Carbonate sand is prone to substantial deformation and particle breakage in marine and geotechnical engineering. Specifically, larger particles are generally more susceptible to crushing. Nevertheless, it remains unclear how these responses evolve as the loading stress path rotates from 0° to 90°. Furthermore, it becomes more complex when the specimens are prepared with different particle sizes or subjected to initial shear stress states. Addressing the above issues is experimentally challenging as the conditions mentioned above are complex, but studying the above issues is necessary as such conditions exist in the field. In this study, we investigated the coupled effects of particle size, loading stress path, and initial shear stress on the shear and breakage behaviors of carbonate sands (CS) through stress-controlled drained triaxial tests. For a given loading stress path, both deformation and particle breakage increase with particle size. Meanwhile, for a given particle size, the deformation mode transitions from a contraction-dominated mode to a shear-dominated mode as the loading stress path rotates counterclockwise from 0° to 90°. The particle breakage index remains nearly constant for loading path angles below 90° under isotropic conditions, and increases markedly at 90°. Under anisotropic consolidation, the particle breakage increases obviously from a loading path angle of 71.6° to 90°. Despite these distinct deformation and breakage patterns, a unique empirical equation was proposed to describe the relationship between particle breakage and soil deformation. These findings advance the understanding of breakage and deformation behaviors under different loading stress paths and provide practical guidance for the design of marine infrastructure on carbonate sand foundations.
Unknown multisource disturbances always exist in nonlinear systems, especially in practical processes, which can enhance the adverse effect on the accuracy and stability of tracking control performance. To deal with this issue, a self-organizing fuzzy tracking control (SFTC) is developed to improve robustness. First, a self-organizing fuzzy neural network (SOFNN), based on the static and dynamic characteristics of neurons, is designed to approximate unknown terms of nonlinear systems. Then, the proposed model can comprehensively describe the system characteristics to assist in the tracking control. Second, a dynamic surface control (DSC) is integrated to avoid the differential explosion problem in traditional backstepping control, enabling high precision and fast trajectory tracking. Third, a Lyapunov function with tracking error, filtering error and estimation error (i.e., weight, center, and width of SOFNN parameters) is constructed to rigorously prove the stability of SFTC. It can be concluded from the theoretical proof that all closed-loop signals are semi-globally uniformly ultimately bounded (SG-UUB), and the tracking error converges to a small bounded region around zero. Finally, several simulation experiments and a real application of SFTC are conducted to validate its feasibility and effectiveness.
Object detection in remote sensing images (RSIs) is a fundamental and valuable task in the field of Earth observation and computer vision. Typically, RSIs are acquired from a bird’s-eye view, leading to inherent characteristics such as complex backgrounds, random and dense distribution of small objects, and large-scale variations. These characteristics significantly constrain the performance of existing object detectors. To address these issues, this paper proposes a lightweight yet effective collaborative context state space model, C2Mamba, which integrates global context and multi-scale local receptive field information, thereby enabling higher accuracy for geospatial object detection. First, a collaborative multi-context Mamba is developed to capture diverse context information, including local, global, and channel-wise features, in a hierarchical interactive manner. This enables the generation of a comprehensive scene description that adaptively focuses on geospatial objects while suppressing interference from complex backgrounds. Second, a Gaussian large-receptive field perception module is designed to progressively aggregate context information and fine-grained details via increasingly larger receptive fields. This process is enhanced by a Gaussian modulation mechanism, which improves sensitivity to subtle yet discriminative features, thereby effectively addressing scale variations of remote sensing objects, especially for small ones. Experimental results on three public datasets (RSOD, NWPU VHR-10.v2, and DIOR) demonstrate that C2Mamba outperforms state-of-the-art methods in both detection accuracy and computational efficiency.
