Abstract:To address the issues of insufficient utilization of spatial information and semantic ambiguity in the inpainting of large missing areas of the existing Transformer-based image inpainting models,a Transformer-local importance-based attention and Mish activation function encoder-decoder network(TLIAM-Net)image inpainting model was proposed.Firstly,the TLIAM-Net model was designed with an encoder-decoder architecture,where Transformer blocks were progressively connected to comprehensively exploit the hierarchical feature information of the image.Subsequently,the local importance-based attention(LIA)mechanism was introduced following each Transformer block,to enhance the model′s spatial information utilization capability.Finally,the Mish activation function was employed within the Transformer blocks,to ensure smoother feature transitions and improved capture of subtle details.The results demonstrated that the TLIAM-Net model achievedthe Fréchet inception distance(FID)value of 12.0116 under mask rates of(0.5,0.6]on Flickr-faces-high quality dataset(FFHQ),representing a 51.17%reduction compared to the multi-level interactive siamese filtering(MISF)model.The accuracy and deblurring performance of image inpainting tasks were significantly improved by the TLIAM-Net model,which could be successfully applied to multiple subtasks including deblurring,denoising,and defect completion,and the cost of manual inpainting was substantially reduced.

Abstract:To address the disturbance of terrestrial ecosystem carbon storage caused by land use changes in the Jialing River Basin,the spatio-temporal evaluation characteristics and driving factors of carbon storage in the basin were systematically assessed based on land use data from 2000 to 2020 by coupling the integrated valuation of ecosystem services and trade-offs(InVEST)model with a geographical detector.Additionally,the potential impact index was applied to reveal the vulnerability of its ecosystem services.The results demonstrated that:From 2000 to 2020,cropland and forest land were the dominant land types in the Jialing River Basin.The areas of cropland and grassland decreased,while forest land and construction land expanded substantially,with significant transitions observed among cropland,forest land,and grassland.The total carbon storage in the basin showed a continuous increasing trend,with a net increase of 0.90×108 t during the 20-year period,and exhibited a spatial distribution pattern characterized by higher values in the north and lower values in the south.The comprehensive land use degree index in the basin decreased by 5.31,the potential impact indices indicated indicating a negative potential impact and consequently an enhanced vulnerability of the carbon storage service.Elevation,annual average temperature,and slope were identified as the primary drivers of the spatial heterogeneity of carbon storage.Interaction detection revealed that two-factor enhancement was the dominant type of factor interaction.The findings of this research provided a scientific framework for territorial spatial optimization and the synergistic advancement of the"Dual Carbon"goals in the ecological barrier zone of the upper Yangtze River.

Abstract:To investigate the mechanisms of selenium metabolism and regeneration in Lilium lancifolium Thunb.scales under selenium-enriched cuttage conditions,scales were treated with sodium selenate at concentrations of 0,3,6,12,and 18mg/kg in a composite substrate for 40 days.Morphological regeneration indices,selenium content,jasmonic acid(JA)and malondialdehyde(MDA)levels,activities of peroxidase(POD)and catalase(CAT),as well as the expression of genes involved in selenium uptake and assimilation(LlSULTR1;1,LlAPS)and JA synthesis(LlAOC,LlLOX)were measured.The results indicated that both total and organic selenium content in the scales increased with rising sodium selenate concentration.The highest organic selenium proportion of(53.2±2.3)%was observed at 6 mg/kg.This treatment also resulted in peak JA content,maximal POD and CAT activities,minimal MDA content,and significantly upregulated expression of the aforementioned genes.Meanwhile,the rooting rate and bulblet induction rate were significantly enhanced,reaching(65.6±5.1)%and(86.3±5.5)%,respectively.However,when sodium selenate exceeded 6mg/kg,the organic selenium ratio decreased,JA synthesis was inhibited,antioxidant capacity declined,and adventitious root formation was significantly suppressed.In conclusion,6mg/kg sodium selenate was the optimal concentration for selenium-enriched cuttage culture of L.lancifolium scales.These findings provided a theoretical basis for improving selenium-enriched cultivation and cutting propagation techniques for L.lancifolium.

