最新刊期
卷 47 ， 期 3 ， 2026
- 摘要：In order to solve the problems of limited information that can be learned by the model due to the scarcity of training samples in few-shot scenarios， the performance of the model is restricted by the scale change of the object， and the catastrophic forgetting of the base class knowledge， a few-shot object detection algorithm with improved Faster R-CNN was proposed by following the transfer learning method. Firstly， in the feature extraction stage， the central coordinate attention mechanism was added to guide the model to learn the core features of the object， and multi-scale fusion and deformable convolution were introduced to enable the model to extract rich feature information on the feature map at different scales. Then， a gradual RPN was introduced， and a two-stage regression method was used to gradually optimize the generated regional suggestions. Finally， the knowledge compensation module was introduced to decouple the prior knowledge from the new knowledge with the help of the decoupling mechanism， so as to maintain a good memory of the base class. Experimental results show that， in the three novel class splits of the PASCAL VOC， compared with the mainstream small sample object detection method FSODCR， the proposed algorithm achieves an average increased in nAP by 8.37 percentage points and 6.63 percentage points in the 2-Shot and 5-Shot settings respectively； on the MS COCO， compared with FSODCR， the proposed algorithm achieves an average increase in nAP and nAP75 of 1.2 percentage points and 0.9 percentage points respectively under the 10-Shot， and nAP and nAP75 have increased by 1.2 percentage points and 0.8 percentage points under the 30-Shot， which significantly improves the accuracy and stability of object detection under few-shot conditions.关键词：object detection;deep learning;few-shot object detection;transfer learning4|0|0更新时间：2026-06-26
- 摘要：Aiming at the current issues in lane detection algorithms based on deep learning， such as insufficient real-time performance， limited global feature modeling capability， and the fact that most related studies remain at the simulation validation stage with a lack of deployment verification for practical embedded systems， this paper proposed a lane detection method based on a lightweight YOLO network and an attention mechanism， and completed the deployment and verification from algorithm design to a real-vehicle system. First， a lightweight Faster-Net was introduced as the backbone network， combined with partial convolution （PConv） modules for structural redesign， which significantly reduced computational complexity while maintaining feature extraction capability. Second， a self-attention mechanism was embedded after the SPPF module of the feature fusion network to enhance the model’s modeling ability for the global structure of lane lines and long-range spatial dependencies， while suppressing interference from complex backgrounds. The results show that the proposed method achieves a precision of 96.83% on the CULane dataset with a single-frame inference time of 11.0 ms， and an accuracy of 96.33% on the TuSimple dataset with a frame rate of 90.9 FPS， outperforming current mainstream algorithms. Finally， the optimized model was deployed on the Jetson Orin NX embedded platform， and functional verification was carried out in a real-vehicle sandbox environment. The results demonstrate that the system exhibits good stability and real-time performance in real-world scenarios. This work not only improves the theoretical performance of lane detection algorithms， but also transitions the technology from dataset validation to practical in-vehicle system implementation， providing a complete and feasible technical pathway for the engineering application of autonomous driving perception technology.关键词：deep learning;autonomous driving;lane detection;self-attention mechanism;Faster-Net;real-vehicle deployment5|0|0更新时间：2026-06-26
- 摘要：Infrared small target detection （ISTD） has important application value in the fields of military early warning， disaster rescue and precision guidance， etc. However， the existing methods have core problems such as an imbalance between local and global features， and insufficient robustness of the model under small sample conditions. To address these issues， this paper proposed a new model called multiscale dynamic fusion with local contrast learning （MDL）， which was based on the hybrid architecture of Transformer and convolutional neural network （CNN）， integrating multi-scale dynamic weighting and adaptive local contrast constraints. Firstly， a dynamic weighted fusion multi-scale context augmentation （MCA） was introduced into the architecture of Transformer， which effectively solved the problem of imbalance between local and global features in small target detection， and at the same time significantly reduced the redundancy of the attention mechanism. Secondly， this model included a dual-branch hybrid dilated convolution （DHDC） module located in the deeper layers of the network， which enhanced the detection accuracy through high-resolution feature reconstruction and semantic information from deeper features as a way to strengthen the target edge representation. In addition， a local contrast convolution （LCC） module was designed， which incorporated the traditional local contrast idea to significantly reduce the model’s dependence on the number of training samples and maintains excellent robustness even on small sample datasets. Finally， a series of experiments were conducted on NUAA， IRSTD-1k and SIRSTAUG datasets. The results show that the method proposed in this paper achieved F1-Scores of 88.98%， 83.51%， and 86.38% on the three datasets， representing improvements of 1.06， 2.48， and 0.76 percentage points compared with the baseline model， respectively.关键词：infrared small target detection;Transformer;attention mechanism;multiscale features;local contrast4|0|0更新时间：2026-06-26