Ultra-thin chips are designed to address the limitations of conventional chips in size scaling, power consumption, and thermal dissipation. Currently, they hold tremendous application potential in fields such as biomedical, artificial intelligence computing, and flexible electronics. However, chips thinner than 50 µm are typically manufactured using thinning or grinding processes, which inevitably introduce severe mismatch stresses and cause significant warpage, thereby limiting device performance and reliability. Moreover, the deformation driven by internal stresses such as misfit stresses or lattice mismatch in functional layers becomes significantly enhanced. This paper focuses on the ultra-thin chip structure, systematically investigating the evolution and suppression strategies of mismatch stresses through experimental measurements and theoretical analysis. White light interferometry was employed to measure warpage profiles and non-uniform curvatures of various ultra-thin chips. In terms of theoretical analysis, a multilayer thin film-substrate system mechanical model was established for mismatch stresses inversion. Furthermore, a patterned backside metallization design provides an effective method for mismatch stress control and chip planarization.
The incorporation of in-situ generated reinforcements into titanium alloys fabricated via selective laser melting (SLM) has been demonstrated to significantly enhance mechanical strength while simultaneously reducing ductility. Post-heat treatment is commonly employed to optimize microstructural characteristics and improve mechanical performance. In this work, the influence of annealing temperature on the microstructure and mechanical properties of SLMed 0.85 vol. [1̅011]_α,//[11̅3̅]_β, (101̅1̅)_α,//(201)_β . Variations in boron (B) atom diffusion coefficients, contents of β phase at different temperatures, and merging of TiBw led to a non-monotonic change in the aspect ratio of TiB whiskers, which initially increased and then decreased with rising temperature. Grain coarsening and reduced dislocation density were identified as the primary mechanisms underlying the reduction in strength, resulting in a progressive decrease in yield strength from 1231 MPa at 750°C to 939 MPa at 1000°C. In contrast, the increase in β-phase content, equiaxed grains, and enhanced interfacial bonding between the reinforcement and the matrix collectively contributed to improved ductility, with elongation increasing steadily from 7.8
The hydrodynamic performance of underwater vehicles is significantly constrained by skin friction drag, which constitutes the predominant component of total resistance. Inspired by the imbricated array architecture of tuna skin scales and the exceptional surface characteristics of zeolitic imidazolate framework-8 (ZIF-8), this study fabricates bio-inspired surfaces with gradient inclination angles via stereolithography 3D printing, followed by surface functionalization using an optimized spray-coating process. Systematic characterization confirms that the resulting modified biomimetic surface (MBS) 30° exhibits a water contact angle of 152.9°, indicating stable superhydrophobicity, and achieves an outstanding drag reduction rate of 92.1
A long short-term memory (LSTM) neural network is integrated into a data-driven adaptive sliding mode control (SMC) strategy to enhance the tracking accuracy and robustness of the system. The designed controller only employs the input and output data without requiring the mathematical model or structural information of the system. Hence, the proposed method is data-driven. Firstly, based on the dynamic linearization method of the model-free adaptive control, a novel controller of multi-input multi-output (MIMO) nonlinear discrete-time systems is designed by combining with the SMC. Then, an LSTM network updated scheme is proposed via the backpropagation algorithm, which achieves real-time adjustment of the switching gain matrix, enabling more accurate execution of the control strategy. Subsequently, the convergence analysis of the adaptive SMC approach is provided through rigorous mathematical proofs, and the numerical simulation demonstrates the effectiveness of the proposed method. Finally, we discuss the existing challenges.
