Abstract:

Abstract:

Abstract: In response to the problem of insufficient attention to the risk degree of potential inundation areas in thesiting of shelters for dam failure floods, a new model for evaluating the sites of shelters was proposed. On the basisof establishing the risk evaluation index system, the improved catastrophe theory evaluation method was used to evaluate the risk degree and classify the risk level of the potential inundation areas. The distance between sheltersand high-risk level areas was taken as the site selection evaluation index, and the disaster risk, location scale, emergency support of shelters were considered. The weights were determined by AHP method, and the TOPSISmethod was applied to evaluate the sites of shelters. Finally, 13 potential inundation areas and 10 alternative shelters downstream of Luhun Reservoir in China were used as examples for validation. The results showed that amongthe areas with high risk levels, Baiyuan Township was a Class Ⅰ hazardous area, and Pengpo Township, MinggaoTownship, Chengguan Street, and Longmen Township were Class Ⅱ hazardous areas. The better evaluated shelterscould meet the requirements of paying attention to areas with high risk degree.

Abstract: In order to analyze the needs of the elderly at the signalized intersection and evaluate the service level ofthe elderly pedestrians, a service level evaluation model based on subjective evaluation and intersection traffic characteristics was established. The traffic characteristic parameters of 33 signal-controlled crosswalks were obtainedthrough field observation and video analysis. A truncated questionnaire was used to obtain the subjective value ofpedestrian service level evaluation of the elderly crossing the street with synchronous conditions. The initial selection was based on nonlinear, linear and fuzzy linear regression models. Based on the data analysis of 30 crosswalks,Pearson correlation coefficient and Spearman correlation coefficient were used to identify the key factors affecting theservice level of the elderly crossing the street. Taking the key factors as input, comparing the nonlinear, linear andfuzzy linear regression models, the results showed that the multivariate nonlinear regression model of elderly pedestrian service level evaluation was better in terms of fit and error, and the data of 3 crosswalks in the verificationgroup further confirm the applicability of the model. The research results could provide some reference for the studyof traffic safety improvement strategies in the aging trend.

Abstract: In the phenomenological model, the sound absorption coefficient of porous materials was used to explaintakes into account the energy dissipation generated by the propagation of sound waves in the void structure. A modelbased on acoustic parameters was constructed to predict the sound absorption coefficient. In order to accurately obtain the five acoustic parameters ( porosity, flow resistance rate, tortuosity factor, viscous/ thermal characteristiclength) of OGFC asphalt mixture to construct the acoustic model, the OGFC asphalt mixture with different porosityand different gradation types was prepared by the combination of measurement and inversion, and the sound absorption coefficient was tested by standing wave tube. The flow resistance measurement equipment suitable for mixturewas developed. Based on the measured porosity, sound absorption coefficient and flow resistance, the inversionprogram was written based on genetic algorithm to invert the tortuosity factor and viscous/ thermal effect characteristic length of OGFC. Secondly, the finite element model of standing wave tube was established to verify the correctness of acoustic parameters. The influence of single factor of acoustic parameters on sound absorption performancewas analyzed. Finally, the sound absorption performance was optimized based on acoustic parameters. The resultsshowed that the higher the porosity of OGFC and the larger the nominal maximum particle size, the larger the average sound absorption coefficient and the peak sound absorption coefficient. The model constructed could better reflect the sound absorption characteristics of the mixture, and the peak sound absorption coefficient and the frequency of occurrence were consistent with the measured values. The increase of porosity, viscous characteristic lengthand thermal effect characteristic length, and the decrease of tortuosity factor would be beneficial to improve thesound absorption performance. The optimization results of the sound absorption performance of the mixture based onthe acoustic parameters showed that the porosity should be controlled at about 22% for the best sound absorptionperformance.

