Abstract:To explore the effects of density regulation on the structure and understory biodiversity of Larix kaempferi plantations,Larix kaempferi plantations in the state-owned Gaoyanzi Forest Farm of Jianshi County,Hubei Province were taken as the research objects.The effects of four different density regulation treatments(light,moderate,heavy thinning and control)on the diversity of understory shrubs and herbs were analyzed.Methods such as fixed sample plot survey,community structure analysis,biodiversity index analysis,analysis of variance and Pearson correlation analysis were adopted to clarify the influence degree.The results showed that a total of 20 fixed sample plots were set up,and 32 species of understory shrub and herb layer plants were investigated,belonging to 26 genera of 21 families.Under different stand density regulation treatments,Symplocos paniculata was the dominant species in the shrub layer,while Indocalamus tessellatus was dominant in the herb layer.The Shannon-Wiener index,species richness and Pielou evenness index of shrub and herb layer plants showed a trend of first increasing and then decreasing with the increase of density regulation intensity,and reached the maximum under moderate thinning.The stand structure diversity had a significant negative impact on the understory shrub layer diversity,but no significant impact on the herb layer diversity.This study indicated that moderate thinning could effectively increase the number of species,improve species diversity,and be beneficial to the stable and sustainable development of Larix kaempferi plantations.

Abstract:To reveal how social norms influenced individual green behavioral decision-making through cognitive and affective mechanisms,a research path of"social norms→environmental affect→green behavioral decision-making"was constructed on the basis of the stimulus-organism-response(SOR)model and the cognition-affect-behavior(CAB)model.Meanwhile,a comparative analysis was conducted between publicity posters with and without social norm information by means of eye-tracking experiments.Furthermore,an integrated mechanism of"visual guidance-affective arousal-behavioral response"was established.The results showed that the posters with social norm information significantly enhanced visual attention metrics,where the total fixation duration and fixation count under the social norm condition were increased by 314.21%and 337.50%,respectively,compared to the posters without social norms.In addition,the Bootstrap mediation test results demonstrated that environmental emotion played a significant mediating role between social norms and green behavioral decision-making.This study found that social norms stimulated environmental emotion by enhancing visual salience,thereby promoting the public's green behavioral decision-making.It was suggested that in the design of environmental protection communication,emphasis should be placed on the visual highlighting and emotional integration of social norms,so as to promote the transformation of behavioral intervention from the"information transmission"mode to the"perceptual experience"mode.

Abstract:To address the insufficient feature extraction capability of lightweight models in lithography hotspot detection,a lithography hotspot detection model was proposed.This model which was named as ShuffleNetV2-MSDA-GHM-AAM(SMGA)adopted the improved shuffle net version 2(ShuffleNetV2)as its backbone network,incorporated a multi-scale dual attention(MSDA)module,and simultaneously integrated the gradient harmonizing mechanism(GHM)and additive angular margin(AAM)based on metric learning.The model's ability to model and perceive contextual information at different scales was enhanced,the inter-class discriminability of the feature embedding space was optimized,and dataset imbalance was alleviated.The experiments were conducted on the dataset of the 2012 international conference on computer-aided design(ICCAD 2012).The results showed that while maintaining a high detection recall of 98.22％,the average number of false alarms of the SMGA model was reduced to 484.This model provided a feasible solution for efficient and low-cost lithography hotspot detection in the integrated circuit design phase,possessing important engineering application value and promotion prospects.

Abstract:To enhance the generalization and adaptability of image restoration models under multiple harsh imaging environments,the research progress on all-in-one image restoration models was discussed systematically.Firstly,the current situation of six categories of all-in-one image restoration models under multiple harsh imaging environments was analyzed.Secondly,the characteristics of these models were summarized.Thirdly,the limitations of these models were discussed.Finally,the development trends were envisaged.It was found that current all-in-one image restoration models had improved image clarity and multi-scene applicability.However,they still suffered from limitations such as poor performance on real-world degraded images,low interpretability,and high model complexity.Future improvements could be achieved through real degradation modeling and self-supervised learning to enhance restoration performance;integrating attention mechanisms,physical modeling,and physics-informed networks to improve interpretability;and adopting modular design and model compression to enable lightweight deployment.This study provided important reference for the research and application of high-performance,general-purpose image restoration models.

