As a fundamental component in computer vision,edges can be categorized into four types based on discontinuities in reflectance,illumination,surface normal,or depth.While deep CNNs have significantly advanced generic e...As a fundamental component in computer vision,edges can be categorized into four types based on discontinuities in reflectance,illumination,surface normal,or depth.While deep CNNs have significantly advanced generic edge detection,real-time multi-class semantic edge detection under resource constraints remains challenging.To address this,we propose a lightweight framework based on PiDiNet that enables fine-grained semantic edge detection.Our model simultaneously predicts background and four edge categories from full-resolution inputs,balancing accuracy and efficiency.Key contributions include:a multi-channel output structure expanding binary edge prediction to five classes,supported by a deep supervision mechanism;a dynamic class-balancing strategy combining adaptive weighting with physical priors to handle extreme class imbalance;and maintained architectural efficiency enabling real-time inference.Extensive evaluations on BSDS-RIND show our approach achieves accuracy competitive with state-of-the-art methods while operating in real time.展开更多
Vehicle-induced response separation is a crucial issue in structural health monitoring(SHM).This paper proposes a block-wise sliding recursive wavelet transform algorithm to meet the real-time processing requirements ...Vehicle-induced response separation is a crucial issue in structural health monitoring(SHM).This paper proposes a block-wise sliding recursive wavelet transform algorithm to meet the real-time processing requirements of monitoring data.To extend the separation target from a fixed dataset to a continuously updating data stream,a block-wise sliding framework is first developed.This framework is further optimized considering the characteristics of real-time data streams,and its advantage in computational efficiency is theoretically demonstrated.During the decomposition and reconstruction processes,information from neighboring data blocks is fully utilized to reduce algorithmic complexity.In addition,a delay-setting strategy is introduced for each processing window to mitigate boundary effects,thereby balancing accuracy and efficiency.Simulated signal experiments are conducted to determine the optimal delay configuration and to verify the algorithm’s superior performance,achieving a lower Root Mean Square Error(RMSE)and only 0.0249 times the average computational time compared with the original algorithm.Furthermore,strain signals from the Lieshi River Bridge are employed to validate the method.The proposed algorithm successfully separates the static trend from vehicle-induced responses in real time across different sampling frequencies,demonstrating its effectiveness and applicability in real-time bridge monitoring.展开更多
An innovative real-time monitoring method for surrounding rock damage based on microseismic time-lapse double-difference tomography is proposed for delayed dynamic damage identification and insufficient detection of a...An innovative real-time monitoring method for surrounding rock damage based on microseismic time-lapse double-difference tomography is proposed for delayed dynamic damage identification and insufficient detection of adverse geological conditions in deep-buried tunnel construction.The installation techniques for microseismic sensors were optimized by mounting sensors at bolt ends which significantly improves signal-to-noise ratio(SNR)and anti-interference capability compared to conventional borehole placement.Subsequently,a 3D wave velocity evolution model that incorporates construction-induced disturbances was established,enabling the first visualization of spatiotemporal variations in surrounding rock wave velocity.It finds significant wave velocity reduction near the tunnel face,with roof and floor damage zones extending 40–50 m;wave velocities approaching undisturbed levels at 15 m ahead of the working face and on the laterally undisturbed side;pronounced spatial asymmetry in wave velocity distribution—values on the left side exceed those on the right,with a clear stress concentration or transition zone located 10–15 m;and systematically lower velocities behind the face than in front,indicating asymmetric rock damage development.These results provide essential theoretical support and practical guidance for optimizing dynamic construction strategies,enabling real-time adjustment of support parameters,and establishing safety early warning systems in deep-buried tunnel engineering.展开更多
The dust cycle is a crucial component of the present-day Martian climate system.This study examines its multitimescale variability using an optimized 50-year simulation with the fully interactive scheme from the Globa...The dust cycle is a crucial component of the present-day Martian climate system.This study examines its multitimescale variability using an optimized 50-year simulation with the fully interactive scheme from the Global Open Planetary Atmospheric Model for Mars(GoMars),a newly developed Mars General Circulation Model(MGCM).GoMars is able to reproduce the diurnal,seasonal,and interannual characteristics of the dust cycle in several key aspects,with high repeatability in diurnal and seasonal variations during non-global dust storm(non-GDS)years.The model’s“climatology”(non-GDS years ensemble mean)captures the seasonal pattern and magnitude of the vertical–meridional dust distribution,validated against Mars Climate Database and Mars Climate Sounder observations.In the absence of direct observations,the GoMars-simulated near-surface wind stress lifting flux is evaluated through comparisons with other MGCMs(e.g.,MarsWRF),revealing consistent seasonal and spatial patterns.As for the diurnal cycle,the peak dust devil lifting flux occurs at 1200–1300 local time,matching the Mars Pathfinder measurements.The model also successfully captures the intense dust devil activity in Amazonis,a region identified as a major dust devil hotspot based on observational data.In GDS years,GoMars effectively reproduces spontaneous GDSs,capturing their observed onset times,locations,and dust transport patterns as exhibited in specific Martian years.The model also simulates significant interannual variability,with irregular GDS intervals along with reasonable dust–atmosphere interactions.展开更多