Abstract:To establish an enzyme linked immunosorbent assay(ELISA)detection method for Tyrophagus putrescentiae,New Zealand white rabbits were immunized with total mite protein to prepare polyclonal antibodies.BALB/c mice were immunized with the mite allergen Tyr p10 protein to prepare monoclonal antibodies.Using the polyclonal antibody as the coating antibody and the monoclonal antibody as the enzyme-labeled antibody,a double-antibody sandwich ELISA detection method for Tyrophagus putrescentiae was established,and its detection sensitivity and reliability in the culture medium and mycelium of Pleurotus ostreatus were verified.The results showed that the purity of the rabbit-sourced mite polyclonal antibody was over 85%,the concentration was 1.70 mg/mL,and the titer was approximately 870.4×103.One cell line(AntiTput-10)was screened from 38 hybridoma cell lines.The purity of the antibody after purification was over 90%,the concentration was 3.1 mg/mL,and the affinity constant was 2.71×1010.The concentration of the coating antibody was determined to be 100μg/mL and the dilution ratio of the enzyme-labeled antibody was 1∶2000 by the checkerboard titration method.The established ELISA detection method can detect 4 Tyrophagus putrescentiae mites per gram of edible fungus culture medium,which has high sensitivity.The proposed method could be applied to mite detection in different scenarios.

Abstract:To address the issue that low-order non-symplectic algorithm easily induced pseudo chaos when simulating the dynamical evolution of strings near black hole event horizons,the construction of high-precision explicit symplectic algorithm was proposed,taking the five-dimensional anti-de Sitter-Schwarzschild(AdS5-Schwarzschild)black hole background as an example.The Hamiltonian of the system was decomposed into five integrable subparts,based on which two types of fourth-order explicit symplectic iterative schemes were designed.The successfully constructed explicit symplectic algorithm not only strictly preserved the symplectic structure of the system but also effectively suppressed the long-term truncation errors accumulated by low-order non-symplectic algorithm,significantly improving accuracy and computational efficiency.Leveraging the high-precision and long-term stable numerical solutions provided by the explicit symplectic algorithm,the analysis of Poincaré sections and Lyapunov exponent diagrams revealed that all stable orbits in the system exhibit ordered motion with no chaotic behavior.The numerical experiments aligned with theoretical analysis,demonstrating that the explicit symplectic algorithm accurately described the long-term dynamical evolution of strings in strong gravitational fields.This research provided a novel pathway for constructing explicit symplectic algorithms in strong gravitational astrophysical systems,holding significant practical implications.

Abstract:Aiming at the accuracy degradation problem caused by modal feature mismatch and thermal imaging overexposure noise in the semantic segmentation of red-green-blue-thermal(RGB-T)images,a cross-modal interaction and dynamic fusion for RGB-T image semantic segmentation(CIDF)model was proposed to realize efficient co-sensing of RGB-T modalities.Firstly,a cross-modal skip connection(CMSC)mechanism was designed to establish multi-scale feature interactions between RGB-T modalities at the encoding stage,and effectively mitigate the inter-modal feature distribution differences through cross-layer feature transfer and adaptive aggregation strategies.Secondly,the multi-modal convolution fusion module(MCFM)was proposed,where dynamic feature preference was achieved through convolution,and the overexposure noise interference in the thermal imaging modality was suppressed.The experiments were based on the publicly available multi-spectral image fusion net for urban road scenes(MFNet)dataset and a self-constructed dataset of photovoltaic inspection scenes.The results showed that the average accuracy of CIDF model reached 72.9%and the mean intersection over union of CIDF model reached 56.6%on the MFNet dataset.In addition,average accuracy and mean intersection over union on the photovoltaic inspection dataset reached 98.9%and 97.4%,respectively.The research results not only confirmed the effectiveness of the CIDF model,but also highlighted its superior performance in cross-scenario applications.

