期刊：
International Journal of Reasoning-based Intelligent Systems,2025年1(1):66-72 ISSN：1755-0556
作者机构：
[Liping Zhong] Library, The Hunan City University, Yiyang, 413000, China
关键词：
binary sort tree;BST;E-book resources;search model;Vector space model;similar distance
摘要：
In order to solve the problems of low classification accuracy and large retrieval error in the classification and retrieval of e-book resources, a classification and retrieval method of e-book resources based on binary sorting tree is designed. Calculate the similar distance between e-book resource data and realise the structural analysis of e-book resource platform. The vector space model is introduced to realise the data integration of e-book resources. The retrieval model is constructed and the stability coefficient is introduced to realise the research. The experimental results show that the classification accuracy of the proposed method is about 99%, the retrieval error is always less than 1%, and the time cost is less than 2 seconds, which has certain advantages.
摘要：
Plant sterilants are used to control rodent populations due to their minimal environmental risk and other ethical considerations. However, their practical utilization is unsatisfactory due to high costs and processing difficulties. Broussonetia papyrifera is a plant material that has shown the potential to inhibit the reproduction of Microtus fortis, a species that causes serious damage to crops in the Dongting Lake region in China. M. fortis was treated with different doses of B. papyrifera leaf methanol extracts. The results show that the growth of sex organs was inhibited, and the males' testosterone levels and sperm quality were reduced. Though there were some positive effects on females, the reproductive parameters of coupled voles were inferior; the most treated couple exhibited an increased reproductive time, fetal counts, and reduced weight. It was also found that M. fortis responded negatively to the extract after a single treatment or long-term repeated treatment compared to a short-term repeated treatment. B. papyrifera leaves showed a higher application potential as a sterilant for male rodents. These findings enrich the study of plant sterilants and provide insights into the utilization of B. papyrifera and the management of rodents. Owing to the effectiveness and accessibility of the leaves, the derived sterilant may be more economical for controlling rodent pests.
摘要：
INTRODUCTION: University campuses, with their abundant natural resources and sports facilities, are essential in promoting walking activities among students, faculty, and nearby communities. However, the mechanisms through which campus environments influence walking activities remain insufficiently understood. This study examines universities in Wuhan, China, using crowdsourced data and machine learning methods to analyze the nonlinear and interactive effects of campus built environments on exercise walking. METHODS: This study utilized crowdsourced exercise walking data and incorporated diverse campus characteristics to construct a multidimensional variable system. By applying the XGBoost algorithm and SHAP (SHapley Additive exPlanations), an explainable machine learning framework was established to evaluate the importance of various factors, explore the nonlinear relationships between variables and walking activity, and analyze the interaction effects among these variables. RESULTS: The findings underscore the significant impact of several key factors, including the proportion of sports land, proximity to water bodies, and Normalized Difference Vegetation Index NDVI, alongside the notable influence of six distinct campus area types. The analysis of nonlinear effects revealed distinct thresholds and patterns of influence that differ from other urban environments, with some variables exhibiting fluctuated or U-shaped effects. Additionally, strong interactions were identified among variable combinations, highlighting the synergistic impact of elements like sports facilities, green spaces, and waterfront areas when strategically integrated. CONCLUSION: This research contributes to the understanding of how campus built environments affect walking activities, offering targeted recommendations for campus planning and design. Recommendations include optimizing the spatial configuration of sports facilities, green spaces, and water bodies to maximize their synergistic impacts on walking activity. These insights can foster the development of inclusive, health-promoting, and sustainable campuses.
摘要：
In order to overcome the problems of high time-consuming and poor recognition accuracy of learning behaviour recognition, this paper proposes an abnormal learning behaviour recognition method for MOOC online education based on background subtraction. Firstly, the characteristics of students' abnormal learning behaviours are collected and extracted. Then, the background difference algorithm is used to obtain the foreground object and background of the learning image, and the image pixels are classified. Finally, the mean background method is used to obtain the learning background, the abnormal behaviour recognition classifier is designed, and the background subtraction method is used to realise the abnormal learning behaviour recognition. The results show that the recognition accuracy of this method is as high as 98.32%, the recognition time is only 0.52 s, and the recognition recall rate is as high as 96.7%, indicating that this method can improve the recognition effect of abnormal learning behaviour.
