Accurate traffic flow prediction has a profound impact on modern traffic management. Traffic flow has complex spatial-temporal correlations and periodicity, which poses difficulties for precise prediction. To address ...Accurate traffic flow prediction has a profound impact on modern traffic management. Traffic flow has complex spatial-temporal correlations and periodicity, which poses difficulties for precise prediction. To address this problem, a Multi-head Self-attention and Spatial-Temporal Graph Convolutional Network (MSSTGCN) for multiscale traffic flow prediction is proposed. Firstly, to capture the hidden traffic periodicity of traffic flow, traffic flow is divided into three kinds of periods, including hourly, daily, and weekly data. Secondly, a graph attention residual layer is constructed to learn the global spatial features across regions. Local spatial-temporal dependence is captured by using a T-GCN module. Thirdly, a transformer layer is introduced to learn the long-term dependence in time. A position embedding mechanism is introduced to label position information for all traffic sequences. Thus, this multi-head self-attention mechanism can recognize the sequence order and allocate weights for different time nodes. Experimental results on four real-world datasets show that the MSSTGCN performs better than the baseline methods and can be successfully adapted to traffic prediction tasks.展开更多
Spatial-temporal traffic prediction technology is crucial for network planning,resource allocation optimizing,and user experience improving.With the development of virtual network operators,multi-operator collaboratio...Spatial-temporal traffic prediction technology is crucial for network planning,resource allocation optimizing,and user experience improving.With the development of virtual network operators,multi-operator collaborations,and edge computing,spatial-temporal traffic data has taken on a distributed nature.Consequently,noncentralized spatial-temporal traffic prediction solutions have emerged as a recent research focus.Currently,the majority of research typically adopts federated learning methods to train traffic prediction models distributed on each base station.This method reduces additional burden on communication systems.However,this method has a drawback:it cannot handle irregular traffic data.Due to unstable wireless network environments,device failures,insufficient storage resources,etc.,data missing inevitably occurs during the process of collecting traffic data.This results in the irregular nature of distributed traffic data.Yet,commonly used traffic prediction models such as Recurrent Neural Networks(RNN)and Long Short-Term Memory(LSTM)typically assume that the data is complete and regular.To address the challenge of handling irregular traffic data,this paper transforms irregular traffic prediction into problems of estimating latent variables and generating future traffic.To solve the aforementioned problems,this paper introduces split learning to design a structured distributed learning framework.The framework comprises a Global-level Spatial structure mining Model(GSM)and several Nodelevel Generative Models(NGMs).NGM and GSM represent Seq2Seq models deployed on the base station and graph neural network models deployed on the cloud or central controller.Firstly,the time embedding layer in NGM establishes the mapping relationship between irregular traffic data and regular latent temporal feature variables.Secondly,GSM collects statistical feature parameters of latent temporal feature variables from various nodes and executes graph embedding for spatial-temporal traffic data.Finally,NGM generates future traffic based on latent temporal and spatial feature variables.The introduction of the time attention mechanism enhances the framework’s capability to handle irregular traffic data.Graph attention network introduces spatially correlated base station traffic feature information into local traffic prediction,which compensates for missing information in local irregular traffic data.The proposed framework effectively addresses the distributed prediction issues of irregular traffic data.By testing on real world datasets,the proposed framework improves traffic prediction accuracy by 35%compared to other commonly used distributed traffic prediction methods.展开更多
Root zone soil moisture(RZSM)plays a critical role in land-atmosphere hydrological cycles and serves as the primary water source for vegetation growth.However,the correlations between RZSM and its associated variables...Root zone soil moisture(RZSM)plays a critical role in land-atmosphere hydrological cycles and serves as the primary water source for vegetation growth.However,the correlations between RZSM and its associated variables,including surface soil moisture(SSM),often exhibit nonlinearities that are challenging to identify and quantify using conventional statistical techniques.Therefore,this study presents a hybrid convolutional neural network(CNN)-long short-term memory neural network(LSTM)-attention(CLA)model for predicting RZSM.Owing to the scarcity of soil moisture(SM)observation data,the physical model Hydrus-1D was employed to simulate a comprehensive dataset of spatial-temporal SM.Meteorological data and moderate resolution imaging spectroradiometer vegetation characterization parameters were used as predictor variables for the training and validation of the CLA model.The results of the CLA model for SM prediction in the root zone were significantly enhanced compared with those of the traditional LSTM and CNN-LSTM models.This was particularly notable at the depth of 80–100 cm,where the fitness(R^(2))reached nearly 0.9298.Moreover,the root mean square error of the CLA model was reduced by 49%and 57%compared with those of the LSTM and CNN-LSTM models,respectively.This study demonstrates that the integration of physical modeling and deep learning methods provides a more comprehensive and accurate understanding of spatial-temporal SM variations in the root zone.展开更多