Abstract:To address the issue of limited generalization in traditional object detection models due to scarce complex defect samples and diverse fault types in the power industry,the application status of few-shot object detection(FSOD)models across the five stages of power industry,namely,power generation,transmission,transformation,distribution,and consumption was systematically studied,and corresponding development trends were analyzed.It was found that FSOD models exhibited excellent performance in defect detection,fault detection,equipment monitoring,and safety inspection through strategies such as meta-learning,transfer learning,data augmentation,and metric learning.However,challenges were encountered in model structure and perception under complex backgrounds,generalization and data fusion,and sample dependency and system engineering,necessitating further optimization.FSOD models could have development in model structure improvements,multimodal data fusion,and domain knowledge integration with system deployment optimization.This study provided an important reference for the deepen application of FSOD models in the power industry.

Abstract:To explore the spatiotemporal pattern of carbon balance amounts and the characteristics of the urban network structure in the Pearl River Delta urban agglomeration,remote sensing data of the Pearl River Delta urban agglomeration from 2001 to 2021 were used to calculate the carbon balance amounts.Spatial autocorrelation and social network analysis methods were then employed to reveal the spatiotemporal distribution and inter-city network structure of carbon balance amounts.The results showed that the carbon balance amounts in the Pearl River Delta urban agglomeration exhibited a declining trend from 2001 to 2021,mainly due to a significant increase in carbon emission amounts,while carbon sequestration amounts changed only slightly,and there was a significant difference in the decrease among cities.Significant spatial heterogeneity of carbon balance amounts was observed,which generally followed a ″high on the periphery,low in the center″ pattern,with spatial autocorrelation dominated by ″high-high clusters″ and ″low-low clusters″.The carbon balance amounts spatial network structure of the Pearl River Delta urban agglomeration was found to be relatively stable.Foshan and Zhaoqing were located at the network core and classified as net beneficiary nodes;Guangzhou and Dongguan functioned as intermediary nodes facilitating resource circulation;the remaining cities acted as bidirectional spillover nodes,guiding carbon balance amounts flow across the cities.This study provided scientific support for establishing cross-regional ecological compensation mechanisms and formulating differentiated carbon reduction strategies.

Abstract:To address the limitations in spatiotemporal feature modeling and the insufficient capture of temporal dependencies in ozone mass concentration prediction,a multi-scale graph spatial feature extraction-temporal attention enhanced gated recurrent unit(MSGSFE-TAGRU)prediction model was proposed.The model consisted of two main modules:MSGSFE and TAGRU modules.In the MSGSFE module,multi-scale graph structures were constructed by integrating graph attention network(GAT)and graph convolutional network(GCN)to effectively extract both local and global spatial features.In the TAGRU module,a temporal attention mechanism was introduced to dynamically focus on key historical time steps,which enhanced the capability of the gated recurrent unit(GRU)in modeling for temporal dependencies.The results demonstrated that,in short-term prediction,the MSGSFE-TAGRU model reduced root mean square error and mean absolute error by 5.22%and 5.76%,respectively,compared to the best-performing graph neural network-GRU(GNN-GRU)model.In medium-and long-term prediction,the proposed model continued to exhibit superior accuracy and stability,validating its effectiveness in spatiotemporal feature modeling and generalization.This study provided a novel methodological framework for high-precision prediction of ozone mass concentration and air quality management.

Abstract:To address the issues of disconnection between teaching content and engineering requirements,monotonous teaching methods,and imperfect evaluation systems in the data structure course for electronic information-related majors,a teaching reform was implemented based on the outcome based education and conceive-design-implement-operate(OBE-CDIO)educational concept,which established a closed-loop instructional design framework,introduced a modular knowledge system,and developed a multidimensional assessment mechanism.The results indicated that significant improvements in students'engineering practical abilities and notable increases in both the excellence rate of final evaluations and the achievement levels of course objectives compared to the pre-reform period were observed after two years of reform and practice.This research provided a replicable implementation pathway for teaching reform in courses of electronic information-related majors.