Abstract:To clarify the spatiotemporal evolution patterns and driving mechanisms of gross primary productivity(GPP)in the Yangtze River Delta urban agglomeration from 2001 to 2023,remote sensing and meteorological data were integrated.A multi-scale analysis of GPP evolution characteristics was conducted using the coefficient of variation,Theil-Sen median trend combined with the Mann-Kendall test,and the Hurst exponent.By comparing the predictive performance of four machine learning models,the optimal light gradient booster machine(LightGBM)model was selected,and the Shapley additive explanations(SHAP)method was employed to reveal the importance and interaction effects of driving factors.It was found that the top 5 multi-year average GPP in the Yangtze River Delta region from 2001 to 2023 was measured at 1130.41g/(m2·a),with a significant upward trend observed in interannual variation(annual average change rate was 6.71g/(m2·a),P<0.05).Spatially,a pattern of ″high in the south and low in the north″ was presented:the southern forested area was designated as a high-value zone,while the northern region dominated by cultivated land and construction land was classified as a low-value zone.The LightGBM model was proven to be superior to the other models.The SHAP results demonstrated that the top 5 contribution degrees of driving factors were ranked as follows:elevation>vapor pressure deficit>nighttime light>slope>aridity index.The enhancement of GPP by the interaction between elevation and climate-anthropogenic factors was particularly notable.The impacts of various driving factors were found to exhibit distinct nonlinear characteristics,resulting in complex effects on vegetation GPP.This study was expected to provide a scientific reference for vegetation conservation and ecological governance in the Yangtze River Delta.

Abstract:To address the problem that traditional target tracking methods were difficult to achieve continuous and effective tracking due to limited observation field of view and external occlusion of the target,a dynamic target real-time tracking method based on two-dimensional galvanometer mirror system(GDTM)was proposed.Firstly,a forward-shifted field of view module was adopted to improve the two-dimensional galvanometer mirror system,which enabled the observation field of view to move rapidly with dynamic targets and expanded the observation range simultaneously.Secondly,a nonlinear constrained window(NCW)was designed to allow dynamic targets to be stably maintained within the local region of interest.Finally,the average peak-to-correlation energy(APCE)was used as the occlusion judgment criterion,and the unscented Kalman filter(UKF)was adopted to predict the optical axis angle of the galvanometer,thereby achieving the tracking of occluded targets.The results showed that in terms of prediction tracking accuracy,the UKF used in GDTM was improved by 28.57%and 16.67%respectively,compared with the Kalman filter(KF)and the extended Kalman filter(EKF).This method was characterized by such advantages as fast response,strong robustness,and low cost,and could continuously and stably track occluded dynamic targets over a large range.

Abstract:To address issues such as large target scale variations,target occlusion,and large model parameters in unmanned aerial vehicle(UAV)aerial images,the context feature enhancement-you only look once version 11(CFE-YOLOv11)model was proposed for aerial image target detection.Firstly,a lightweight downsampling convolution(LDC)module was designed,which enhanced feature information through dual-path downsampling and improved cross-channel information interaction using channel shuffling,thereby reducing the number of model parameters.Secondly,a convolutional three-scale kernel-adaptive dual-path separated and enhancement attention(C3k2-SEA)module based on separated and enhanced attention was designed to improve the feature extraction capability of the model.Meanwhile,multi-branch spatial and channel attention(MSCA)module was proposed to enhance the model′s ability to extract features of targets at different scales.Finally,dual loss optimization(DLO)module was used to optimize the detection effect of the model through a differential gradient gain allocation mechanism.The results showed that,compared with the YOLOv11 model,the CFE-YOLOv11 model achieved 2.1 and 2.0 percentage points,respectively,in mean average precision(mAP)at an intersection over union(IoU)threshold of 0.50 were achieved by the CFE-YOLOv11 model on the visual drones(VisDrone)and remote sensing object detection(RSOD)datasets,respectively,while the number of parameters was reduced by 15.4%.The CFE-YOLOv11 model not only improved the detection accuracy of aerial images but also alleviated the problems of model missed detection and false detection,providing an efficient solution for accurate detection of aerial image targets.