Titanium-based semiconductors are known for their high chemical stability and suitable band gap widths.However,the conventional experimental screening methods are inefficient due to the wide variety of materials.To sp...Titanium-based semiconductors are known for their high chemical stability and suitable band gap widths.However,the conventional experimental screening methods are inefficient due to the wide variety of materials.To speed up the selection process,this work focuses on interpretable feature learning and band gap prediction for titanium-based semiconductors.First,titanium compounds were selected from the Materials Project database by machine learning,and elemental features were extracted using the Magpie descriptors.Then,principal component analysis(PCA)was applied to reduce the data dimensionality,creating a representative dataset.Meantime,heatmaps and SHAP(SHapley Additive exPlanations)methods were used to demonstrate the influence of key features such as electronegativity,covalent radius,period number,and unit cell volume on the bandgap,understanding the relationship between the material’s properties and performance.After comparing different machine learning models,including Random Forest(RF),Support Vector Machines(SVM),Linear Regression(LR),and Gradient Boosting Regression(GBR),the RF was found to be the most accurate for band gap prediction.Finally,the model performance was improved through parameter tuning,showing high accuracy.These findings provide strong data support and design guidance for the development of materials in fields like photocatalysis and solar cells.展开更多
Predicting monsoon climate is one of the major endeavors in climate science and is becoming increasingly challenging due to global warming. The accuracy of monsoon seasonal predictions significantly impacts the lives ...Predicting monsoon climate is one of the major endeavors in climate science and is becoming increasingly challenging due to global warming. The accuracy of monsoon seasonal predictions significantly impacts the lives of billions who depend on or are affected by monsoons, as it is essential for the water cycle, food security, ecology, disaster prevention, and the economy of monsoon regions. Given the extensive literature on Asian monsoon climate prediction, we limit our focus to reviewing the seasonal prediction and predictability of the Asian Summer Monsoon (ASM). However, much of this review is also relevant to monsoon predictions in other seasons and regions. Over the past two decades, considerable progress has been made in the seasonal forecasting of the ASM, driven by an enhanced understanding of the sources of predictability and the dynamics of seasonal variability, along with advanced development in sophisticated models and technologies. This review centers on advances in understanding the physical foundation for monsoon climate prediction (section 2), significant findings and insights into the primary and regional sources of predictability arising from feedback processes among various climate components (sections 3 and 4), the effects of global warming and external forcings on predictability (section 5), developments in seasonal prediction models and techniques (section 6), the challenges and limitations of monsoon climate prediction (section 7), and emerging research trends with suggestions for future directions (section 8). We hope this review will stimulate creative activities to enhance monsoon climate prediction.展开更多
The goal of this study was to determine the spatiotemporal characteristics of mangrove distribution and fragmentation patterns from 1988 through 2019 in Dongzhaigang.Land cover datasets were generated for Dongzhaigang...The goal of this study was to determine the spatiotemporal characteristics of mangrove distribution and fragmentation patterns from 1988 through 2019 in Dongzhaigang.Land cover datasets were generated for Dongzhaigang for multiple years via a decision tree method based on a classification and regression tree(CART)algorithm using Landsat time series images.Spatiotemporal transform and fragmentation patterns of mangrove distribution were separately assessed with a transfer matrix of land cover types and a landscape pattern index.The classification method combined with multi-band images showed good accuracy,with overall accuracy higher than 90%.Mangrove areas in 1988,1999,2009,and 2019 were 2050,1875,1818,and 1750 ha,respectively,with decreases mainly due to conversion to aquaculture ponds and farmland.A mangrove growth index(MGI)was proposed,reflecting the water-mangrove relationship,showing positive mangrove growth from 1988–2009 and negative growth from 2009–2019.Study results indicated anthropogenic factors play a leading role in the extent and scale of mangrove effects over the past 30 years.According to the analysis results,corresponding management and protection measures are proposed to provide reference for the sustainable development of Dongzhaigang Mangrove Wetland ecosystem.展开更多