Abstract: Waste wood, crop residues, livestock manure and waste oil, were widely used to produce bio-oil. Andthen partially or completely replace the petroleum-based asphalt. But the road performance of this asphalt was insufficient. In this study, the development of green and sustainable bio-asphalt to partially replace petroleum-based asphalt with wood-based phenolic resin made by straw liquefaction products was proposed. Firstly, the separation ofstraw components was carried out to extract cellulose and lignin. Straw and its primary components were liquefied toexplore the main factors influencing the liquefaction of straw. Then the liquefied products were combined with formaldehyde to synthesize lignin-based phenolic resin WPR. And the phenolic resin PR, which was obtained from thereaction with phenol, was compared and analyzed. Finally, the resin was put into matrix asphalt with different ratios to prepare bio-asphalt, and the performance of the bio-asphalt was analyzed by the three major indexes of eachbio-asphalt and viscosity. The results showed that the liquefaction rate of straw mainly depended on the liquefactionrate of cellulose, and its liquefaction reaction was complex, with a reaction kinetic level of 1. 71. From the resinyield and liquefaction products, WPR, PR FT-IR spectral analysis showed that the WPR resin synthesis rate washigher and had a better reactivity than the PR. Only when mixed with the liquefaction products, the performance ofthe bio-asphalt prepared was poorer. However, when mixed with WPR resin, the bio-asphalt′s high temperaturestability, low-temperature cracking resistance, deformation resistance, and temperature stability were better thanmatrix asphalt.

Abstract: A monthly runoff prediction model( STL-VMD-SVM) based on a secondary decomposition using loess( STL) and variational mode decomposition (VMD) combined with a support vector machine( SVM) was proposed to address the nonlinear and non-stationary characteristics of runoff sequences. This model utilized STL to decompose the original runoff sequence into seasonal, trend, and residual terms of different frequencies and decomposedthe residual term into IMFs through VMD. An SVM model was established to predict seasonal, trend, and IMFs.The sum of the predicted values of all IMFs was the predicted value of the residual term, and the product of seasonal, trend, and residual terms was the final predicted value of the original runoff series. Based on the monthly runofftime series of Heishiguan Station and Gaocun Station on the mainstream of the Yellow River in the Yiluo River Basin, an example application and universality evaluation were conducted, and compared with the BP neural networkmodel and the long shortterm memory neural network model( LSTM) . The results showed that for the runoff prediction of Heishiguan Station in the Yiluo River Basin, the NSE, MAPE, RMSE, and R in the validation period of theproposed model were 0. 977, 13. 705%, 0. 327 and 0. 991, respectively, and their prediction accuracy was betterthan that of the single model and the primary decomposition model. The secondary decomposition of STL-VMDcould effectively improve the prediction accuracy of the model. The NSE, MAPE, RMSE, and R during the validation period in the runoff prediction at Gaocun Station on the mainstream of the Yellow River were 0. 979, 8. 509%,3. 263, and 0. 989, respectively, which also achieved good prediction results.

Abstract: Address issues such as the inadequate consideration of inter-feature correlations in existing intrusion detection methods and the need for improved detection accuracy on high-dimensional discrete datasets, a network intrusion detection method MBGAN based on spatial features and generative adversarial networks was proposed. Initially, a transformation approach was devised to convert one-dimensional data into two-dimensional grayscale images, enabling convolutional kernels to capture richer contextual information. Subsequently, a bidirectional generative adversarial network model was employed for anomaly detection. The model was trained using network traffic images, incorporating the minimum Wasserstein distance and gradient penalty techniques to mitigate mode collapseand instability during generative adversarial network training. Experimental verification showed that the detection accuracy of the proposed method on the NSL-KDD, UNSW-NB15 and CICIDIS2017 datasets was 97. 4%, 92. 3% and94. 8%, the recall rates were 97. 2%, 93. 1% and 95. 6%, and the F1 were 97. 3%, 93. 0% and 95. 2%, respectively, which were better than those of other methods.

Abstract: In the process of image super-resolution reconstruction, high frequency features might be ignored, whichwould lead to insufficient extraction features and fuzzy texture details in the reconstructed image. To solve this problem, an image super-resolution reconstruction network based on double feature extraction and attention mechanismwas proposed. In particular, in this study, a two-branch network for feature extraction was proposed to solve theproblem that high frequency features and multi-scale features could not be effectively extracted and uniformly fusedduring image reconstruction. In addition, in order to make the network obtain more accurate high-frequency features, a local spatial attention module was proposed, and combined with channel attention. A residual fusion attention module was constructed to improve the network′s ability to locate high-frequency features. Finally, the atrouspyramid module was designed to enlarge the receptive field of the network and enable the multi-scale feature extraction. Experiments were carried out on four benchmark datasets, and the results were better than the current advanced methods. Especially when the super-resolution multiple was 4, the proposed method improved the optimalPSNR by 0. 16, 0. 08, 0. 03 and 0. 20 dB, respectively, compared with the current mainstream models. The experimental results shown that the proposed method achieved better improvement in visual effect and quantitative analysis