- 摘要：Traditional table tennis rubbers are prone to friction performance drift under prolonged ball-striking friction and thermal-oxidative ageing conditions， and often present the challenge of balancing grip capability （friction） with rebound （durability）. To enhance the friction stability and ageing resistance of rubber surfaces， this study utilized the metal-organic framework material HKUST-1 as a precursor. Through high-temperature carbonisation， a Cu/C composite porous carbon material （MOFs-C） was prepared. This material was then incorporated into a styrene-butadiene rubber （SBR） matrix at varying ratios to construct composite rubber films. Characterisation via SEM， XRD and other methods indicates that MOFs-C retains its octahedral-like particle morphology and crystalline features after carbonisation， while developing a porous， rough surface structure conducive to enhancing specific surface area and interfacial interaction sites. In accelerated ageing and friction tests simulating actual usage conditions， the coefficient of friction for pure SBR membranes increased from 0.75 to 1.32， whereas the SBR/MOFs-C composite membranes maintained a coefficient of friction within the range of 0.77 to 0.82. The results indicate that the synergistic interaction between the porous structure of MOFs-C and the filler-rubber interface reduces the effective diffusion of oxygen within the rubber and inhibits thermal-oxidative degradation of molecular chains. This enables the rubber to maintain more stable surface tribological behaviour under repeated friction conditions. This study provides a reference for the application of MOF-derived porous carbon materials in regulating the durable friction properties of sports equipment rubber.关键词：metal-organic framework materials;table tennis rubber;friction coefficient;anti-aging4|0|0更新时间：2026-06-26
- 摘要：Neutral aqueous organic redox flow batteries （AORFBs） represent a compelling candidate for scalable energy storage integrated with renewable power generation， largely due to the earth-abundant， tunable， and sustainable nature of organic electroactive compounds. However， current redox-active species still face critical limitations， including insufficient solubility， inadequate voltage window compatibility， and poor cycling stability， hindering their practical deployment in neutral-pH aqueous electrolytes. In this work， a new class of pyridinium derivatives of thiadiazole intermediates， namely 4，4'-（1，3，4-thiadiazole-2，5-diyl）bis（1-（3-trimethylammonio）propyl）pyridin-1-ium （BnSn） and 4，4'-（1，3，4-thiadiazole-2，5-diyl）bis（1-methylpyridin-1-ium） （BnSm） were designed and synthesized. By introducing a 1，3，4-thiadiazole bridging unit into the molecular framework， the conjugation system was systematically extended， and an intermolecular hydrogen-bonding network based on nitrogen lone pair electrons was synergistically constructed. The resulting molecule achieves exceptional aqueous solubility （0.89 mol·L-1）， a well-positioned low potential （-0.98 V）， and outstanding electrochemical reversibility under neutral conditions. When employed as the negative electrolyte， it delivers remarkable long-term stability. After 500 constant-current charge-discharge cycles at 0.05 mol·L-1 and a current density of 600 A·m⁻²， the capacity retention rate remained above 70%， corresponding to an average cycle decay rate of only 0.016% and a daily decay rate of 0.16%. Combined density functional theory calculations and experimental mechanistic analysis indicate that the 1，3，4-thiadiazole bridge not only enhances the intramolecular electron delocalization to improve the structural rigidity and thermodynamic stability of the redox center， but also effectively inhibits the irreversible dimerization， hydrolysis and free radical-mediated degradation pathways of active species by virtue of its strong polarity and hydrogen bond donor-acceptor dual functional properties. Thus， it simultaneously optimizes the dissolution kinetics， charge transfer efficiency and cycle durability.关键词：energy storage;flow battery;aqueous organic flow battery;redox species;intermolecular hydrogen bonding4|0|0更新时间：2026-06-26