Abstract:To address the problem that nonlinear networked systems were often subjected to aperiodic denial-of-service attacks and false data injection attacks,the switching observer-based dynamic event-triggered control for a class of interval type-2 fuzzy networked systems was investigated.Firstly,considering the intermittent nature of aperiodic denial-of-service attacks and the unmeasurability of some system states,a nonlinear system based on the interval type-2 fuzzy model with switching observers was constructed.Meanwhile,to better save communication resources,a more advantageous dynamic event-triggered control method was employed.Subsequently,by applying the Lyapunov stability theory,a stochastic asymptotic stability condition for the controlled system and a controller design approach were derived.Finally,the stability of the controlled system when both types of network attacks occur was verified through a simulation example.This research provided key support for interval type-2 fuzzy systems to enhance their anti-interference capability and operational reliability when coping with hybrid network attacks,and possessed significant theoretical and practical application value.

Abstract:To address the issue of redundant computation in non-region of interest(ROI)areas when existing workpiece classification models processed high-resolution images,a workpiece classification network model with ROI adaptive contour-driven cropping(ACDC-ClassNet)was proposed.This model utilized contour detection to localize the largest contour and its center within the image.Based on this,a standardized square ROI cropping region was generated,effectively eliminating background interference.A pre-trained residual network-50 layers(ResNet-50)model was adopted,and its classification head was adjusted to match the number of workpiece categories,enabling efficient feature focusing and classification.The results demonstrated that this ROI cropping strategy achieved an average reduction of 72.15%in the area of redundant regions,allowing the model to focus more effectively on workpiece details.Compared to the original ResNet-50,ACDC-ClassNet was observed to achieve improvements of 3.83,4.04,3.64,and 4.13 percentage points in accuracy,precision,recall,and F1-score,respectively.Furthermore,the strategy was also shown to outperform efficient network(EfficientNet)and vision transformer(ViT)models,achieving accuracy gains of 8.40 and 4.27 percentage points.ACDC-ClassNet provided a novel technical pathway for efficient visual inspection in industrial settings.

Abstract:To address the issue that Android malware was often obfuscated to evade detection,a method named sensitive function call graphs detector(SFCG_Detector)based on representation learning of sensitive function call graphs(SFCG)was proposed for detecting Android malware.Firstly,function call graphs were extracted through static analysis,and a graph pruning strategy was designed to preserve critical nodes related to sensitive application programming interfaces(API).This reduced graph complexity while retaining behavioral semantics.For node representation,semantic features and structural importance of nodes were captured using a bidirectional encoder representation from transformers(BERT)model and Katz centrality,respectively.Subsequently,the SFCG were hierarchically learned using the graph sample and aggregation(GraphSAGE)network to generate graph-level embeddings to support the classification task.Experimental results on the Canadian institute for cybersecurity malware dataset 2020(CICMalDroid 2020)demonstrated that SFCG_Detector achieved a malware detection F1-score of 98.75%and recall of 98.89%.Compared with other methods,SFCG_Detector could effectively identify Android malware and improve the performance significantly.

Abstract:A systematic survey was conducted to investigate the composition of the fern flora in the Hubei Xingdou Mountain National Nature Reserve,with the aim of clarifying its geographical distribution patterns and elucidating its conservation significance.The study was carried out from 2013 to 2022,involving field investigations and specimen collection across 97 permanent plots,each measuring 20m×20m.Fern species within the reserve were identified and cataloged at the family and genus levels.Based on these data,distribution zones at both the family and genus levels were delineated,medicinal species were tallied,and comparisons with neighboring reserves were made.A total of 156 fern species belonging to 64 genera and 29 families were recorded.4 dominant families(Dryopteridaceae,Polypodiaceae,Athyriaceae,and Thelypteridaceae)were identified,accounting for 67.31%of the total species.The distribution at the family level was dominated by tropical elements(82.76%),while the genus-level distribution exhibited characteristics of a transitional zone,with both tropical and East Asian-North American temperate elements present.Only 7 medicinal fern species were confirmed,representing 4.49%of the total,indicating considerable potential for further development.Compared to surrounding protected areas,the species richness of ferns in Xingdou Mountain was relatively low,which may be related to the monospecific composition of the Metasequoia community structure.The fern flora of Xingdou Mountain was found to have an ancient origin and continuous lineage,though diversity levels remain suboptimal.It was recommended that the species composition of the tree layer be optimized,conservation of rare and endangered species be enhanced,and systematic evaluation of medicinal active components be conducted.The research would provide a scientific basis for the conservation of fern diversity,sustainable use of medicinal resources,and vegetation management within the reserve.