Abstract:In order to screen out poplar(Populus L.)cultivars suitable for afforestation in arid areas of Hubei Province,eight main planted cultivars in Jianghan Plain,namely,P.euramericana'Nanlin 895',P.deltoides'Danhong',P.deltoides'Huashi 1',P.deltoides'Huashi 2',P.euramericana'I-214',P.deltoides'Harvard',P.deltoides'Zhongqian 3',and P.deltoides'Zhongshi 8'were selected,and their growth and physiological indicators were determined through drought stress experiments.The results showed that all poplar cultivars presented with growth inhibition and decreased photosynthetic efficiency.Among them,P.deltoides'Harvard'was the most severely affected,while P.deltoides'Huashi 2'was the least affected.The cluster analysis indicated that P.deltoides'Huashi 1'and P.deltoides'Huashi 2'had good drought resistance performances.The membership function analysis showed that the drought resistance ranking of the poplar cultivars was P.deltoides'Huashi 2'>P.deltoides'Danhong'>P.euramericana'Nanlin 895'>P.deltoides'Zhongqian 3'>P.deltoides'Huashi 1'>P.deltoides'Zhongshi 8'>P.euramericana'I-214'>P.deltoides'Harvard'.Therefore,P.deltoides'Huashi 2'had better drought resistance and was recommended for afforestation in arid areas of Hubei Province.It was advisable to avoid planting P.euramericana'I-214'and P.deltoides'Harvard',which had weaker drought resistance.This research could provide a reference for the selection of poplar afforestation cultivars in the arid areas of Hubei Province.

Abstract:To address the challenges of inconspicuous features and significant scale variations in steel surface defects that make detection difficult,a global feature interaction fusion-you only look once(GFIF-YOLO)model was proposed for steel surface defect detection.Firstly,the feature pyramid network(FPN)in the YOLO version 8 nano(YOLOv8n)was replaced with a gather-and-distribute(GD)mechanism,which enhanced feature fusion capability while mitigating information loss during cross-layer transmission.Secondly,a global attention mechanism(GAM)was embedded into the final layer of the backbone network to strengthen the ability of global feature representation.Finally,an efficient multi-scale convolution head(EMSCHead)module was introduced to reduce model parameters and computational costs while improving detection accuracy.The results demonstrated that the GFIF-YOLO model achieved mean average precision of 82.0%and 66.7%on the Northeastern University detection(NEU-DET)and the generic component 10-class detection(GC10-DET)dataset,respectively,outperforming the baseline YOLOv8n by 3.0 and 3.4 percentage points.The GFIF-YOLO model showed good performance and could effectively complete the detection task of steel surface defects.

Abstract:To address the low detection accuracy,difficulty in recognition,and large parameter size of models for surface defect detection in vertical shaft guide rails,a multi-enhanced lightweight efficient-you only look once version 11 nano(MELE-YOLOv11n)model was proposed.Firstly,the efficient channel attention dual-stream(ECA-DS)and the depthwise separable convolution-efficient multi-scale attention(DWEMA)modules were designed to enhance the model's adaptability to target defect features in complex mine environments and improve detection capability.Secondly,the convolutional three-scale kernel-adaptive dual-path-omni-kernel(C3K2-OK)module was introduced to capture feature map information at different scales,alleviating information loss.Finally,to address the high computational load of the original detection head,the lightweight shared detail-enhanced convolutional detection head(Detect-LSDECD)module was developed,where shared strategies were combined with detail-enhanced convolution to enhance lightweight performance.The results demonstrated that compared to the YOLOv11n model,the mean average precision of MELE-YOLOv11n model on vertical shaft guide rails surface defect dataset increased by 2.5%,and the number of parameters reduced by 0.3×106.The MELE-YOLOv11n model met the balance between accuracy and lightweight requirements,providing strong technical support for the automated detection of surface defects in vertical shaft guide rails.

Abstract:To address the difficulty in coal flow detection caused by strong dust interference and complex illumination in underground coal mines,a real-time segmentation model of underground coal flow based on the coal flow U-shaped network(CFU-Net)was developed.Firstly,the backbone network structure of U-Net was improved based on the lightweight mobile network version 4(MobileNetV4),and the operational efficiency of the model was thereby enhanced.Secondly,a lightweight multi-scale feature preservation(LMFP)module and a dynamic sampling(DySample)operator were employed to compensate for feature loss caused by channel compression and to strengthen the model's detail reconstruction capability.Finally,a dynamic feature interaction(DFI)module was introduced to refine the skip connections of U-Net.Through this module,adaptive deep fusion of cross-layer features was realized,and the model's perception capability for features at different scales was enhanced.The results showed that the mean intersection over union(mIoU)of the CFU-Net model was 95.10%,with the detection speed increased to 70.20 frames/s,thus possessing the advantages of high accuracy and high speed.When compared with mainstream models such as the pyramid scene parsing network(PSPNet)and the deep dual-resolution network(DDRNet),significant advantages were demonstrated.The research confirmed that real-time segmentation model of underground coal flow based on CFU-Net could satisfy coal flow detection requirements in real situations.