- 摘要：During the process of hydrogen production from water-splitting assisted by the urea oxidation reaction （UOR）， developing a high-performance nickel-based catalyst for UOR is a key challenge that must be addressed urgently to enhance hydrogen production efficiency. In this work， the Co/Cr co-doped nickel sulfide composite （Co，Cr-NiS/CC） was prepared on carbon cloth （CC） by the hydrothermal-calcination method. The composition， morphology， structure， and surface chemical state of the composite were characterized by X-ray diffraction （XRD）， scanning electron microscopy （SEM）， transmission electron microscopy （TEM）， and X-ray photoelectron spectroscopy （XPS）. The electrochemical performance of the composite was tested in the electrolyte of 1.00 mol·L-1 KOH and 0.33 mol·L-1 urea. Meanwhile， density functional theory （DFT） calculations were conducted on the NiS （001） and Co，Cr-NiS （001） models. The structure-activity relationship was revealed through the analysis of the density of states （DOS）， the d-band center， and the energy barrier of the rate-determining step （RDS） for the UOR. Electrochemical tests demonstrate that the Co，Cr-NiS/CC composite exhibits excellent catalytic activity for UOR. Specifically， at the current densities of 100 A·m-2 and 1 000 A·m-2， the required potentials are 1.321 V and 1.378 V， respectively， which are lower than those of the other control samples. At a potential of 1.527 V， after 200 h of chronoamperometry （CA） testing， the retention rate of current density is 64.83%. DFT calculations indicate that the co-doping of Co and Cr enhances the intensity of the DOS at the Fermi energy （EF） of the composite， shifts the d-band center of the composite upwards， reduces the energy barrier of the RDS of UOR to 0.331 eV， and optimizes the electronic structure of NiS and the adsorption energy of reaction intermediates. This work proposes a new approach to constructing an efficient non-metallic UOR catalyst.关键词：Co，Cr-NiS/CC;co-doping;urea oxidation reaction;alkaline electrolyte;density functional theory4|0|0更新时间：2026-06-26
- 摘要：In response to the overcapacity of bulk chemicals such as ethylene glycol， methanol， and ethanol in the coal chemical industry， the development of high-value-added ethylene glycol ether products has emerged as an important technological pathway. Due to the differences in molecular structure and physicochemical properties between methanol and ethanol， the thermodynamic behaviors of their dehydration reactions with ethylene glycol may vary， which in turn affects the product distribution and the selection of process conditions. To reveal the intrinsic differences in the two reaction systems， a systematic heterogeneous thermodynamic calculation was conducted for the methanol-ethylene glycol and ethanol-ethylene glycol dehydration systems based on the Gibbs free energy minimization principle. The UNIFAC activity coefficient model was employed to correct for liquid-phase non-ideality， and the genetic algorithm was utilized for global optimization. The effects of operating parameters on the equilibrium conversion rate， product selectivity， and phase equilibrium composition of the two reaction systems were investigated under the conditions of 100 mol of ethylene glycol feed， temperatures ranging from 20 to 300 °C， pressures from 0.1 to 0.7 MPa， and the feed molar ratio of methanol or ethanol to ethylene glycol ranging from 1∶1 to 7∶1. The results show that there are significant differences in the heterogeneous temperature ranges of the two systems. At 0.3 MPa and a feed molar ratio of 4∶1， the heterogeneous temperature range of the methanol-ethylene glycol system is 50~110 ℃ （low starting point， wide range）， while that of the ethanol-ethylene glycol system is 100~120 ℃ （high starting point， narrow range）. Pressure mainly exerts an indirect influence on reaction equilibrium by affecting the temperature range of phase transitions， and it has a negligible effect on equilibrium conversions rate and product selectivities. With the increase of the feed molar ratio， both systems mainly produce diethers， the selectivities of ethylene glycol dimethyl ether and ethylene glycol diethyl ether continuously increase and approach 100%.关键词：methanol-ethylene glycol;ethanol-ethylene glycol;ethylene glycol ether;Gibbs free energy minimization;heterogeneous thermodynamic4|0|0更新时间：2026-06-26
- 摘要：Natural surfactants have significant advantages such as biodegradability， non-toxicity， and easy availability， making their extraction processes and technologies a focus of research. This study employed the high-gravity foam separation technology to extract saponins from Sapindus mukorossi， aiming to develop an efficient separation and purification process. The effects of different operating conditions on the separation efficiency were investigated. The results indicate that under the optimal conditions of high-gravity factor of 30， feed mass concentration of 0.1 g·L-1， gas flow rate of 150 L·h-1， liquid loading volume of 4 L， and liquid flow rate of 250 mL·min-1， the recovery rate of saponin from Sapindus mukorossi by high-gravity foam separation reaches 90%， the enrichment ratio reaches 6.38， and the purity is increased from 65% to 95% in a single pass. Compared with traditional foam separation methods， the recovery rate is increased by 36 percentage points， the enrichment ratio is improved by twofold， and the purity is increased by 30 percentage points. Furthermore， the structure and purity of commercially available saponins were compared with those of the self-prepared saponins. The results indicate that the structures of the two types of saponins are consistent， and the purity of the self-prepared saponins is higher than that of the commercially available ones. The high-efficiency separation of saponins achieved in this study using high-gravity foam separation technology is primarily attributed to the intensification of the interfacial adsorption process in foam separation by the high-gravity field. This process alters the morphology， size， and distribution of the foam， enhances the effective contact area between saponins and bubbles， and thereby enables the efficient separation and purification of saponins from Sapindus mukorossi. The high-gravity foam separation extraction process employed in this study offers significant advantages， including the absence of any new chemical additives， and environmental friendliness. This approach addresses the issues of low separation efficiency and environmental pollution associated with current extraction methods， providing new insights and techniques for the high-performance separation and purification of natural surfactants.关键词：high-gravity;foam separation;natural surfactants;saponin from Sapindus mukorossi9|0|0更新时间：2026-06-26