Abstract:To investigate the physiological responses of the″Changlin series″Camellia oleifera varieties under drought stress,a controlled drought stress experiment lasting 70 days followed by a 7-day rehydration treatment was conducted.The drought resistance of five major Changlin series varieties(C.oleifera'Changlin 03,04,18,40,and 53')was evaluated.Leaf gas exchange parameters,chlorophyll fluorescence,relative chlorophyll content,relative membrane permeability,as well as physiological and biochemical indicators such as malondialdehyde,proline,superoxide anion,and hydrogen peroxide content were measured.The results revealed significant differences in drought resistance among the varieties.C.oleifera'Changlin 40'was found to exhibit the stronger drought resistance,with superior maintenance of photosynthetic efficiency,photosystem II functionality,and cellular integrity,along with lower reactive oxygen species accumulation and effective osmotic regulation.In contrast,C.oleifera'Changlin 53'was the more sensitive,showing a significant decline in photosynthetic activity and severe oxidative stress.A membership function analysis based on the drought resistance coefficients of the five varieties at 70 days of drought stress ranked their drought resistance in the following order:C.oleifera'Changlin 40'>C.oleifera'Changlin 04'>C.oleifera'Changlin 03'>C.oleifera'Changlin 18'>C.oleifera'Changlin 53'.This study provided a scientific basis for the selection and cultivation of drought-resistant C.oleifera varieties in arid regions,contributing to the enhancement of stress tolerance in the C.oleifera industry.

Abstract:To address the issue that high-density salt-and-pepper noise severely damaged image details,a novel image restoration algorithm for removing salt-and-pepper noise with the plug-and-play alternating direction method of multipliers(SP3A)was proposed.Firstly,the noise locations were identified by a noise detection filter,and the image denoising problem was transformed into an image inpainting problem.Secondly,a mathematical model based on variational energy was established.Using the plug-and-play alternating direction method of multipliers(PnP-ADMM)as the solution framework,the regularization subproblem was replaced by an existing denoiser,and the SP3A algorithm was derived.Finally,the convergence of the SP3A algorithm was demonstrated under the condition that the denoiser was bounded.Numerical experiments showed that the SP3A algorithm was improved by 0.74 dB in peak signal-to-noise ratio(PSNR)and 0.12 in structural similarity index measure(SSIM),compared with L0-overlapping group sparse total variation(L0-OGSTV)algorithm under high noise levels.This study confirmed that the detailed texture parts of images were effectively restored by the SP3A algorithm,and a new solution was provided for the removal of high-density salt-and-pepper noise.

Abstract:To investigate the trait response mechanisms of evergreen and deciduous tree species along different altitude gradients(1250～1650 meters),Hubei Xingdou Mountain National Nature Reserve was selected as the study area.A total of 230 plant samples(148 evergreen plant samples and 82 deciduous plant samples)were collected,involving 37 plant species.The leaf functional traits of different leaf habit tree species were measured for research.The results showed that chlorophyll was significantly positively correlated with leaf thickness(P<0.01),and leaf area was influenced by both leaf width(R2=0.64～0.79)and leaf length(R2=0.13～0.80),with evergreen tree species having a higher explanatory power for leaf area;The leaf thickness and chlorophyll content of evergreen plants were significantly higher than those of deciduous plants(P<0.001),while the leaf width and specific leaf area of deciduous plants were significantly larger(P<0.05);The overall chlorophyll content and the overall leaf thickness of deciduous plants decreased compared to evergreen plants.No significant changes were found in leaf width,length,leaf area,specific leaf area,and dry matter content across all altitude gradients.The study revealed that evergreen and deciduous tree species adapt to mountainous environments through a trade-off strategy:evergreen tree species maintain sustained photosynthetic capacity with thick leaves and high chlorophyll content,while deciduous tree species achieved rapid resource acquisition with wide leaves and high specific leaf area.This study provided a research basis for vegetation adaptation mechanisms in subtropical evergreen deciduous broad-leaved mixed forests.