Aquatic organisms have evolved sophisticated hair-like sensory systems capable of detecting hydrodynamic flows with remarkable sensitivity and directionality. However, existing artificial hair-like flow sensors still face the challenges of directional discrimination and broadband sensitivity in underwater hydrodynamic perception. This study proposes a multi-mode coupled hair-like sensor (MMCHS) to overcome these limitations by mimicking the functional coordination found in seal whisker arrays. The MMCHS integrates four sensor units of distinct resonant frequencies into a centrally symmetric spiral-beam base. This architecture enables the fusion of their discrete resonant peaks into a collaborative broadband response, while the parameter-customized spiral beams introduce intrinsic structural anisotropy, thereby providing a physical mechanism for flow-direction identification. Through dynamic modeling and finite element analysis, the working principle enabling broadband perception and directional discrimination is characterized. Prototypes were fabricated and experimentally characterized under controlled hydrodynamic stimuli. Results demonstrate that the MMCHS exhibits a broad high-sensitivity frequency band, a linear response over flow velocities from 1 to 100 mm/s, and robust operation even under strong background noise (signal-to-noise ratio, SNR = −20 dB). When coupled with a basic classifier, the sensor achieves over 99
Photovoltaics (PV) represent the most widely deployed solar energy technology and a cornerstone of sustainable power systems. However, conventional PV technologies still struggle with low solar-to-energy efficiency, thermally induced performance degradation, and carbon footprints throughout their life cycles. To address these challenges, this work presents the proposed Carbon-Swallow Photovoltaic Module, an integrated architecture comprising a PV layer, a thermo-carbon adjustment layer, and a microalgae carbon-fixation layer. This tri-layered design enables the simultaneous production of solar electricity, CO2 sequestration, and sustainable high-value biomass. Notably, the thermo-carbon adjustment layer serves as a multi-functional interface that passively enhances energy and mass transfer within the system. By regulating vapor and CO2 fluxes during diurnal and nocturnal cycles, the system achieves synergistic PV cooling and intensified microalgal photosynthesis. Outdoor experimental results on a 2 m2 of test rig demonstrate that the system increased CO2 replenishment by 18.5
To mitigate greenhouse gas emissions from fossil fuel combustion, the transition to green energy is critical for the transportation sector. Ammonia (NH3) has attracted significant attention as a zero-carbon fuel due to its high energy density and carbon-free combustion. However, its practical application is hindered by ignition difficulties and combustion instabilities. Hydrogen (H2) blending has been proposed as an effective strategy to overcome these limitations. Yet, hydrogen enrichment also introduces irregular combustion phenomena, such as knocking and pre-ignition, which constrain engine performance. In this study, fundamental optical diagnostics combined with transient synchronous measurements were employed to investigate the mechanisms governing these irregular combustion behaviors. Results show that pure ammonia combustion under spark ignition is dominated by regular flame propagation, whereas hydrogen blending promotes irregular combustion. When the hydrogen blending ratio reaches ∼20
Flexible electric heaters require materials with precise temperature control and rapid thermal response capabilities. In this work, we systematically investigate the influence of material dimensionality and nanoparticle deposition density on electrothermal conversion in silver nanomaterials. We found that variations in the contact modes among zero-dimensional (0D) silver nanoparticles significantly affect their electrothermal performance. Specifically, a lower deposition density results in relatively slender sintering necks, which enhance electron-phonon interactions and localized Joule heating effects, thus improving the electrothermal conversion efficiency. In contrast, a higher deposition density leads to more robust and thicker sintering necks, which reduces efficiency. Extending the study to one-dimensional (1D) silver nanowires, quasi-two-dimensional (quasi-2D) silver flakes, and three-dimensional (3D) silver films, comparative analyses reveal a clear dimension-dependent hierarchy, with the order of electrothermal conversion efficiency being 0D > 1D > quasi-2D > 3D. As a demonstration, the fabricated Ag NPs/PI heater (Rs ∼3 Ω/sq, 2 cm × 2 cm) exhibits superior performance, with a rapid thermal response (∼250°C within 10 s), stable low-voltage operation (< 5 V), and high-temperature stability (∼250°C). This dimensional comparison reveals fundamental correlations between nanomaterial structure and electrothermal performance, establishing design principles for next-generation flexible heating systems.
Very low frequency (VLF) communications have demonstrated significant application potential for cross-medium applications, including underground/underwater propagation, emergency communication, and navigation systems, owing to their superior penetration capability and low transmission loss in different media. However, the bulky size and short-range transmission remain huge obstacles to practical implementation. Addressing this issue may require high input power, radiation efficiency, and reception sensitivity. Therefore, a co-designed metamaterial transmitter and receiver scheme is proposed for miniature long-distance VLF communication. These metamaterial antennas, employing the synthetic optimization of isolation transformer, coupling resonance, and superposition amplification techniques, are co-designed for high-efficiency radiation and high reception sensitivity. Theoretical derivation, simulations, and experiments reveal the performance enhancement in input power, transmission efficiency, and transmission distance. Through these synergistic improvements in transmitting and receiving, the proposed VLF communication system enables reliable kilometer-level transmission and 48 m air-ground cross-medium transmission while maintaining a compact antenna size within λ/10000. Our work provides a universal optimization method for VLF antenna design, paving the way for applications in portable VLF devices, cross-medium communication, and indoor navigation.