Online examinations have become a dominant assessment mode,increasing concerns over academic integrity.To address the critical challenge of detecting cheating behaviours,this study proposes a hybrid deep learning appr...Online examinations have become a dominant assessment mode,increasing concerns over academic integrity.To address the critical challenge of detecting cheating behaviours,this study proposes a hybrid deep learning approach that combines visual detection and temporal behaviour classification.The methodology utilises object detection models—You Only Look Once(YOLOv12),Faster Region-based Convolutional Neural Network(RCNN),and Single Shot Detector(SSD)MobileNet—integrated with classification models such as Convolutional Neural Networks(CNN),Bidirectional Gated Recurrent Unit(Bi-GRU),and CNN-LSTM(Long Short-Term Memory).Two distinct datasets were used:the Online Exam Proctoring(EOP)dataset from Michigan State University and the School of Computer Science,Duy Tan Unievrsity(SCS-DTU)dataset collected in a controlled classroom setting.A diverse set of cheating behaviours,including book usage,unauthorised interaction,internet access,and mobile phone use,was categorised.Comprehensive experiments evaluated the models based on accuracy,precision,recall,training time,inference speed,and memory usage.We evaluate nine detector-classifier pairings under a unified budget and score them via a calibrated harmonic mean of detection and classification accuracies,enabling deployment-oriented selection under latency and memory constraints.Macro-Precision/Recall/F1 and Receiver Operating Characteristic-Area Under the Curve(ROC-AUC)are reported for the top configurations,revealing consistent advantages of object-centric pipelines for fine-grained cheating cues.The highest overall score is achieved by YOLOv12+CNN(97.15%accuracy),while SSD-MobileNet+CNN provides the best speed-efficiency trade-off for edge devices.This research provides valuable insights into selecting and deploying appropriate deep learning models for maintaining exam integrity under varying resource constraints.展开更多
The highly dynamic nature,strong uncertainty,and coupled multiple safety constraints inherent in carrier aircraft recovery operations pose severe challenges for real-time decision-making.Addressing bolter scenarios,th...The highly dynamic nature,strong uncertainty,and coupled multiple safety constraints inherent in carrier aircraft recovery operations pose severe challenges for real-time decision-making.Addressing bolter scenarios,this study proposes an intelligent decision-making framework based on a deep long short-term memory Q-network.This framework transforms the real-time sequencing for bolter recovery problem into a partially observable Markov decision process.It employs a stacked long shortterm memory network to accurately capture the long-range temporal dependencies of bolter event chains and fuel consumption.Furthermore,it integrates a prioritized experience replay training mechanism to construct a safe and adaptive scheduling system capable of millisecond-level real-time decision-making.Experimental demonstrates that,within large-scale mass recovery scenarios,the framework achieves zero safety violations in static environments and maintains a fuel safety violation rate below 10%in dynamic scenarios,with single-step decision times at the millisecond level.The model exhibits strong generalization capability,effectively responding to unforeseen emergent situations—such as multiple bolters and fuel emergencies—without requiring retraining.This provides robust support for efficient carrier-based aircraft recovery operations.展开更多
This study presents a hybrid CNN-Transformer model for real-time recognition of affective tactile biosignals.The proposed framework combines convolutional neural networks(CNNs)to extract spatial and local temporal fea...This study presents a hybrid CNN-Transformer model for real-time recognition of affective tactile biosignals.The proposed framework combines convolutional neural networks(CNNs)to extract spatial and local temporal features with the Transformer encoder that captures long-range dependencies in time-series data through multi-head attention.Model performance was evaluated on two widely used tactile biosignal datasets,HAART and CoST,which contain diverse affective touch gestures recorded from pressure sensor arrays.TheCNN-Transformer model achieved recognition rates of 93.33%on HAART and 80.89%on CoST,outperforming existing methods on both benchmarks.By incorporating temporal windowing,the model enables instantaneous prediction,improving generalization across gestures of varying duration.These results highlight the effectiveness of deep learning for tactile biosignal processing and demonstrate the potential of theCNN-Transformer approach for future applications in wearable sensors,affective computing,and biomedical monitoring.展开更多
Aiming at the problems of lagging curriculum,weak practice,and single evaluation in the cultivation of HarmonyOS Development talents,this study constructs a“teacher-machine-student”ternary interactive teaching model...Aiming at the problems of lagging curriculum,weak practice,and single evaluation in the cultivation of HarmonyOS Development talents,this study constructs a“teacher-machine-student”ternary interactive teaching model based on the Congyou platform.Through the building block curriculum system,the HarmonyOS technology stack is decoupled into dynamic capability units,and a multi-disciplinary cross-case library is jointly built with Huawei,which significantly improves the synchronization of teaching content and industrial technology.This paper innovatively designs an AI collaborative teaching system,which employs knowledge graphs to plan learning paths,utilizes virtual equipment clusters to simulate development environments,and establishes a“diagnosis-feedback-enhancement”closed loop through AI-based review,thereby effectively improving students’development efficiency and code reuse rate.A three-dimensional evaluation model integrating task outcomes,process performance,and innovation is constructed,incorporating indicators such as code standardization and an innovation index to strengthen the cultivation of engineering thinking and innovative ability.Furthermore,a data-driven support platform is built to generate student competency profiles,open up the“credit-competency-certification”pathway,promote the transformation of course achievements into contributions to the Huawei ecosystem,and significantly shorten the job adaptation cycle for graduates.The research results provide a replicable paradigm for the cultivation of domestic operating system talents.展开更多
Reliable traffic flow prediction is crucial for mitigating urban congestion.This paper proposes Attentionbased spatiotemporal Interactive Dynamic Graph Convolutional Network(AIDGCN),a novel architecture integrating In...Reliable traffic flow prediction is crucial for mitigating urban congestion.This paper proposes Attentionbased spatiotemporal Interactive Dynamic Graph Convolutional Network(AIDGCN),a novel architecture integrating Interactive Dynamic Graph Convolution Network(IDGCN)with Temporal Multi-Head Trend-Aware Attention.Its core innovation lies in IDGCN,which uniquely splits sequences into symmetric intervals for interactive feature sharing via dynamic graphs,and a novel attention mechanism incorporating convolutional operations to capture essential local traffic trends—addressing a critical gap in standard attention for continuous data.For 15-and 60-min forecasting on METR-LA,AIDGCN achieves MAEs of 0.75%and 0.39%,and RMSEs of 1.32%and 0.14%,respectively.In the 60-min long-term forecasting of the PEMS-BAY dataset,the AIDGCN out-performs the MRA-BGCN method by 6.28%,4.93%,and 7.17%in terms of MAE,RMSE,and MAPE,respectively.Experimental results demonstrate the superiority of our pro-posed model over state-of-the-art methods.展开更多