- 摘要：To address the high cost issue of measuring blast shock waves using the self-developed effect target method and the limitation of insufficient experimental validation in finite element analysis， this paper proposed a closed-loop research methodology integrating effect target experiments with Abaqus/Explicit finite element simulation， adapted for predicting the dynamic response of a 40 mm diameter micro effect target. By integrating the Brode formula， a modified Friedlander waveform and the CONWEP load algorithm， and combined with the Johnson-Cook dynamic constitutive model， a “load-structure-material” chain simulation model was constructed to systematically investigate the plastic dynamic response of 1060 aluminum thin plates under blast loading. The simulation results were validated by both shock tube and field effect target experiments， with the relative error in predicting central deflection being less than 6%. Parametric sensitivity analysis indicates that the material yield strength is the most critical parameter affecting accuracy， and a 0.5 mm mesh achieves the optimal balance between efficiency and accuracy. Compared with existing typical models， the proposed method significantly reduces the prediction error from 10.3%~16.1% to 5.7%. This method can complete the virtual optimization of target plate parameters before experiments， reducing the number of experiments by more than 40%， providing an efficient and low-cost technical approach for the anti-explosion design of military equipment.关键词：blast loading;finite element analysis;Abaqus/Explicit;effect target;Johnson-Cook model;parametric sensitivity9|0|0更新时间：2026-06-26
- 摘要：To investigate the dispersion process and flow field characteristics of energetic metal powder driven by explosive detonation， a charge structure for driving the dispersion of the energetic metal powder was designed， and the explosive-driven dispersion process was captured using a high-speed camera. The particle cloud velocity was calculated， and its combustion effect upon impact with the steel target was studied. Based on the FLUENT-EDEM fluid-structure interaction method， a physical model for the explosive dispersion of energetic metal powder particles was established， and a simulation study on the flow field characteristics of the particle cloud under shock wave action was conducted. The results indicate that during the initial stage of the explosive dispersion test， the particle cloud diffuses conically at a speed of approximately 400 m/s， and upon high-velocity impact with the steel target， a secondary explosion occurs， producing intense flames which demonstrate a significant afterburning effect of the energetic particles. The dispersion morphology and velocity of the particle cloud in the initial stage， as calculated by the simulation model， are in good agreement with the high-speed photographic observations. The particle cloud
- 摘要：Medication stocking recommendation， which proactively provides medication suggestions by forecasting disease evolution， is critical for optimizing preventive healthcare inventory management. To address challenges such as the long-tail distribution of medical codes in electronic health records （EHRs）， insufficient multi-source relational mining， and the lack of inter-task coordination mechanisms， it is essential to enhance representation learning capabilities under complex medical associations. This paper proposed a multi-relational enhanced medication stocking recommendation method based on hyperbolic graph contrastive learning （HGCL-MR）. The method first integrated diagnosis ontologies， medication ontologies， and clinical statistical patterns to construct a heterogeneous multi-source prior knowledge graph for representing complex medical associations. On this basis， a dual-view hyperbolic graph contrastive learning mechanism was designed， which utilized the Poincaré ball model to align ontological knowledge with statistical features in a unified hyperbolic space. By leveraging the negative curvature properties of hyperbolic space， the model precisely captured the hierarchical structure of medical codes and enhanced the representation capability for long-tail nodes. To explicitly model inter-task interaction logic， a learnable association matrix with soft constraints was introduced to achieve collaborative optimization between disease prediction and medication recommendation through cross-task knowledge transfer and prior guidance. Finally， a hierarchical multi-task decoding architecture was adopted， utilizing a sequential learning network to capture dynamic patient disease evolution and performing cross-space mapping from disease prediction vectors to the medication recommendation space via the association matrix. Experimental results on the MIMIC-III dataset demonstrated that HGCL-MR improved Jaccard scores by approximately 3.5% and 3.6% in the two respective tasks， and increased F1 scores by about 1.9% and 1.8%. Ablation studies validated the gains of hyperbolic geometric representation in alleviating label imbalance. The model achieved balanced performance when the weight parameters were set to