Stroke often results in upper and lower limb motor dysfunction, leading to prolonged rehabilitation training and assessment. Existing rehabilitation assessment methods rely on manual clinical scales, which have several limitations, such as strong subjectivity, partiality of the assessment, reliance on specialist physicians, and difficulty in capturing subtle changes. To address these challenges, we developed a fine-grained, comprehensive, quantitative automatic rehabilitation assessment system based on the depth camera for patients with upper and lower limb motor dysfunction. A joint motion capture platform is designed to collect visual temporal data of joint movements by using YOLOv8-Pose and the depth camera. We then proposed the trajectory-ROM-smooth (TRS) scoring algorithm based on joint motion data, which comprehensively and interpretably assesses patients’ rehabilitation level from the perspectives of joint tracking, flexion-extension, and muscle control ability. In the evaluation experiment of 21 stroke patients, the proposed system shows results highly consistent with the manual assessments of experienced physicians. More importantly, the TRS assessment results are more fine-grained and are able to sensitively capture subtle improvements of joints during the rehabilitation process. In conclusion, this work provides a fine-grained, comprehensive, and interpretable quantitative assessment for patients with motor dysfunction.
Precise monitoring of ambient humidity is critical for human health, environmental regulation, and a wide range of industrial processes. Here, we report a high-performance capacitive humidity sensor based on poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) complex, in which trifluoroacetic acid (TFA) functions as a dual-role modulator. TFA enhances baseline electronic conductivity by disrupting the insulating PSS encapsulation, thereby optimizing the PEDOT structure, while simultaneously introducing latent charge-carrying ions (H+ and CF3COO−) into the PEDOT matrix. Upon exposure to moisture, these ions become hydrated and mobile, resulting in a pronounced increase in ionic conductivity. This humidity-gated surge in mobile ions enables the formation of a prominent electric double layer (EDL) at the electrode interface, generating a large and measurable capacitive response. The resulting sensor exhibits a broad detection range (7
Bionic compound eyes have emerged as a promising solution for high-precision and real-time imaging in complex scenarios, yet existing designs face challenges including poor imaging quality, low recognition efficiency, and difficulties in target tracking. To address these issues, this study proposes an ellipsoidal-ommatidia negative-meniscus-substrate compound eye (EONMSCE) tailored for planar optoelectronic detection. The EONMSCE features a honeycomb full-fill arrangement of ellipsoidal ommatidia, which optimizes imaging uniformity and enhances the brightness of edge ommatidia. High-precision fabrication of the EONMSCE was achieved via two-photon polymerization (TPP) lateral additive manufacturing. Optical performance validation using spot diagrams and modulation transfer function (MTF) analysis confirmed that the EONMSCE meets preset imaging requirements, with a measured field of view (FOV) of 82° that aligns well with the theoretical value of 82.2°. Compared to traditional spherical compound eyes, the EONMSCE increases edge ommatidia imaging brightness by up to 48
Lunar rover mobility inevitably induces dust lifting, posing risks to onboard instruments. This study develops a physics-based model that incorporates van der Waals forces, electrostatic interactions, and Coulomb forces to analyze the detachment and dispersion behavior of lunar dust. Numerical simulations using MATLAB are conducted to determine the critical detachment radius under varying driving speeds and slip rates. Results indicate that increasing either speed or slip rate reduces the detachment threshold from 19.11 to 17.79 µm. Further analysis of the simulated dust-dispersion images reveals a clear stratification in ejection height as a function of particle size. Under high-velocity and high-slip-rate conditions, small particles (radius < 20 µm) dominate the upper portion of the dust plume, reaching vertical propagation heights of up to 1.8 m, which substantially exceed the height of the rover itself. Validation against Apollo mission imagery confirms the model’s applicability. These findings provide theoretical guidance for dust mitigation strategies in future lunar rover missions.