Accurate fingertip detection is critical for translating hand gestures into actionable commands in vision-based human‒computer interaction(HCI)systems.However,challenges such as complex backgrounds,dynamic hand postur...Accurate fingertip detection is critical for translating hand gestures into actionable commands in vision-based human‒computer interaction(HCI)systems.However,challenges such as complex backgrounds,dynamic hand postures,and real-time processing constraints hinder reliable detection.This paper introduces a robust framework integrating three key innovations:(1)an adaptive Gaussian mixture model(GMM)enhanced with neighborhood pixel connectivity for precise motion extraction;(2)a weighted YCbCr color-space shadow removal algorithm to eliminate false positives;and(3)a centroid distance method refined with circularity constraints for accurate fingertip localization.Extensive experiments demonstrate a recognition accuracy of 97.26%across diverse scenarios,including varying illuminations,occlusions,and hand rotations.The algorithm processes each frame in 23.43 ms on average,satisfying real-time requirements.Comparative evaluations against state-of-the-art methods reveal significant improvements in precision(8.3%),recall(6.1%),and F-measure(7.8%).This work advances HCI applications such as virtual keyboards,gesture-controlled interfaces,and augmented reality systems.展开更多
The purpose of this study was to find a way to promote the collaboration and interaction of students and bring about the growth of learners through feedback while taking advantage of real-time interactive class via vi...The purpose of this study was to find a way to promote the collaboration and interaction of students and bring about the growth of learners through feedback while taking advantage of real-time interactive class via video conferencing tools.Although real-time interactive class with using video conferencing tools had great advantages,but there were also limitations of active interaction.To this end,real-time interactive tool and cloud-based educational platform were applied to create cases of learner participation classes and analyze the cases.The convergence of real-time interactive class tools and cloud tools has been able to draw students’participation and collaboration in non-face-to-face situations,and it can be seen that it is very helpful in creating learner-centered educational activities based on communication and interaction with students.Through this,the application of the cloud-based educational platform in real-time interactive class could lead students to participate and collaborate even in non-face-to-face situations.展开更多
In view of the limitations of the mathematical method used in the container terminal logistics system, this paper uses Unity3D to establish a computer simulation model for the container automated yard, which dynamical...In view of the limitations of the mathematical method used in the container terminal logistics system, this paper uses Unity3D to establish a computer simulation model for the container automated yard, which dynamically displays the operation process of the container automated yard logistics system in real time. Through the plane four-parameter coordinate conversion method and by taking the Shanghai urban construction coordinate system as the medium, it completes the conversion from the satellite positioning reference ellipsoid coordinates to the three-dimensional virtual scene coordinates. The example results show that the method is reliable and practical, improves the accuracy and efficiency of positioning, and provides a reliable reference basis for the container terminal logistics system.展开更多
In the foundry industries,process design has traditionally relied on manuals and complex theoretical calculations.With the advent of 3D design in casting,computer-aided design(CAD)has been applied to integrate the fea...In the foundry industries,process design has traditionally relied on manuals and complex theoretical calculations.With the advent of 3D design in casting,computer-aided design(CAD)has been applied to integrate the features of casting process,thereby expanding the scope of design options.These technologies use parametric model design techniques for rapid component creation and use databases to access standard process parameters and design specifications.However,3D models are currently still created through inputting or calling parameters,which requires numerous verifications through calculations to ensure the design rationality.This process may be significantly slowed down due to repetitive modifications and extended design time.As a result,there are increasingly urgent demands for a real-time verification mechanism to address this issue.Therefore,this study proposed a novel closed-loop model and software development method that integrated contextual design with real-time verification,dynamically verifying relevant rules for designing 3D casting components.Additionally,the study analyzed three typical closed-loop scenarios of agile design in an independent developed intelligent casting process system.It is believed that foundry industries can potentially benefit from favorably reduced design cycles to yield an enhanced competitive product market.展开更多
Large portions of the tunnel boring machine(TBM)construction cost are attributed to disc cutter consumption,and assessing the disc cutter's wear level can help determine the optimal time to replace the disc cutter...Large portions of the tunnel boring machine(TBM)construction cost are attributed to disc cutter consumption,and assessing the disc cutter's wear level can help determine the optimal time to replace the disc cutter.Therefore,the need to monitor disc cutter wear in real-time has emerged as a technical challenge for TBMs.In this study,real-time disc cutter wear monitoring is developed based on sound and vibration sensors.For this purpose,the microphone and accelerometer were used to record the sound and vibration signals of cutting three different types of rocks with varying abrasions on a laboratory scale.The relationship between disc cutter wear and the sound and vibration signal was determined by comparing the measurements of disc cutter wear with the signal plots for each sample.The features extracted from the signals showed that the sound and vibration signals are impacted by the progression of disc wear during the rock-cutting process.The signal features obtained from the rock-cutting operation were utilized to verify the machine learning techniques.The results showed that the multilayer perceptron(MLP),random subspace-based decision tree(RS-DT),DT,and random forest(RF)methods could predict the wear level of the disc cutter with an accuracy of 0.89,0.951,0.951,and 0.927,respectively.Based on the accuracy of the models and the confusion matrix,it was found that the RS-DT model has the best estimate for predicting the level of disc wear.This research has developed a method that can potentially determine when to replace a tool and assess disc wear in real-time.展开更多
Real-time train rescheduling plays a vital role in railway transportation as it is crucial for maintaining punctuality and reliability in rail operations.In this paper,we propose a rescheduling model that incorporates...Real-time train rescheduling plays a vital role in railway transportation as it is crucial for maintaining punctuality and reliability in rail operations.In this paper,we propose a rescheduling model that incorporates constraints and objectives generated through human-computer interaction.This approach ensures that the model is aligned with practical requirements and daily operational tasks while facilitating iterative train rescheduling.The dispatcher’s empirical knowledge is integrated into the train rescheduling process using a human-computer interaction framework.We introduce six interfaces to dynamically construct constraints and objectives that capture human intentions.By summarizing rescheduling rules,we devise a rule-based conflict detection-resolution heuristic algorithm to effectively solve the formulated model.A series of numerical experiments are presented,demonstrating strong performance across the entire system.Furthermore,theflexibility of rescheduling is enhanced through secondary analysis-driven solutions derived from the outcomes of humancomputer interactions in the previous step.This proposed interaction method complements existing literature on rescheduling methods involving human-computer interactions.It serves as a tool to aid dispatchers in identifying more feasible solutions in accordance with their empirical rescheduling strategies.展开更多