Abstract:To address the limitations of self-attention-based Transformer models in sequential recommendation tasks,including insufficient dynamic interest capture and quadratic computational complexity growth with sequence length,a sequential recommendation model based on Mamba2 and adaptive time-frequency analysis(M2ATFSRec)was proposed.The model was designed to enhance dynamic interest modeling capability while reducing computational complexity and improving recommendation accuracy.Firstly,adaptive time-frequency analysis was employed to extract time-frequency features from user historical behavior sequences,explicitly encoding multi-scale periodic patterns of interests.Secondly,Mamba2's selective state space mechanism was utilized to achieve efficient dynamic interest modeling for long sequences.The M2ATFSRec was experimentally evaluated on three datasets,namely,the movielens 1 million ratings(MovieLens-1M),the Amazon beauty products(Amazon-Beauty),and the Amazon video games(Amazon-Video-Games).In terms of the normalized discounted cumulative gain(NDCG)metric,it was found that M2ATFSRec achieved improvements of 6.42%,22.76%,and 33.22%respectively compared to towards efficient sequential recommendation with selective state space model(Mamba4Rec).The model had better recommendation performance in long sequence scenarios.

Abstract:To address the challenges of excessive parameter size and training difficulties caused by complex neural network architectures in image super-resolution(SR)tasks,a lightweight image SR reconstruction model based on improved multi-scale feature pyramid network(MFPNet)was proposed.First,an improved MFPNet architecture was designed.The feature representation spaces at different scales were constructed through iterative downsampling operations,which effectively enhanced the network's ability to capture multi-granularity detail features of images.Second,position aware circular convolution(ParC)was adopted as the primary feature extraction module,reducing the parameter count while expanding the network's receptive field size.Finally,a dynamic attention block(DAB)was developed.Through an attention guidance layer(AGL),the weighting of efficient channel attention(ECA)and spatial attention(SA)modules was dynamically adjusted,improving the network's capability for restoring texture details.The experimental results demonstrated that,compared with other state-of-the-art models,the structural similarity index measure(SSIM)maximum of 0.9613 and the peak signal-to-noise ratio(PSNR)maximum of 38.11 dB were achieved by improved MFPNet model.This research confirmed that the improved MFPNet model could be used to image reconstruction tasks with more natural detail textures.

Abstract:To explore the coexistence mechanisms between Sus scrofa and Elaphodus cephalophus distributed in the same region,Sus scrofa and Elaphodus cephalophus in the Hubei Mulinzi National Nature Reserve were monitored using infrared camera technology from January to December in 2020,and the data were analyzed using kernel estimation and overlap calculation methods.The results indicate:①There were differences in the daily activity rhythms of Sus scrofa during the rainy and dry seasons.In the dry season,Sus scrofa exhibited a bimodal activity pattern,with activity intensity peaks occurring between 02:00-04:00 and 17:00-19:00.During the rainy season,they showed a multimodal activity pattern,with the main activity intensity peaks occurring between 00:00-06:00 and 18:00-20:00,and secondary activity intensity peaks between 07:00-08:00 and 09:00-12:00.Compared to the dry season,the rainy season showed not only an increase in the number of activity intensity peaks but also an enhancement in activity intensity.② Also,there were differences in the daily activity rhythms of Elaphodus cephalophus during the rainy and dry seasons.Both seasons exhibited the multi-peak activity patterns.There were the main activity intensity peaks occurring between 08:00-10:00 and 18:00-19:00,and the secondary activity intensity peaks between 00:00-01:00 and 02:00-04:00 in the rainy season.The main activity intensity peaks in the dry season were concentrated between 00:00-04:00 and 18:00-20:00,with secondary activity intensity peaks between 06:00-08:00 and 12:00-14:00.Compared to the rainy season,the dry season showed an enhancement in activity intensity.③ The overlap of activity patterns between Sus scrofa and Elaphodus cephalophus was higher during the rainy season and lower during the dry season.The two species achieved resource sharing and avoided competition by adjusting their activity times.This study provided a theoretical basis and scientific support for the protection and management of co-occurring species within the Hubei Mulinzi National Nature Reserve.