Abstract:To address the security vulnerabilities caused by the lack of authentication and encryption mechanisms in vehicular controller area network(CAN),a CAN bus anomaly detection model based on temporal-spatial attention bi-stream network(TSABiNet)was proposed.The model employed a parallel dual-stream architecture:Firstly,semantic features of CAN messages were extracted through the content stream utilizing an embedding layer and one-dimensional convolutional neural network(CNN).Secondly,temporal dependencies of message sequences were modeled through the temporal stream based on bidirectional long short-term memory(BiLSTM)networks and self-attention mechanisms.Finally,adaptive weight allocation and deep fusion of content and temporal features were achieved through a cross attention fusion(CAF)module utilizing cross-attention mechanisms.The model was validated on the car-hacking dataset(CHD).The results demonstrated that,compared to deep CNN and CAN anomaly detection using LSTM autoencoder(CANnolo)model,the TSABiNet model achieved F1 score improvements on denial of service(DoS)attack of 5.40 and 2.48 percentage points,respectively.The research findings validated the effectiveness of the TSABiNet model for CAN bus anomaly detection applications.

Abstract:To elucidate the antioxidant mechanisms of the hyper-selenium-accumulating plant Cardamine hupingshanensis under selenium stress,bioinformatics methods were used to identify and analyze the thioredoxin-y(TRX-y)gene family.Five members of the TRX-y gene family were identified in C.hupingshanensis,distributed across four chromosomes.All members contained the conserved active center"WCGPC"and were localized in chloroplasts.Phylogenetic analysis revealed that they clustered in Class I with AtTRX-y from Arabidopsis thaliana,and the ChTRX-y isoform possesses a specific motif.Selenium stress expression profiling revealed that high selenium concentration(80000μg/L)significantly upregulated the expression of ChTRX-y1-1 and ChTRX-y1-2 in roots(24 hours)and leaves(9 hours),suggesting that these were core selenium-responsive genes.ChTRX-y1-1 was further cloned,and the pET28b-ChTRX-y1-1 recombinant vector was constructed.Expression in Escherichia coli BL21 was achieved using induction conditions of a temperature of 25℃,a molar concentration of 0.1 mmol/L Isopropyl-beta-D-thiogalactopyranoside and a duration of 10 h.A highly pure recombinant protein of approximately 59 kDa was successfully obtained,reaching a concentration of 1.26 mg/mL after ultrafiltration.This study revealed the key role of ChTRX-y1-1 in selenium stress and established a prokaryotic expression system for its expression,providing a molecular basis for further investigation into the selenium metabolism mechanism in C.hupingshanensis.

Abstract:To extract polyphenols from Oenanthe javanica efficiently and explore its antioxidant activity,ultrasonic-assisted ionic liquid method was used,and the effects of ultrasonic time,ionic liquid concentration,ultrasonic temperature and liquid-material ratio on the extraction yield were investigated by single factor experiments.The extraction process was further optimized by Box-Behnken response surface design,and the antioxidant activity of the extract was evaluated by DPPH·,ABTS+·,·OH and O2-·scavenging experiments.The results showed that the optimum extraction process was ultrasonic time of 20 min,ionic liquid concentration of 0.31 mol/L,ultrasonic temperature of 62℃and liquid-material ratio of 31∶1.The extraction yield of polyphenols reached 56.91mg/g,and the relative error with the predicted value of the model was 0.65%,which verified the prediction efficiency and reliability of the mathematical model.The antioxidant activity experiment showed that the scavenging effect of Oenanthe javanica polyphenols on 4 kinds of free radicals was concentration-dependent,among which the scavenging effect on ABTS+·was the strongest,and its median inhibition concentration(IC50)was only 53.19mg/L,while its scavenging effect on O2-·free radicals was relatively weak,with IC50 reaching 302.78 mg/L.At the appropriate concentration,the scavenging rates of 4 free radicals were all above 72%,especially for ABTS+·,which could reach 92.67%of vitamin C.The research showed that the ultrasonic-assisted ionic liquid method was a feasible technology for efficient extraction of polyphenols from Oenanthe javanica,and the polyphenols from Oenanthe javanica had broad-spectrum and efficient antioxidant activity,which provided scientific support for its functional development and utilization.