The Internet of Things(IoT)and mobile technology have significantly transformed healthcare by enabling real-time monitoring and diagnosis of patients.Recognizing Medical-Related Human Activities(MRHA)is pivotal for he...The Internet of Things(IoT)and mobile technology have significantly transformed healthcare by enabling real-time monitoring and diagnosis of patients.Recognizing Medical-Related Human Activities(MRHA)is pivotal for healthcare systems,particularly for identifying actions critical to patient well-being.However,challenges such as high computational demands,low accuracy,and limited adaptability persist in Human Motion Recognition(HMR).While some studies have integrated HMR with IoT for real-time healthcare applications,limited research has focused on recognizing MRHA as essential for effective patient monitoring.This study proposes a novel HMR method tailored for MRHA detection,leveraging multi-stage deep learning techniques integrated with IoT.The approach employs EfficientNet to extract optimized spatial features from skeleton frame sequences using seven Mobile Inverted Bottleneck Convolutions(MBConv)blocks,followed by Convolutional Long Short Term Memory(ConvLSTM)to capture spatio-temporal patterns.A classification module with global average pooling,a fully connected layer,and a dropout layer generates the final predictions.The model is evaluated on the NTU RGB+D 120 and HMDB51 datasets,focusing on MRHA such as sneezing,falling,walking,sitting,etc.It achieves 94.85%accuracy for cross-subject evaluations and 96.45%for cross-view evaluations on NTU RGB+D 120,along with 89.22%accuracy on HMDB51.Additionally,the system integrates IoT capabilities using a Raspberry Pi and GSM module,delivering real-time alerts via Twilios SMS service to caregivers and patients.This scalable and efficient solution bridges the gap between HMR and IoT,advancing patient monitoring,improving healthcare outcomes,and reducing costs.展开更多
In recent years,railway construction in China has developed vigorously.With continuous improvements in the highspeed railway network,the focus is gradually shifting from large-scale construction to large-scale operati...In recent years,railway construction in China has developed vigorously.With continuous improvements in the highspeed railway network,the focus is gradually shifting from large-scale construction to large-scale operations.However,several challenges have emerged within the high-speed railway dispatching and command system,including the heavy workload faced by dispatchers,the difficulty of quantifying subjective expertise,and the need for effective training of professionals.Amid the growing application of artificial intelligence technologies in railway systems,this study leverages Large Language Model(LLM)technology.LLMs bring enhanced intelligence,predictive capabilities,robust memory,and adaptability to diverse real-world scenarios.This study proposes a human-computer interactive intelligent scheduling auxiliary training system built on LLM technology.The system offers capabilities including natural dialogue,knowledge reasoning,and human feedback learning.With broad applicability,the system is suitable for vocational education,guided inquiry,knowledge-based Q&A,and other training scenarios.Validation results demonstrate its effectiveness in auxiliary training,providing substantial support for educators,students,and dispatching personnel in colleges and professional settings.展开更多
Along with process control,perception represents the main function performed by the Edge Layer of an Internet of Things(IoT)network.Many of these networks implement various applications where the response time does no...Along with process control,perception represents the main function performed by the Edge Layer of an Internet of Things(IoT)network.Many of these networks implement various applications where the response time does not represent an important parameter.However,in critical applications,this parameter represents a crucial aspect.One important sensing device used in IoT designs is the accelerometer.In most applications,the response time of the embedded driver software handling this device is generally not analysed and not taken into account.In this paper,we present the design and implementation of a predictable real-time driver stack for a popular accelerometer and gyroscope device family.We provide clear justifications for why this response time is extremely important for critical applications in the acquisition process of such data.We present extensive measurements and experimental results that demonstrate the predictability of our solution,making it suitable for critical real-time systems.展开更多
Fertilization or atmospheric deposition of nitrogen(N)and phosphorus(P)to terrestrial ecosystems can alter soil N(P)availability and the nature of nutrient limitation for plant growth.Changing the allocation of leaf P...Fertilization or atmospheric deposition of nitrogen(N)and phosphorus(P)to terrestrial ecosystems can alter soil N(P)availability and the nature of nutrient limitation for plant growth.Changing the allocation of leaf P fractions is potentially an adaptive strategy for plants to cope with soil N(P)availability and nutrient-limiting conditions.However,the impact of the interactions between imbalanced anthropogenic N and P inputs on the concentrations and allocation proportions of leaf P fractions in forest woody plants remains elusive.We conducted a metaanalysis of data about the concentrations and allocation proportions of leaf P fractions,specifically associated with individual and combined additions of N and P in evergreen forests,the dominant vegetation type in southern China where the primary productivity is usually considered limited by P.This assessment allowed us to quantitatively evaluate the effects of N and P additions alone and interactively on leaf P allocation and use strategies.Nitrogen addition(exacerbating P limitation)reduced the concentrations of leaf total P and different leaf P fractions.Nitrogen addition reduced the allocation to leaf metabolic P but increased the allocation to other fractions,while P addition showed opposite trends.The simultaneous additions of N and P showed an antagonistic(mutual suppression)effect on the concentrations of leaf P fractions,but an additive(summary)effect on the allocation proportions of leaf P fractions.These results highlight the importance of strategies of leaf P fraction allocation in forest plants under changes in environmental nutrient availability.Importantly,our study identified critical interactions associated with combined N and P inputs that affect leaf P fractions,thus aiding in predicting plant acclimation strategies in the context of intensifying and imbalanced anthropogenic nutrient inputs.展开更多
基金supported by the National Natural Science Foundation of China 62402171.