Abstract:To address the problem of how to leverage the advantages of quantum computing to perform practical key recovery attacks on block ciphers,the required quantum resources for key recovery attacks were optimized based on Grover algorithm,building upon improvements to the quantum circuit of the Yo-yo block cipher.Firstly,based on the substitution(S)box lookup table,the S box quantum circuit was successfully implemented using the Dorcis tool.Secondly,by directly substituting input variables,swap gates in the S box quantum circuit were eliminated.Thirdly,this method was applied to the encryption and key expansion algorithms to remove permutation operations.Fourthly,by backtracking the ciphertext from the second round of iteration,the value obtained in the first round was used to construct the target function for the Grover algorithm,thereby avoiding the second round of iteration.Finally,the Grover algorithm was applied to perform quantum key recovery attacks on the Yo-yo block cipher,successfully retrieving the correct key.The correctness of the quantum circuit was verified using the Qiskit Aer quantum simulator.The results showed that,compared with the Vu method(VM),the optimized Grover algorithm attacking the quantum circuit of Yo-yo reduced the consumption of controlled-not(CNOT)gates,not(NOT)gates,Toffoli gates and swap(SWAP)gates by an average of 22%,13%,33%and 68%respectively.The algorithm effectively reduced the quantum implementation cost,thus reducing the resource consumption required for quantum key recovery attacks.

Abstract:To address the issues of sparse lesion distribution,variable morphology,and small inter-class differences in pneumoconiosis X-ray chest films,a dual-branch pneumoconiosis staging model based on dynamic convolution and Mamba(DC-Mamba)was proposed.First,the model enhanced the extraction of local features of small fibrotic lesions through the adaptive kernel parameter generation strategy of the dynamic convolution branch.Meanwhile,the global spatial dependencies of multi-regional lesions were captured by leveraging the sequential modeling capability of the Mamba branch.Second,a feature fusion module with dual-path attention collaboration mechanism was designed to integrate local details and global contextual information.The model was validated on 1760 real anonymized patient X-ray chest films.The results showed that the accuracy of DC-Mamba model reached 78.3%,the recall was 79.0%,and the F1 score was 77.6%,all of which were superior to contrast models.DC-Mamba significantly improved the model's understanding of the overall distribution of pulmonary lesions and the detection of subtle lesions,thereby enhancing the early screening and precise staging capacities of pneumoconiosis.

Abstract:To analyze the codon usage bias of the chloroplast genome of Cymbopogon citratus(DC.)Stapf and its evolutionary driving factors,the codon patterns were systematically analyzed,providing a basis for subsequent research on the functions of the genome and the development of germplasm resources.The complete sequence of the chloroplast genome of Cymbopogon citratus was obtained from the national center for biotechnology information(NCBI)database,and 50 protein-coding sequences(CDS)were screened.Combined with methods such as effective number of codons plot(ENC-plot),parity rule 2 plot(PR2-plot),and neutrality plot,the codon usage bias and influencing factors were explored.The overall guanine and cytosine(GC)content of the chloroplast genome of Cymbopogon citratus was 38.70%,and the average GC content at the third base position(GC3)was 29.60%,both of which were lower than those of other species in the same family.The average value of the codon adaptation index(CAI)was 0.17,and the average value of the effective number of codons(ENC)was 48.62,indicating that the codon usage bias was weak.The 30 frequently used codons were screened out,among which 26 were highly expressed codons.Furthermore,12 optimal codons with both high frequency and high expression density were determined,such as leucine and alanine.This study indicated that natural selection was the main driving force for the formation of the codon usage bias of the chloroplast genome of Cymbopogon citratus.