Abstract: In order to address the difficulty of existing word-level semantic matching models in understanding sentence-level scientific dataset metadata, a sentence-level semantic matching ( CSDSM) model for Chinese scientificdatasets was proposed. The model used the CSL dataset to train and generate the CoSENT pre-training model basedon SimCSE and CoSENT. Building upon the CoSENT model, a multi-head self-attention mechanism was introducedfor feature extraction, and the final output was obtained by weighting the cosine similarity and KNN classificationresults. Experimental data from the National Earth System Science Data Center′s open semantic metadata information was used as a self-built scientific dataset. The experimental results showed that compared to the Chinese BERTmodel, the proposed model improved the Spearman′s ρ index by 0. 044 8, 0. 029 0, 0. 177 7 and 0. 050 9 on thepublic datasets AFQMC, LCQMC, Chinese-STS-B, and PAWS-X, respectively. Additionally, F1 and Acc on theself-built scientific dataset were improved by 0. 078 8 and 0. 063 4 respectively. The proposed model effectively addresses the problem of sentence-level semantic matching in scientific datasets.

Abstract: Aiming at the unloading problem of mobile tasks for the limited edge server resources to maximize thesatisfaction of numerous mobile tasks with deadline requirements, a model for cloud-edge-device collaboration wasproposed to offload mobile tasks. Firstly, the model analyze the factors that affect the service demand of mobiletasks and the service guarantee of virtual machines, and give the measurement method, as well as the measurementmethod of the service matching degree between mobile tasks and virtual machines. Secondly, a mobile task offloading strategy was designed for on-demand allocation of physical resources in a dynamic cloud-edge environment.Based on the improved Hungarian algorithm, the purpose of this strategy was to find an offloading plan that couldmaximize service matching for a batch of tasks, and to further optimize the offloading plan by eliminating resourcecompetition through a limited number of iterations. Finally, the algorithm in this study was compared with theP2PITS algorithm, the ALBOA algorithm and the ESSDSA algorithm from many aspects. Experimental resultsshowed that compared with the P2PITS algorithm, the algorithm in this study reduced the virtual machine load rateby 30. 1%, the average waiting time by 13%, compared with the ALBOA algorithm, the algorithm in this study reduce the average completion time by 38. 6% on average, compared with the ESSDSA algorithm, the algorithm inthis study increased the execution success rate by 3. 5% on average. The proposed algorithm could effectively improve resource utilization and reduce the average completion time of tasks while meeting user deadline requirements.

Abstract: For unlabeled networks, the link prediction method based on graph neural networks had poor performance when using its efficient modeling mechanism for link prediction tasks. An unsupervised link prediction algorithm (ALIP) was proposed. It could approximate the graph neural network framework to simulate the efficientmodeling mechanism and learning process of graph neural network algorithms, and to solve the problem of insufficient modeling caused by missing network node labels. Firstly, referring to the input layer of GCN, the structuralinformation and node attributes of the network were fused. Secondly, matrix factorization is used to replace the hidden layer of GCN and simulate forward propagation. Then the ideas of identity mapping and vector optimization toachieve vector transformation and model optimization to obtain the network node representation vector, which wereused to simulate the back propagation of GCN. Finally, the similarity matrix for performance evaluation of link prediction tasks was calculated. On the Citeseer dataset, DBLP dataset and Cora dataset, the experimental resultsshowed that ALIP algorithm had a maximum AUC value of 98. 01%, and its performance was superior to the other23 link prediction algorithms. The effectiveness and feasibility of the algorithm, in this study provide a new solutionfor complex unlabeled network link prediction tasks.