文摘As a fundamental component in computer vision,edges can be categorized into four types based on discontinuities in reflectance,illumination,surface normal,or depth.While deep CNNs have significantly advanced generic edge detection,real-time multi-class semantic edge detection under resource constraints remains challenging.To address this,we propose a lightweight framework based on PiDiNet that enables fine-grained semantic edge detection.Our model simultaneously predicts background and four edge categories from full-resolution inputs,balancing accuracy and efficiency.Key contributions include:a multi-channel output structure expanding binary edge prediction to five classes,supported by a deep supervision mechanism;a dynamic class-balancing strategy combining adaptive weighting with physical priors to handle extreme class imbalance;and maintained architectural efficiency enabling real-time inference.Extensive evaluations on BSDS-RIND show our approach achieves accuracy competitive with state-of-the-art methods while operating in real time.
基金the support of the Major Science and Technology Project of Yunnan Province,China(Grant No.202502AD080007)the National Natural Science Foundation of China(Grant No.52378288)。
文摘Vehicle-induced response separation is a crucial issue in structural health monitoring(SHM).This paper proposes a block-wise sliding recursive wavelet transform algorithm to meet the real-time processing requirements of monitoring data.To extend the separation target from a fixed dataset to a continuously updating data stream,a block-wise sliding framework is first developed.This framework is further optimized considering the characteristics of real-time data streams,and its advantage in computational efficiency is theoretically demonstrated.During the decomposition and reconstruction processes,information from neighboring data blocks is fully utilized to reduce algorithmic complexity.In addition,a delay-setting strategy is introduced for each processing window to mitigate boundary effects,thereby balancing accuracy and efficiency.Simulated signal experiments are conducted to determine the optimal delay configuration and to verify the algorithm’s superior performance,achieving a lower Root Mean Square Error(RMSE)and only 0.0249 times the average computational time compared with the original algorithm.Furthermore,strain signals from the Lieshi River Bridge are employed to validate the method.The proposed algorithm successfully separates the static trend from vehicle-induced responses in real time across different sampling frequencies,demonstrating its effectiveness and applicability in real-time bridge monitoring.
基金support of the National Natural Science Foundation of China(No.52274176)the Guangdong Province Key Areas R&D Program(No.2022B0101070001)+5 种基金Chongqing Elite Innovation and Entrepreneurship Leading talent Project(No.CQYC20220302517)the Chongqing Natural Science Foundation Innovation and Development Joint Fund(No.CSTB2022NSCQ-LZX0079)the National Key Research and Development Program Young Scientists Project(No.2022YFC2905700)the Chongqing Municipal Education Commission“Shuangcheng Economic Circle Construction in Chengdu-Chongqing Area”Science and Technology Innovation Project(No.KJCX2020031)the Fundamental Research Funds for the Central Universities(No.2024CDJGF-009)the Key Project for Technological Innovation and Application Development in Chongqing(No.CSTB2025TIAD-KPX0029).
文摘An innovative real-time monitoring method for surrounding rock damage based on microseismic time-lapse double-difference tomography is proposed for delayed dynamic damage identification and insufficient detection of adverse geological conditions in deep-buried tunnel construction.The installation techniques for microseismic sensors were optimized by mounting sensors at bolt ends which significantly improves signal-to-noise ratio(SNR)and anti-interference capability compared to conventional borehole placement.Subsequently,a 3D wave velocity evolution model that incorporates construction-induced disturbances was established,enabling the first visualization of spatiotemporal variations in surrounding rock wave velocity.It finds significant wave velocity reduction near the tunnel face,with roof and floor damage zones extending 40–50 m;wave velocities approaching undisturbed levels at 15 m ahead of the working face and on the laterally undisturbed side;pronounced spatial asymmetry in wave velocity distribution—values on the left side exceed those on the right,with a clear stress concentration or transition zone located 10–15 m;and systematically lower velocities behind the face than in front,indicating asymmetric rock damage development.These results provide essential theoretical support and practical guidance for optimizing dynamic construction strategies,enabling real-time adjustment of support parameters,and establishing safety early warning systems in deep-buried tunnel engineering.
基金jointly supported by the National Natural Science Foundation of China(Grant No.42475135)the Key Technology Research Project of TW-3(TW3006)the IAP’s basic scientific research project during the 14th Five-Year Plan Period.
文摘The dust cycle is a crucial component of the present-day Martian climate system.This study examines its multitimescale variability using an optimized 50-year simulation with the fully interactive scheme from the Global Open Planetary Atmospheric Model for Mars(GoMars),a newly developed Mars General Circulation Model(MGCM).GoMars is able to reproduce the diurnal,seasonal,and interannual characteristics of the dust cycle in several key aspects,with high repeatability in diurnal and seasonal variations during non-global dust storm(non-GDS)years.The model’s“climatology”(non-GDS years ensemble mean)captures the seasonal pattern and magnitude of the vertical–meridional dust distribution,validated against Mars Climate Database and Mars Climate Sounder observations.In the absence of direct observations,the GoMars-simulated near-surface wind stress lifting flux is evaluated through comparisons with other MGCMs(e.g.,MarsWRF),revealing consistent seasonal and spatial patterns.As for the diurnal cycle,the peak dust devil lifting flux occurs at 1200–1300 local time,matching the Mars Pathfinder measurements.The model also successfully captures the intense dust devil activity in Amazonis,a region identified as a major dust devil hotspot based on observational data.In GDS years,GoMars effectively reproduces spontaneous GDSs,capturing their observed onset times,locations,and dust transport patterns as exhibited in specific Martian years.The model also simulates significant interannual variability,with irregular GDS intervals along with reasonable dust–atmosphere interactions.