摘要：
With the development of artificial intelligence, there has been a wave of innovation in the field of music creation and performance. In this paper, we propose a novel generative adversarial network architecture, which can generate adaptive piano accompaniment in real time. In the network, the generator generates piano accompaniment based on the input melody, and the discriminator evaluates the quality difference between the generated accompaniment and the professional accompaniment, and improves the quality of the accompaniment through the adversarial learning mechanism. The network design focuses on harmonizing the accompaniment with the main melody, introduces music theory knowledge constraints such as harmonic acoustic principles, and uses recurrent neural networks to capture the temporal characteristics of music, so that the rhythm and dynamics of the accompaniment change synchronously with the main melody. To validate the model, we collected piano repertoire of different styles and conducted extensive experiments. The results show that the generative adversarial network has good adaptability, can automatically adjust the accompaniment style and complexity according to the melody characteristics, and the generation speed is up to 5 seconds per song, and the adaptability is up to 90%, which is significantly better than the traditional method. User studies have shown that this network-generated accompaniment is audibly indistinguishable from that performed by a professional pianist, highlighting its potential for practical applications.
With the development of artificial intelligence, there has been a wave of innovation in the field of music creation and performance. In this paper, we propose a novel generative adversarial network architecture, which can generate adaptive piano accompaniment in real time. In the network, the generator generates piano accompaniment based on the input melody, and the discriminator evaluates the quality difference between the generated accompaniment and the professional accompaniment, and improves the quality of the accompaniment through the adversarial learning mechanism. The network design focuses on harmonizing the accompaniment with the main melody, introduces music theory knowledge constraints such as harmonic acoustic principles, and uses recurrent neural networks to capture the temporal characteristics of music, so that the rhythm and dynamics of the accompaniment change synchronously with the main melody.
To validate the model, we collected piano repertoire of different styles and conducted extensive experiments. The results show that the generative adversarial network has good adaptability, can automatically adjust the accompaniment style and complexity according to the melody characteristics, and the generation speed is up to 5 seconds per song, and the adaptability is up to 90%, which is significantly better than the traditional method. User studies have shown that this network-generated accompaniment is audibly indistinguishable from that performed by a professional pianist, highlighting its potential for practical applications.
作者机构：
[Yao, Qi; Ren, Heng] Hunan City Univ, Sch Civil Engn, Yiyang 413000, Peoples R China.;[Wei, Mingxing; Li, Peng; Ren, Heng] Hunan Univ Sci & Technol, Coll Resources Environm & Safety Engn, Xiangtan 411201, Peoples R China.;[Zhu, Yongjian] Key Lab Gas & Roof Disaster Prevent & Control Sout, Xiangtan 411201, Peoples R China.
通讯机构：
[Ren, H ] H;Hunan City Univ, Sch Civil Engn, Yiyang 413000, Peoples R China.;Hunan Univ Sci & Technol, Coll Resources Environm & Safety Engn, Xiangtan 411201, Peoples R China.
关键词：
Excavation disturbance;Weakening of damage;Rock-anchor combination;Failure mode;Stress intensity factor
摘要：
In response to the insufficient research on the damage weakening evolution law of surrounding rock during roadway (tunnel) excavation and retreat, as well as the inadequate understanding of the anchoring mechanism of rock bolts in such rock masses, a series of triaxial loading-unloading and uniaxial reloading experiments were conducted on sandstone specimens using the RMT-150C rock mechanics testing system. Microstructural characteristics of the damaged rock mass were investigated through scanning electron microscope (SEM) analysis, revealing the damage weakening mechanism of the rock. Through laboratory experiments and theoretical analysis, in combination with the failure characteristics of rock-anchor composite bodies, the underlying causes of changes in failure modes were analyzed to explore the anchoring mechanism of bolts in damaged rock. The experimental results indicate that during the triaxial loading-unloading tests, the degree of rock damage varied depending on the axial unloading point, exhibiting a trend of first decreasing and then increasing. For damaged rock masses, the presence of bolts did not alter the failure modes, which manifested as splitting failure at low damage levels and shear slip failure at high damage levels. However, the bolts did influence the characteristics of the failure, with the failure features of damaged rock being more diverse under the influence of bolts. The failure mode of the rock is related to the degree of damage, which can be represented by the ratio η of wing crack length to main crack length. When η is large, splitting failure occurs, whereas smaller η leads to shear slip failure. Bolts can reduce the stress intensity factor at the crack tip to some extent, inhibiting the propagation of wing cracks. This study effectively explains, from a microscopic perspective, the relationship between the failure modes of rock-anchor composite bodies and the anchoring effect of bolts.
作者机构：
[Deng, Chao; Li, Liuxi; Xu, Zhichao; Zhou, Yi] Hunan City Univ, Hunan Engn Res Ctr Struct Safety & Disaster Preven, Yiyang 413000, Peoples R China.;[Deng, Chao] Hunan City Univ, Hunan Engn Res Ctr Dev & Applicat Ceramsite Concre, Yiyang 413000, Hunan, Peoples R China.;[Hu, Huanxiao] Cent South Univ, Sch Geosci & Info Phys, Key Lab Metallogen Predict Nonferrous Met & Geol E, Minist Educ, Changsha 410083, Peoples R China.;[Yin, Quan] NYU, Dept Civil & Urban Engn, Brooklyn, NY USA.;[Chen, Juan] Hunan Univ Sci & Technol, Coll Elect & Informat Engn, Xiangtan 411201, Hunan, Peoples R China.