- 摘要：Precise segmentation of colon polyps is crucial for early screening and clinical diagnosis of colorectal cancer. However， large variations in polyp size， shape， and boundary characteristics， along with low contrast and blurred edges， make it challenging to balance segmentation accuracy and computational efficiency. To address these issues， a lightweight dynamic multi-scale segmentation network （LDMS-Net） was proposed. The model adopted a U-shaped encoder-decoder architecture and introduced a dynamic multi-scale module （DMM） in the encoder. Specifically， global average pooling and a two-layer fully connected network were used to generate adaptive weights， enabling dynamic fusion of multiple convolution branches with different dilation rates （1，2， 3） and a standard convolution branch for adaptive multi-scale feature representation. A structural re-parameterization strategy was further applied to merge the multi-branch structure into a single
- 摘要：Event extraction aims to extract structured information from unstructured text. However， when handling complex overlapping and nested events， existing single-stage models face challenges such as task cascading error propagation， long-tail class recognition bottlenecks， and feature representation redundancy. To address these issues， this paper proposed an improved model based on structured cost-sensitive learning and feature redundancy elimination gating. First， in the design of the loss function， the structured focal joint loss was proposed. By introducing a focusing parameter， this mechanism dynamically adjusted the weights of hard samples. We also designed novel task-dependent weighting factors to construct hierarchical gradient constraints， which algorithmically prevented the propagation of upstream trigger detection errors to argument roles. Second， to tackle noise interference during feature fusion， we designed a summation-based gated event fusion layer. This layer utilized Sigmoid gating signals to generate additive interference， replacing traditional linear concatenation strategies to achieve endogenous redundancy elimination while maintaining stable feature dimensionality. Experimental results on the Chinese financial and biomedical datasets show that the F1 score of the proposed model on FewFC improves by 1.6 percentage points compared to the OneEE baseline， outperforming the generative model Degree and the LLM correction strategy LC4EE. On the Genia13 dataset， the argument identification F1 score reaches 80.2%， improvement of 3.4 percentage points. Ablation studies indicate that removing the structured loss leads to a 1.8 percentage points decrease in F1 score. The improved method proposed in this study effectively solves the problems of missing logical constraints and feature noise while retaining the high efficiency of parallel inference， and enhances the accuracy of structured semantic parsing in complex scenarios. However， for extreme long-tail classes with very few training samples， the generalization ability of the model remains to be further enhanced through few-shot learning.关键词：event extraction;multi label classification;focal loss;structured loss;overlapping events;nested events4|0|0更新时间：2026-06-26
- 摘要：The critical nonlocal Choquard equation can be used to describe the theory of the polaron at rest， and can also be used as a self-gravitational model to describe the motion of a single particle in the self-gravitational field. In recent years， the study of the existence， multiplicity and related properties of solutions of Choquard equations has become a hot issue in the field of nonlinear analysis. This paper studied the existence of solutions for a class of critical Kirchhoff-Choquard equations with logarithmic term. Firstly， the Hardy-Littlewood-Sobolev exponent was classified and an appropriate truncation function was introduced to overcome the lack of compactness of the embedding caused by the critical exponent. Secondly， it was difficult to verify the boundedness of the PS（Palais-Smale） sequence due to the uncertainty of the logarithmic term and the fact that the logarithmic term does not satisfy the Ambrosetti-Rabinowitz type condition. Therefore， the local PS condition was obtained by using the concentration compactness principle， and the existence theorem of infinitely many solutions of the problem was established by means of the symmetric mountain pass lemma. Finally， the existence of the local minimal solution with negative energy was considered on the appropriate sphere （a constrained region of function space）， and then the additional conditions were applied to the parameters. It is proved that the local minimal solution is also the ground state solution. The results of this paper extend the research on Kirchhoff-Choquard equation or logarithmic equation in the relevant literature， and analyze the influence of the parameters in the Kirchhoff operator and the logarithmic perturbation term and Hardy-Littlewood-Sobolev embedding constants on the number of solutions.关键词：Kirchhoff-Choquard equation;critical exponent;logarithmic term perturbation;truncation function;symmetric mountain pass lemma4|0|0更新时间：2026-06-26
- 摘要：The strength of correlation of non-strongly-
-
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0