文摘Online examinations have become a dominant assessment mode,increasing concerns over academic integrity.To address the critical challenge of detecting cheating behaviours,this study proposes a hybrid deep learning approach that combines visual detection and temporal behaviour classification.The methodology utilises object detection models—You Only Look Once(YOLOv12),Faster Region-based Convolutional Neural Network(RCNN),and Single Shot Detector(SSD)MobileNet—integrated with classification models such as Convolutional Neural Networks(CNN),Bidirectional Gated Recurrent Unit(Bi-GRU),and CNN-LSTM(Long Short-Term Memory).Two distinct datasets were used:the Online Exam Proctoring(EOP)dataset from Michigan State University and the School of Computer Science,Duy Tan Unievrsity(SCS-DTU)dataset collected in a controlled classroom setting.A diverse set of cheating behaviours,including book usage,unauthorised interaction,internet access,and mobile phone use,was categorised.Comprehensive experiments evaluated the models based on accuracy,precision,recall,training time,inference speed,and memory usage.We evaluate nine detector-classifier pairings under a unified budget and score them via a calibrated harmonic mean of detection and classification accuracies,enabling deployment-oriented selection under latency and memory constraints.Macro-Precision/Recall/F1 and Receiver Operating Characteristic-Area Under the Curve(ROC-AUC)are reported for the top configurations,revealing consistent advantages of object-centric pipelines for fine-grained cheating cues.The highest overall score is achieved by YOLOv12+CNN(97.15%accuracy),while SSD-MobileNet+CNN provides the best speed-efficiency trade-off for edge devices.This research provides valuable insights into selecting and deploying appropriate deep learning models for maintaining exam integrity under varying resource constraints.
基金supported by the National Natural Science Foundation of China(Grant No.62403486)。
文摘The highly dynamic nature,strong uncertainty,and coupled multiple safety constraints inherent in carrier aircraft recovery operations pose severe challenges for real-time decision-making.Addressing bolter scenarios,this study proposes an intelligent decision-making framework based on a deep long short-term memory Q-network.This framework transforms the real-time sequencing for bolter recovery problem into a partially observable Markov decision process.It employs a stacked long shortterm memory network to accurately capture the long-range temporal dependencies of bolter event chains and fuel consumption.Furthermore,it integrates a prioritized experience replay training mechanism to construct a safe and adaptive scheduling system capable of millisecond-level real-time decision-making.Experimental demonstrates that,within large-scale mass recovery scenarios,the framework achieves zero safety violations in static environments and maintains a fuel safety violation rate below 10%in dynamic scenarios,with single-step decision times at the millisecond level.The model exhibits strong generalization capability,effectively responding to unforeseen emergent situations—such as multiple bolters and fuel emergencies—without requiring retraining.This provides robust support for efficient carrier-based aircraft recovery operations.
文摘This study presents a hybrid CNN-Transformer model for real-time recognition of affective tactile biosignals.The proposed framework combines convolutional neural networks(CNNs)to extract spatial and local temporal features with the Transformer encoder that captures long-range dependencies in time-series data through multi-head attention.Model performance was evaluated on two widely used tactile biosignal datasets,HAART and CoST,which contain diverse affective touch gestures recorded from pressure sensor arrays.TheCNN-Transformer model achieved recognition rates of 93.33%on HAART and 80.89%on CoST,outperforming existing methods on both benchmarks.By incorporating temporal windowing,the model enables instantaneous prediction,improving generalization across gestures of varying duration.These results highlight the effectiveness of deep learning for tactile biosignal processing and demonstrate the potential of theCNN-Transformer approach for future applications in wearable sensors,affective computing,and biomedical monitoring.
文摘Aiming at the problems of lagging curriculum,weak practice,and single evaluation in the cultivation of HarmonyOS Development talents,this study constructs a“teacher-machine-student”ternary interactive teaching model based on the Congyou platform.Through the building block curriculum system,the HarmonyOS technology stack is decoupled into dynamic capability units,and a multi-disciplinary cross-case library is jointly built with Huawei,which significantly improves the synchronization of teaching content and industrial technology.This paper innovatively designs an AI collaborative teaching system,which employs knowledge graphs to plan learning paths,utilizes virtual equipment clusters to simulate development environments,and establishes a“diagnosis-feedback-enhancement”closed loop through AI-based review,thereby effectively improving students’development efficiency and code reuse rate.A three-dimensional evaluation model integrating task outcomes,process performance,and innovation is constructed,incorporating indicators such as code standardization and an innovation index to strengthen the cultivation of engineering thinking and innovative ability.Furthermore,a data-driven support platform is built to generate student competency profiles,open up the“credit-competency-certification”pathway,promote the transformation of course achievements into contributions to the Huawei ecosystem,and significantly shorten the job adaptation cycle for graduates.The research results provide a replicable paradigm for the cultivation of domestic operating system talents.