Abstract: Considering the factors such as time delay in the propagation of rumors and the inability to spread rumorsdue to the forced silence of network regulators, in this study, based on the SIR model, combined with the rumorrefuging mechanism and the space theory, nodes in the network were divided into susceptible node S, infectivenode I, rumor-refuging node C and recovered node R, a new SICR rumor propagation model was proposed. Firstly,the dynamic equation of rumor propagation in homogeneous network structure was given by means of average fieldtheory, the existence of equilibrium point was analyzed, and the basic reproduction number of the model was calculated by using the next generation matrix method. It was found that the basic reproduction number was related to thepropagation rate, average degree, migration rate, migration rate, forced silence rate, and recovery rate of infectivenodes. Secondly, the local asymptotic stability of the equilibrium point was analyzed by Routh-Hurwitz criterion,and the global asymptotic stability was analyzed by LaSalle′s invariance principle. Finally, the correctness of thetheoretical results was verified by numerical simulation experiments. The simulation results showed that SICR modelconsidering the rumor-refuting mechanism could suppress the rumor propagation better than SIR model. Based onDataset_R6 dataset, the parameters of the model were fitted by least square method, and the R2of the model was0. 950 8.

Abstract: In order to explore the key risk factors of high-rise building construction safety and the coupling effects ofrisk elements, the complex network model and the N-K model were combined, 158 high-rise building constructionaccident cases from 2015 to 2022 were selected according to the completeness of the accident case investigation report, and the causes of 84 of them were coded according to the grounded theory, and 5 risk elements including human, machine, material, pipe and environment and 23 risk factors were identified through statistical analysis. Thecoupling risk values of various coupling forms of risk elements were calculated according to the N-K model, and thedegree value distribution of each risk factor in the complex network model was obtained by using Ucinet software.The risk coupling network model of high-rise building construction was drawn by Netdraw, and the potential riskchain of nodes of 23 risk factors was analyzed, and the coupling form of potential risk chain and the coupling degreevalue of N-K model were combined to modify the standardized output. The remaining 74 accident cases were used toverify the results, and the verification results were roughly consistent with the previous 84 accident cases. The results showed that the greater the number of risk elements involved in coupling, the greater the risk coupling value,and the coupling form involving equipment elements had the largest coupling value. The lack of site management,the lack of safety education and training, the incompleteness of safety management system, inporper wearing of protective equipment, and the lack of investigation and management of hidden danger were the key risk factors thatneed to be focused on prevention and control.

Abstract: Based on the spatially-averaged large eddy simulation method, the wind interference effect of the tandemdouble hemispherical domes was numerically studied considering different center spacing. The effectiveness of thepresent numerical simulation method and parameter settings was firstly verified by comparison between results oflarge eddy simulation and the wind tunnel test on the tandem double hemispherical dome with center spacing of160 m. Then the wind pressure distribution characteristics on the tandem double hemispherical domes surface withthe center spacing of 160 m and 190 m were compared and analyzed to study the influence of the interferenceeffect. In conjunction with the simulated unsteady flow field, the mechanism of the influence on wind loads withdifferent center spacing was investigated. Similar tendency was observed for the distribution of the mean and fluctuating wind pressure coefficients of the tandem double spherical shell roof with different center spacing. The blockingeffect of the upstream roof weakened the positive pressure on the windward surface of the downstream roof. However, the blocking effect of the downstream roof would intensify the flow separation of the upstream roof when thespacing was small, resulting in a significant increase in the local wind suction on the top skylight. As the spacingincreased, the separated vortex on the leeward surface of the upstream roof could change from small-scale strip vortex to large-scale arc-shaped one, leading to more significant wind pressure fluctuations on the roof skylight, upstream leeward surface, and downstream windward surface.

Abstract: To address the issues of insufficient precision, inadequate representation of regional characteristics, andthe lack of consideration for geological spatial anisotropy in traditional geological modeling interpolation algorithms,in this study a three-dimensional geological interpolation method was proposed to take into account geological spatialanisotropy. Building upon the traditional IDW interpolation, this method involved constructing virtual boreholes,simulating the creation of an original stratum point set, and adjusting the parameters of the IDW interpolation ellipsesearch range to ultimately achieve an adaptive transformation of the optimal search parameters. Additionally, in thisstudy the concept of " directional marker points" was introduced to achieve a secondary optimization of the searcharea through " expansion" and " collapse" . Geological borehole datas from the Zhengzhou Urban Active Fault Detection Project, were used to analyze the regional adaptability of each algorithm layer by layer, implement the implicit surface representation of the three-dimensional geological model, and conduct cross-validation with commonlyused interpolation algorithms. The experiment results demonstrated that the proposed method possessed good regional adaptability and could effectively represent the anisotropic characteristics of urban underground spatial structures.While ensuring interpolation accuracy, it could effectively expresse the actual extension of strata within the region.