Abstract:To improve the monitoring and early warning levels and prevention and control efficiency of instability and collapse disasters of unstable rock masses under complex geological conditions,through extensive retrieval and in-depth analysis of relevant domestic and international literature,a systematic study was conducted on the geological disasters of unstable rock masses,as well as the current research status,existing problems,and development trends of monitoring and early warning technologies for unstable rock masses.The results indicated that rockfall monitoring and early warning technologies had evolved from traditional manual observation to real-time sensor-based monitoring using three-dimensional laser scanning and fiber optic sensing,and further integrated machine learning(ML)and deep learning(DL)for intelligent identification.Microseismic monitoring and multi-physical field coupling simulation had also become important supplements.However,current technologies still suffered from limited accuracy in identifying disaster-breeding features,the absence of dynamic criteria for instability precursors,and restricted timeliness of early warnings.Future research could focus on multi-source data fusion and the integration of mechanism-driven and data-driven modeling to establish an efficient and intelligent early warning and emergency response system,promoting a shift in rockfall disaster prevention from passive management to active control.

Abstract:To address the problem that mask optimization accuracy and efficiency were difficult to balance under advanced process nodes,an end-to-end inverse lithography mask optimization model based on the improved lightweight mobile neural network version 2(MobileNetV2)was proposed.Firstly,MobileNetV2 was adopted as the backbone network of the model,giving full play to its advantages of small parameter size and efficient feature extraction,which could effectively adapt to the processing requirements of large-scale integrated circuit data.Secondly,coordinate attention(CA)modules were embedded multiple times in the feature extraction stage,through which spatial position information and channel features were fully fused,effectively enhancing the model′s ability to model mask edge details and target structures.Finally,a content-aware reassembly of features(CARAFE)module was integrated into the decoder part,enabling efficient and dynamic feature reassembly,which effectively alleviated the problems of information loss and edge blurring in traditional upsampling.The results showed that compared with the neural inverse lithography technology(Neural-ILT),attention-accelerated inverse lithography technology(A2-ILT),and differentiable optical proximity correction(DiffOPC)models,the improved MobileNetV2 model significantly reduced the turnaround time by 97.3%,94.2%,and 96.1%respectively while decreasing the process variation band index.The model was able to balance accuracy,efficiency,and process adaptability,effectively meeting the practical needs of integrated circuit mask design and manufacturing.

Abstract:To address the problem that existing event causality identification methods failed to consider the noise interference generated after the introduction of external knowledge,leading to increase ambiguity in event representation and thereby affecting the identification performance,an event causality identification model based on external vocabulary and hypergraph denoising(EHDM)was proposed.Firstly,contextual knowledge was retrieved from an external vocabulary to enrich the semantic information of events,encoding event descriptions with this contextual knowledge.Secondly,a hypergraph was constructed based on knowledge features corresponding to multiple relationships within the event background knowledge.Features were further processed through a hypergraph convolutional neural network and a multi-head attention mechanism to obtain denoised event feature representations.Thirdly,context-based feature representations were encoded from the events and their contexts,which were then fused with the denoised event feature representations via a gate unit.Finally,the fused feature representations were input into a multi-layer perceptron to obtain prediction values,thereby achieving causality identification.The results demonstrated that EHDM achieved an F1 score improvement of 1.5 percentage points over relation graph convolutional networks(RGCN)on the intra-sentential aspect of the causal-timebank(CTB)dataset.On the intra-sentential aspect of the event story line(ESL)dataset,EHDM achieved an F1 score improvement of 2.4 percentage points over RGCN and also recorded increases of 2.1 and 3.0 percentage points in cross-sentence and overall F1 scores,respectively,compared to the event relation graph Transformer model.This research confirmed EHDM′s effective application in the field of event causal relationship identification.