文摘Reliable traffic flow prediction is crucial for mitigating urban congestion.This paper proposes Attentionbased spatiotemporal Interactive Dynamic Graph Convolutional Network(AIDGCN),a novel architecture integrating Interactive Dynamic Graph Convolution Network(IDGCN)with Temporal Multi-Head Trend-Aware Attention.Its core innovation lies in IDGCN,which uniquely splits sequences into symmetric intervals for interactive feature sharing via dynamic graphs,and a novel attention mechanism incorporating convolutional operations to capture essential local traffic trends—addressing a critical gap in standard attention for continuous data.For 15-and 60-min forecasting on METR-LA,AIDGCN achieves MAEs of 0.75%and 0.39%,and RMSEs of 1.32%and 0.14%,respectively.In the 60-min long-term forecasting of the PEMS-BAY dataset,the AIDGCN out-performs the MRA-BGCN method by 6.28%,4.93%,and 7.17%in terms of MAE,RMSE,and MAPE,respectively.Experimental results demonstrate the superiority of our pro-posed model over state-of-the-art methods.
基金funded by the Jiaying University Research Start-Up Fund(grant number 323E0431).
文摘Accurate fingertip detection is critical for translating hand gestures into actionable commands in vision-based human‒computer interaction(HCI)systems.However,challenges such as complex backgrounds,dynamic hand postures,and real-time processing constraints hinder reliable detection.This paper introduces a robust framework integrating three key innovations:(1)an adaptive Gaussian mixture model(GMM)enhanced with neighborhood pixel connectivity for precise motion extraction;(2)a weighted YCbCr color-space shadow removal algorithm to eliminate false positives;and(3)a centroid distance method refined with circularity constraints for accurate fingertip localization.Extensive experiments demonstrate a recognition accuracy of 97.26%across diverse scenarios,including varying illuminations,occlusions,and hand rotations.The algorithm processes each frame in 23.43 ms on average,satisfying real-time requirements.Comparative evaluations against state-of-the-art methods reveal significant improvements in precision(8.3%),recall(6.1%),and F-measure(7.8%).This work advances HCI applications such as virtual keyboards,gesture-controlled interfaces,and augmented reality systems.
文摘The purpose of this study was to find a way to promote the collaboration and interaction of students and bring about the growth of learners through feedback while taking advantage of real-time interactive class via video conferencing tools.Although real-time interactive class with using video conferencing tools had great advantages,but there were also limitations of active interaction.To this end,real-time interactive tool and cloud-based educational platform were applied to create cases of learner participation classes and analyze the cases.The convergence of real-time interactive class tools and cloud tools has been able to draw students’participation and collaboration in non-face-to-face situations,and it can be seen that it is very helpful in creating learner-centered educational activities based on communication and interaction with students.Through this,the application of the cloud-based educational platform in real-time interactive class could lead students to participate and collaborate even in non-face-to-face situations.
文摘In view of the limitations of the mathematical method used in the container terminal logistics system, this paper uses Unity3D to establish a computer simulation model for the container automated yard, which dynamically displays the operation process of the container automated yard logistics system in real time. Through the plane four-parameter coordinate conversion method and by taking the Shanghai urban construction coordinate system as the medium, it completes the conversion from the satellite positioning reference ellipsoid coordinates to the three-dimensional virtual scene coordinates. The example results show that the method is reliable and practical, improves the accuracy and efficiency of positioning, and provides a reliable reference basis for the container terminal logistics system.
基金the financial support of the Natural Science Foundation of Hubei Province,China (Grant No.2022CFB770)。
文摘In the foundry industries,process design has traditionally relied on manuals and complex theoretical calculations.With the advent of 3D design in casting,computer-aided design(CAD)has been applied to integrate the features of casting process,thereby expanding the scope of design options.These technologies use parametric model design techniques for rapid component creation and use databases to access standard process parameters and design specifications.However,3D models are currently still created through inputting or calling parameters,which requires numerous verifications through calculations to ensure the design rationality.This process may be significantly slowed down due to repetitive modifications and extended design time.As a result,there are increasingly urgent demands for a real-time verification mechanism to address this issue.Therefore,this study proposed a novel closed-loop model and software development method that integrated contextual design with real-time verification,dynamically verifying relevant rules for designing 3D casting components.Additionally,the study analyzed three typical closed-loop scenarios of agile design in an independent developed intelligent casting process system.It is believed that foundry industries can potentially benefit from favorably reduced design cycles to yield an enhanced competitive product market.
文摘Large portions of the tunnel boring machine(TBM)construction cost are attributed to disc cutter consumption,and assessing the disc cutter's wear level can help determine the optimal time to replace the disc cutter.Therefore,the need to monitor disc cutter wear in real-time has emerged as a technical challenge for TBMs.In this study,real-time disc cutter wear monitoring is developed based on sound and vibration sensors.For this purpose,the microphone and accelerometer were used to record the sound and vibration signals of cutting three different types of rocks with varying abrasions on a laboratory scale.The relationship between disc cutter wear and the sound and vibration signal was determined by comparing the measurements of disc cutter wear with the signal plots for each sample.The features extracted from the signals showed that the sound and vibration signals are impacted by the progression of disc wear during the rock-cutting process.The signal features obtained from the rock-cutting operation were utilized to verify the machine learning techniques.The results showed that the multilayer perceptron(MLP),random subspace-based decision tree(RS-DT),DT,and random forest(RF)methods could predict the wear level of the disc cutter with an accuracy of 0.89,0.951,0.951,and 0.927,respectively.Based on the accuracy of the models and the confusion matrix,it was found that the RS-DT model has the best estimate for predicting the level of disc wear.This research has developed a method that can potentially determine when to replace a tool and assess disc wear in real-time.