The temperature difference is the driving force for heat conduction. Conventional liquid-cooling technologies rely on a single conduction path between the heat source and the coolant to establish a stable temperature difference. This reliance leads to relatively long startup times and may cause temperature overshoot in components exposed to high heat fluxes during intermittent operation. To address this limitation, this study proposes an ammonia–water desorption-based cooling system. By leveraging the flash evaporation of an ammonia–water solution, the system actively creates a localized low-temperature region and a larger thermal driving force, thereby enabling rapid cooling of pulsed heat sources. Experimental results characterizing the dynamic thermal response show that pressure-difference-driven flash evaporation during startup can reduce the desorption chamber temperature by 10°C–15°C within 30 s, while sustaining effective heat transfer for more than 500 s. Parametric analysis further indicates that increasing the initial concentration and filling charge of the desorption solution extends the duration of reliable heat dissipation. However, excessive filling increases the ammonia desorption path length and impedes bubble detachment, which does not necessarily enhance desorption performance. Within an appropriate range, increasing the cooling water flow rate or lowering its inlet temperature in the absorption chamber further improves desorption cooling capacity, resulting in an increase of more than 25
Liquid-gas phase change cooling technology has emerged as an effective strategy for addressing the thermal management challenges of high-power chip arrays within confined spaces. However, conventional two-phase cooling systems, which rely on hardware design and structural optimization, are often constrained by limited flexibility, complex integration, and high costs. To overcome these limitations, this study adopts the concept of “software cooling” and investigates a 4×4 high-power chip array cooled by a two-phase parallel-channel cooling plate. A non-time-varying simulation method enabling rapid and coupled evaluation of the thermal-hydraulic performance of the two-phase cooling plate and the temperature of the mounted chips is employed to analyse the effect of chip power allocation on the overall system performance. Steady-state multi-objective optimization is conducted using the NSGA-II genetic algorithm, targeting the total power of the chip array, the variance of power among chips, and the pressure drop across the cooling plate. By adjusting the power distribution strategy among chips at the software level without altering the hardware configuration, substantial improvements in cooling performance are achieved. The results demonstrate that, compared with the conventional core-based fixed scheduling strategy and under identical cooling conditions with the average chip temperature constrained to below 373.15 K, the optimized solution yields an approximately 12.3
Two electromechanical arms are combined to control the movements of a brush for plotting different spatial patterns and attractors. The brush is attached to the end of one arm of the electromechanical devices, and two electromechanical arms are forced to move along perpendicular directions in a plane. A hyperchaotic Lorenz system is used as the signal source. Its two output variables are encoded and used as driving currents in the coils of the robot arms’ electromechanical devices, and changes to those currents modify the movements of the brush. A scale transformation is applied to the physical equations and physical energy to obtain a dimensionless theoretical model and an energy function. Field energy from the signal sources is shunted to the electromechanical devices. The moving arms generate motional electromotive forces and provide feedback to the driving circuits (signal sources); that is, the coils of the robot arms are considered as load circuits (additive branch circuits) of the driving circuits. The trajectory of the brush seldom produces complete patterns and attractors like those in the driving system, so two controllers are introduced to assist in adaptive drawing. The Hamilton energy function H1 of the multi-wing Lorenz system (MWLS) is provided, and the energy is characteristic of the driving system. An energy-based adaptive control law is proposed that dynamically adjusts the system parameters based on the MWLS’s energy distribution, thereby significantly improving flexibility. Bifurcation parameters, control intensity, and adaptive control parameters affect the dynamic characteristics of a signal source and the coupled robot arms. The energy-guided adaptive control scheme enables switching during drawing by a mechanical arm, approaching the desired graphic, and providing greater control freedom. Our results provide theoretical support for two-dimensional drawing by mechanical arms.
