excellent signal-to-noise ratio (SNR). Furthermore, this approach proves resilient to interferences from elastic Rayleigh and Mie scattering, facilitating measurements in particle-laden environments or enabling simul taneous measurements with particle image velocimetry (PIV)-a crucial combination to explore the fundamentals of NH3/H2 flame dynamics.
Much of the existing research on formation maneuvering control is based on a leader-follower framework, in which one or more leaders track a desired trajectory or velocity, and the rest adapt to it. However, if a leader fails, the desired maneuvering is typically compromised. Alternatively, some studies on formation maneuvering control introduce designed mismatches within a leaderless framework, but these mismatches must be determined offline, and most such methods only achieve local convergence to the desired maneuver. To address these issues, this paper investigates the problem of angle-rigid formation maneuvering control with simultaneous scaling, rotation, and translation in leaderless multi-agent systems. Without requiring offline-designed mismatches for a predetermined maneuver, a formation maneuvering control law is proposed, generating a virtual velocity signal derived from angle constraints within the desired triangles. Combined with a distributed estimation algorithm, the proposed method ensures that the closed-loop multi-agent system globally achieves the desired angle rigid formation maneuvering in a fully distributed manner. Furthermore, to enable implementation in agents’ local coordinate frames, a finite-time orientation alignment algorithm is developed. The effectiveness of the proposed control strategies is validated through three simulation examples.
Magnesium (Mg)- based alloys and their composites have recently gained significant interest in the biomedical field as potential biodegradable materials due to their unique mechanical properties, bioresorbability, biocompatibility, and biological activity. Nevertheless, the widespread biomedical applications of Mg-based alloys and composites are still limited mostly by their superior corrosion rates and subsequent loss in mechanical integrity. In recent years, numerous research studies have been conducted to develop biodegradable Mg-based alloys and magnesium-based metal matrix composites (Mg/MMCs) with enhanced corrosion resistance and mechanical properties. In this paper, an effort has been made to discuss the various methodologies for processing biodegradable Mg/MMCs for important clinical applications. The major processing technologies for biodegradable Mg/MMCs, including liquid-state processing (e.g., stir casting), solid-state processing (e.g., powder metallurgy (PM)), in-situ processing, and modern additive manufacturing (e.g., powder bed fusion (PBF), selective laser melting (SLM), and wire arc additive manufacturing (WAAM)), are first briefly introduced. Subsequently, biodegradable Mg-based alloying designs and the current trend in biodegradable Mg-based alloys, including Mg/Ca, Mg/Zn, Mg/Cu, Mg/Sr, and Mg/RE, essential for biomedical applications, are reviewed in detail. This review article also comprehensively discusses the design of reinforcement materials for producing biodegradable Mg/MMCs for clinical applications. The current trends on biodegradable Mg/MMCs, including calcium phosphate (CaP)-based bioceramics, Si-containing bioceramics, biodegradable magnesium oxide (MgO), and carbon materials reinforced Mg/MMCs, are discussed. Special emphasis has been placed on the production techniques and various behaviors (e.g., mechanical properties, microstructure, biocompatibility, and corrosion behaviors) displayed by micro/nano-sized particles reinforced Mg/MMCs. The potential engineering applications of Mg/MMCs are also introduced. Ultimately, this review highlights the prospects of biodegradable Mg-based alloys and Mg/MMCs biomaterials in various biomedical applications. This review will serve as a valuable reference for young researchers and industry personnel with a comprehensive understanding of biodegradable Mg/MMCs utilized in biomedical applications.
With the rapid growth of online video platforms and the escalating volume of video content, the need for proficient video understanding tools has increased significantly. Given the remarkable capabilities of large language models (LLMs) in language and multimodal tasks, this survey provides a detailed overview of recent advances in video understanding that harness the power of LLMs (Vid-LLMs). The emergent capabilities of Vid-LLMs are surprisingly advanced, particularly their ability for open-ended multi-granularity (abstract, temporal, and spatiotemporal) reasoning combined with common-sense knowledge, suggesting a promising path for future video understanding. We examine the unique characteristics and capabilities of Vid-LLMs, categorizing the approaches into three main types: Video Analyzer x LLM, Video Embedder x LLM, and (Analyzer + Embedder) x LLM. We identify five subtypes based on the functions of LLMs in Vid-LLMs: LLM as Summarizer, LLM as Manager, LLM as Text Decoder, LLM as Regressor, and LLM as Hidden Layer. This survey also presents a comprehensive study of the tasks, datasets, benchmarks, and evaluation methods for Vid-LLMs. Additionally, it explores the extensive applications of Vid-LLMs in various domains, highlighting their remarkable scalability and versatility in real-world video understanding challenges. Additionally, it summarizes the limitations of existing Vid-LLMs and outlines directions for future research. For more information, readers are encouraged to visit the repository at https://github.com/yunlong10/Awesome-LLMs-for-Video-Understanding