基金supported by the China Fundamental Research Funds for the Central Universities(2022JBQY006)。
文摘Real-time train rescheduling plays a vital role in railway transportation as it is crucial for maintaining punctuality and reliability in rail operations.In this paper,we propose a rescheduling model that incorporates constraints and objectives generated through human-computer interaction.This approach ensures that the model is aligned with practical requirements and daily operational tasks while facilitating iterative train rescheduling.The dispatcher’s empirical knowledge is integrated into the train rescheduling process using a human-computer interaction framework.We introduce six interfaces to dynamically construct constraints and objectives that capture human intentions.By summarizing rescheduling rules,we devise a rule-based conflict detection-resolution heuristic algorithm to effectively solve the formulated model.A series of numerical experiments are presented,demonstrating strong performance across the entire system.Furthermore,theflexibility of rescheduling is enhanced through secondary analysis-driven solutions derived from the outcomes of humancomputer interactions in the previous step.This proposed interaction method complements existing literature on rescheduling methods involving human-computer interactions.It serves as a tool to aid dispatchers in identifying more feasible solutions in accordance with their empirical rescheduling strategies.
基金funded by the ICT Division of theMinistry of Posts,Telecommunications,and Information Technology of Bangladesh under Grant Number 56.00.0000.052.33.005.21-7(Tracking No.22FS15306)support from the University of Rajshahi.
文摘The Internet of Things(IoT)and mobile technology have significantly transformed healthcare by enabling real-time monitoring and diagnosis of patients.Recognizing Medical-Related Human Activities(MRHA)is pivotal for healthcare systems,particularly for identifying actions critical to patient well-being.However,challenges such as high computational demands,low accuracy,and limited adaptability persist in Human Motion Recognition(HMR).While some studies have integrated HMR with IoT for real-time healthcare applications,limited research has focused on recognizing MRHA as essential for effective patient monitoring.This study proposes a novel HMR method tailored for MRHA detection,leveraging multi-stage deep learning techniques integrated with IoT.The approach employs EfficientNet to extract optimized spatial features from skeleton frame sequences using seven Mobile Inverted Bottleneck Convolutions(MBConv)blocks,followed by Convolutional Long Short Term Memory(ConvLSTM)to capture spatio-temporal patterns.A classification module with global average pooling,a fully connected layer,and a dropout layer generates the final predictions.The model is evaluated on the NTU RGB+D 120 and HMDB51 datasets,focusing on MRHA such as sneezing,falling,walking,sitting,etc.It achieves 94.85%accuracy for cross-subject evaluations and 96.45%for cross-view evaluations on NTU RGB+D 120,along with 89.22%accuracy on HMDB51.Additionally,the system integrates IoT capabilities using a Raspberry Pi and GSM module,delivering real-time alerts via Twilios SMS service to caregivers and patients.This scalable and efficient solution bridges the gap between HMR and IoT,advancing patient monitoring,improving healthcare outcomes,and reducing costs.
基金the Talent Fund of Beijing Jiaotong University(Grant No.2024XKRC055).
文摘In recent years,railway construction in China has developed vigorously.With continuous improvements in the highspeed railway network,the focus is gradually shifting from large-scale construction to large-scale operations.However,several challenges have emerged within the high-speed railway dispatching and command system,including the heavy workload faced by dispatchers,the difficulty of quantifying subjective expertise,and the need for effective training of professionals.Amid the growing application of artificial intelligence technologies in railway systems,this study leverages Large Language Model(LLM)technology.LLMs bring enhanced intelligence,predictive capabilities,robust memory,and adaptability to diverse real-world scenarios.This study proposes a human-computer interactive intelligent scheduling auxiliary training system built on LLM technology.The system offers capabilities including natural dialogue,knowledge reasoning,and human feedback learning.With broad applicability,the system is suitable for vocational education,guided inquiry,knowledge-based Q&A,and other training scenarios.Validation results demonstrate its effectiveness in auxiliary training,providing substantial support for educators,students,and dispatching personnel in colleges and professional settings.
文摘Along with process control,perception represents the main function performed by the Edge Layer of an Internet of Things(IoT)network.Many of these networks implement various applications where the response time does not represent an important parameter.However,in critical applications,this parameter represents a crucial aspect.One important sensing device used in IoT designs is the accelerometer.In most applications,the response time of the embedded driver software handling this device is generally not analysed and not taken into account.In this paper,we present the design and implementation of a predictable real-time driver stack for a popular accelerometer and gyroscope device family.We provide clear justifications for why this response time is extremely important for critical applications in the acquisition process of such data.We present extensive measurements and experimental results that demonstrate the predictability of our solution,making it suitable for critical real-time systems.
基金supported by the National Natural Science Foundation of China(No.41473068)supported by China Postdoctoral Science Foundation(No.2022M722667)。
文摘Fertilization or atmospheric deposition of nitrogen(N)and phosphorus(P)to terrestrial ecosystems can alter soil N(P)availability and the nature of nutrient limitation for plant growth.Changing the allocation of leaf P fractions is potentially an adaptive strategy for plants to cope with soil N(P)availability and nutrient-limiting conditions.However,the impact of the interactions between imbalanced anthropogenic N and P inputs on the concentrations and allocation proportions of leaf P fractions in forest woody plants remains elusive.We conducted a metaanalysis of data about the concentrations and allocation proportions of leaf P fractions,specifically associated with individual and combined additions of N and P in evergreen forests,the dominant vegetation type in southern China where the primary productivity is usually considered limited by P.This assessment allowed us to quantitatively evaluate the effects of N and P additions alone and interactively on leaf P allocation and use strategies.Nitrogen addition(exacerbating P limitation)reduced the concentrations of leaf total P and different leaf P fractions.Nitrogen addition reduced the allocation to leaf metabolic P but increased the allocation to other fractions,while P addition showed opposite trends.The simultaneous additions of N and P showed an antagonistic(mutual suppression)effect on the concentrations of leaf P fractions,but an additive(summary)effect on the allocation proportions of leaf P fractions.These results highlight the importance of strategies of leaf P fraction allocation in forest plants under changes in environmental nutrient availability.Importantly,our study identified critical interactions associated with combined N and P inputs that affect leaf P fractions,thus aiding in predicting plant acclimation strategies in the context of intensifying and imbalanced anthropogenic nutrient inputs.