Abstract: At present, the manual adjustment of hyper-parameter for current wind power prediction model was slowand unreliability. In order to achieve the prediction effect, the model used in wind power prediction needs to selectthe appropriate hyper-parameters for the model. Based on this, in this study, a multi-unit wind power predictionmodel was proposed based on long short-term memory ( LSTM) . Firstly, the Spearman correlation method was usedto quantitative analysis. Secondly, the principal component analysis ( PCA) was used to reduce the dimension ofthe input features as well as extract the key information. In addition, considering the difficulty of choosing parameters for LSTM, in this study, particle swarm optimization ( PSO) algorithm was used to optimize the number of hidden layer neurons in each layer of LSTM. For the problem of wind power prediction of multiple units, in this study,a single wind turbine was used to find the most excellent model in a single unit, and applied the prediction model tomulti-unit prediction. Experiments showed that compared with other models, the root mean square error of the proposed method was reduced by 11. 8%, and the mean absolute error was reduced by 5. 03%.

Abstract: The safe operation of smart grid was the primary premise to ensure continuous and efficient power supply. Therefore, a graph neural network (GNN) based power system operation state analysis and evaluation modelwas proposed. Firstly, long short-term memory network was used to fill missing data, to ensure that the model hadgood performance in stability assessment and fault location. Secondly, a binary classifier for evaluating the stablestate of power grid operation and a multi classifier for locating faulty components were designed based on GNN. Dueto the ability of the proposed model to fully explore the spatiotemporal characteristics of power grid operation data,the proposed model exhibited superior performance compared to other methods under different measurement conditions. Experimental results showed that when the time series length of data was 0. 1 seconds, the stability assessment and fault location accuracy of the proposed model were 0. 985 5 and 0. 981 4, respectively, and higher thanthe comparative models. When only half of the component data can be measured, the accuracy of the proposedmodel for stability assessment, bus fault location, and generator fault location were 0. 998 0, 0. 960 9, and0. 981 2, respectively, and higher than the comparative models.

Abstract: In order to improve the power generation efficiency of the gas turbine,and to solve the problem of thehigh cost of the operation and maintenance (O&M) of the compressor water washing system, a digital twin-basedO&M decision study of the compressor was conducted. And a health management framework for gas turbine in power plants based on digital twin was proposed. Based on that the compressor operation data was processed, and theextreme gradient boosting algorithm was used to build a prediction model for the washing cycle, and some of the parameters within the dataset were selected as inputs to the model, with the gas consumption as the output quantity,analyzed the change rule and its relationship with the input quantity. the water washing cycle and water washing recovery rate were calculated and compared, and the appropriate water washing cycle for O&M decision of the compressor was derived. The model prediction results showed that: the average R2_score of the eight water washing gasconsumption prediction reached 0. 98, and the prediction results were accurate. Among the eight times of waterwashing, the second and third water washing cycles were appropriate, and the third water washing recovery rate wasoptimal, resulting in the guiding hours of gas turbine compressor water washing cycle of 1 824 hours. Comparedwith the average water washing cycle of the power plant in the actual implementation, the cost of water washingcould be reduced by 21. 9 million yuan per time.

Abstract: In order to investigate particulate matter ( PM) emission characteristics from the combustion of commercial biomass pellet, a fixed-bed reactor was used to conduct combustion experiments of wood dust, cotton stalk andbamboo dust. The particle size distributions and main element composition of PM were analyzed. And the influenceof element content on PM emission was discussed. It was found that the yields of PM10from high to low was cottonstalk, wood dust and bamboo dust, and the yields were 27. 76, 20. 83 and 9. 65 mg / m3, respectively. The PMswere mainly composed of submicron particles ( PM1 ) , and the proportion of PM1to PM10 was more than 90%. PM1was mainly composed of alkali metal chloride and sulfide, while PM1-10 was mainly composed of compounds formedby calcium magnesium silicate. Correlation analysis showed that there was a positive correlated between biomass ashcontent and PM1yield, while the content of Mg +Ca and n ( Mg +Ca) / n ( Si) were linearly correlated with PM1-10 yield.