通讯机构：
[Deng, C ] H;Hunan City Univ, Hunan Engn Res Ctr Struct Safety & Disaster Preven, Yiyang 413000, Peoples R China.;Hunan City Univ, Hunan Engn Res Ctr Dev & Applicat Ceramsite Concre, Yiyang 413000, Hunan, Peoples R China.
摘要：
The development and modification of grouting materials constitute crucial factors influencing the effectiveness of grouting. Given the pivotal role of water in the hydration of cement-based composite materials and construction processes, this study proposes an exploratory approach using green, economical magnetized water technology to enhance the performance of cement grouts. The research systematically investigates the effects of magnetized water on the fundamental grouting properties (stability, rheological behavior, and stone body strength) of cement grouts, prepared under varying magnetization conditions (including magnetic intensity, water flow speed, and cycle times). Through the conduct of specific physicochemical tests on water, the study elucidates the mechanism through which magnetized water influences these properties. The results indicate that magnetized water positively impacts the stability of cement grouts, significantly reducing their absolute viscosity, apparent viscosity, plastic viscosity, and yield stress, thus markedly affecting the rheological characteristics of the cement grouts. Additionally, magnetized water notably enhances the flexural and compressive strength of the cement grout stone body, with a particularly significant improvement in early strength. From a quantum mechanics perspective, a magnetization mechanism based on the competition between the strengthening of hydrogen bonds between water molecule clusters and the weakening or breaking of hydrogen bonds within clusters is introduced, providing a theoretical basis for explaining the variability observed in water magnetization experiments.
作者机构：
[Liu, Xiqun; Mo, Ping; Xu, Zhenggang; He, Xinwu; Huang, Tian; Liu, Jiajia; Li, Youwen] Hunan City Univ, Coll Informat & Elect Engn, Hunan Engn Res Ctr Ecol Environm Intelligent Monit, Yiyang 413000, Peoples R China.;[Xu, Zhenggang] Northwest A&F Univ, Coll Forestry, Yangling 712100, Peoples R China.;[Mo, Ping] Hunan Univ Arts & Sci, Coll Life & Environm Sci, Changde 415000, Peoples R China.
通讯机构：
[Huang, T ] H;Hunan City Univ, Coll Informat & Elect Engn, Hunan Engn Res Ctr Ecol Environm Intelligent Monit, Yiyang 413000, Peoples R China.
关键词：
bean goose;clustering algorithms;habitat identification;spatial-temporal random partitioning;time complexity
摘要：
With the acceleration of social development and urbanization, birds' natural habitats have been greatly disturbed and threatened. Satellite tracking technology can collect much bird activity data, providing important data support for habitat protection research. However, satellite data are usually characterized by discontinuity, extensive periods, and inconsistent frequency, which challenges cluster analysis. Habitat research frequently employs clustering techniques, but conventional clustering algorithms struggle to adjust to these data features, particularly when it comes to time dimension changes and irregular data sampling. T-DBSCAN, an enhanced clustering algorithm, is suggested to accommodate this intricate data need. T-DBSCAN is improved based on the traditional DBSCAN algorithm, which combines a quadtree structure to optimize the efficiency of spatial partitioning and introduces a convex hull algorithmic strategy to perform the boundary identification and clustering processing, thus improving the efficiency and accuracy of the algorithm. T-DBSCAN is made to account efficiently for the uniformity of data sampling and changes in the time dimension. Tests demonstrate that the algorithm outperforms conventional habitat identification accuracy and processing efficiency techniques. It can also manage large amounts of discontinuous satellite tracking data, making it a dependable tool for studying bird habitats.
作者机构：
[Ziwei Zhou] School of Civil Engineering, Hunan City University, Yiyang, Hunan 413000, China;Hunan Engineering Research Center of Development and Application of Ceramsite Concrete Technology, Hunan City University, Yiyang 413000, China;[Jianxin Peng] School of Civil Engineering, Changsha University of Science and Technology, Changsha, Hunan 410114, China;[Yiming Yang] School of Civil Engineering, Hunan City University, Yiyang, Hunan 413000, China<&wdkj&>Hunan Engineering Research Center of Development and Application of Ceramsite Concrete Technology, Hunan City University, Yiyang 413000, China
通讯机构：
[Jianxin Peng] S;School of Civil Engineering, Changsha University of Science and Technology, Changsha, Hunan 410114, China
摘要：
This study aims to propose a hybrid physics-based and data-driven model for predicting flexural capacity of corroded prestressed concrete (PC) structures, aiming to improve the rationality and applicability of flexural capacity assessment. First, a self-adaptive physics-informed deep neural network framework is developed. Then, based on a collected database containing 100 sets of measured data, the proposed model is compared with three deep learning (DL) models, two machine learning (ML) models and two traditional mechanism-based prediction models. After that, the score analysis, error analysis, and generalization performance evaluation are conducted. Finally, the importance analysis of input parameters is analyzed by using SHapley Additive exPlanations (SHAP) method. The results indicate that the proposed method significantly improves prediction accuracy compared to DL and ML models, with the mean absolute percentage error (MAPE) reduced by 28.88% compared to the second-best model. Compared to conventional mechanism-based models, the proposed method achieves substantial reductions in mean absolute error (MAE), root mean squared error (RMSE), and MAPE, with the MAPE reduced by over 83%. Moreover, the proposed method exhibits good generalization performance, with the average relative error between predicted and measured values being less than 6%. In addition, SHAP analysis reveals that the beam height, cross-sectional area of tensile non-prestressing reinforcement and prestressing reinforcement have a significant influence on the prediction results among all input parameters. These findings not only enhance the reliability of flexural capacity prediction for corroded PC structures but also provide a solid data foundation for similar research. The proposed hybrid framework also offers a new perspective for predicting other structural performance indexes.
This study aims to propose a hybrid physics-based and data-driven model for predicting flexural capacity of corroded prestressed concrete (PC) structures, aiming to improve the rationality and applicability of flexural capacity assessment. First, a self-adaptive physics-informed deep neural network framework is developed. Then, based on a collected database containing 100 sets of measured data, the proposed model is compared with three deep learning (DL) models, two machine learning (ML) models and two traditional mechanism-based prediction models. After that, the score analysis, error analysis, and generalization performance evaluation are conducted. Finally, the importance analysis of input parameters is analyzed by using SHapley Additive exPlanations (SHAP) method. The results indicate that the proposed method significantly improves prediction accuracy compared to DL and ML models, with the mean absolute percentage error (MAPE) reduced by 28.88% compared to the second-best model. Compared to conventional mechanism-based models, the proposed method achieves substantial reductions in mean absolute error (MAE), root mean squared error (RMSE), and MAPE, with the MAPE reduced by over 83%. Moreover, the proposed method exhibits good generalization performance, with the average relative error between predicted and measured values being less than 6%. In addition, SHAP analysis reveals that the beam height, cross-sectional area of tensile non-prestressing reinforcement and prestressing reinforcement have a significant influence on the prediction results among all input parameters. These findings not only enhance the reliability of flexural capacity prediction for corroded PC structures but also provide a solid data foundation for similar research. The proposed hybrid framework also offers a new perspective for predicting other structural performance indexes.
通讯机构：
[Yang, N ] H;Hunan City Univ, Coll Architecture & Urban Planning, Yiyang 413002, Peoples R China.;Key Lab Key Technol Digital Urban Rural Spatial Pl, Yiyang 413002, Peoples R China.;Key Lab Urban Planning Informat Technol Hunan Prov, Yiyang 413002, Peoples R China.
关键词：
production-living-ecological space (PLES);land-use transition;spatiotemporal pattern;driving force;Hunan Province
摘要：
China's rapid economic growth has increased tensions between production, living, and ecological spaces (PLES), making sustainable land-use planning difficult. Therefore, PLES evolution and processes are a focus of current research. Remote sensing data with land-use transition matrices, centroid migration, standard deviation ellipses, spatial autocorrelation, and geographic detectors were used to study the dynamics of PLES in Hunan Province from 1990 to 2020, elucidate its mechanisms and main influencing factors, and provide a comprehensive understanding of its evolutionary characteristics. The main conclusions of our analysis are as follows: (1) Ecological space was the dominant land-use type, while production space increased, putting strain on natural areas. (2) Living space increased by 40.73% over three decades, mostly comprising manufacturing space, highlighting urban expansion. (3) Despite land-use changes, Loudi City's PLES centroid remained central. (4) Standard deviation ellipses showed spatial shrinkage with directional stability, implying enhanced land usage within borders rather than outward growth. (5) The geographic detector analysis showed that the GDP, population density, slope, and elevation influenced these spatial changes. Economic prosperity drove urban expansion, but the slope and elevation limited development to accessible locations. These findings provide policymakers with essential information for balancing urbanization and ecological preservation and provide a case study for sustainable PLES design in rapidly developing regions.