The objective of the current study is to investigate an adaptive predictive observer-based autopilot for a skid-to-turn(STT)missile model with uncertainties and unknown dynamic equations.A predictive control for the S...The objective of the current study is to investigate an adaptive predictive observer-based autopilot for a skid-to-turn(STT)missile model with uncertainties and unknown dynamic equations.A predictive control for the STT missile is designed based on nonlinear model predictive control(NMPC)using Taylor series expansion,after which,via a neural network(NN),unknown functions are approximated.The present study also evaluates an adaptive optimal observer of a new strategy-based nonlinear system.Specifically,to estimate the missile states such as normal acceleration and its derivatives for the future,originally the Taylor series states expansion was gained to any specified order,based on their receding horizons.To address the problem of prediction error,an analytic solution was prepared that led to a closed form regarding the nonlinear optimal observer.Out of the gains resulting from the analytic solution,as developed for the problem of prediction error,the selection of the proposed observer gain was optimally conducted to meet the stability condition.Thus,combining the adaptive predictive autopilot and the adaptive optimal observer scheme was implemented to secure the performance,which needed only estimated normal acceleration and its derivatives.Meanwhile,no angular velocity measurement or wind angle estimation was required.Ultimately,the proposed technique was found effective,as confirmed by the qualitative simulation results.展开更多
Transit managers can use Intelligent Transportation System technologies to access large amounts of data to monitor network status.However,the presentation of the data lacks structural information.Existing single-netwo...Transit managers can use Intelligent Transportation System technologies to access large amounts of data to monitor network status.However,the presentation of the data lacks structural information.Existing single-network description technologies are ineffective in representing the temporal and spatial characteristics simultaneously.Therefore,there is a need for complementary methods to address these deficiencies.To address these limitations,this paper proposes an approach that combines Network Snapshots and Temporal Paths for the scheduled system.A dual information network is constructed to assess the degree of operational deviation considering the planning tasks.To validate the effectiveness,discussions are conducted through a modified cosine similarity calculation on theoretical analysis,delay level description,and the ability to identify abnormal dates.Compared to some state-of-the-art methods,the proposed method achieves an average Spearman delay correlation of 0.847 and a relative distance of 3.477.Furthermore,case analyses are invested in regions of China's Mainland,Europe,and the United States,investigating both the overall and sub-regional network fluctuations.To represent the impact of network fluctuations in sub-regions,a response loss value was developed.The times that are prone to fluctuations are also discussed through the classification of time series data.The research can offer a novel approach to system monitoring,providing a research direction that utilizes individual data combined to represent macroscopic states.Our code will be released at https://github.com/daozhong/STPN.git.展开更多
Pinus radiata(D.Don)dominates New Zealand's forestry industry,constituting 91%of plantations,and is among the world's most important plantation species.Given the socio-economic and environmental importance of ...Pinus radiata(D.Don)dominates New Zealand's forestry industry,constituting 91%of plantations,and is among the world's most important plantation species.Given the socio-economic and environmental importance of this species,it is important to have accurate and precise projections over time to make efficient decisions for forest management and greenfield investments in afforestation projects,especially for permanent carbon forests.Future projections of any natural resource systems rely on modeling;however,the acceleration of climate change makes future projections of yield less certain.These challenges also impact national expectations of the contribution planted forests will provide to address climate change and meet international commitments under the Paris Agreement.Using a large national-scale set of contemporary ground-measured data(2013–2023),this study investigates the performance of two growth models developed over 30 years ago that are widely used by NZ plantation growers:1)the Pumice Plateau Model 1988(PPM88)and 2)the 300-index(including a model variant of regional drift).Model simulations were made using the FORECASTER modeling suite with geographic boundaries to adjust for drift in space and time.Basal area(BA,m^(2)⋅ha^(-1))and volume(m^(3)⋅ha^(-1))were simulated,and standard errors and goodness-of-fit metrics calculated up to a typical rotation age of 30 years.Model residuals were then separated and analysed for the main plantation growing regions.The models overpredicted observed growth by between 6.8%and 16.2%,but model predictions and errors varied significantly between regions.The results of this study provided clear evidence of divergence between the outputs of both models and the measured data.Finally,this study suggests future measures to address challenges posed by these discrepancies that will provide better information for forest management and investment decisions in a changing climate.展开更多
The genetic basis of early-stage salt tolerance in alfalfa(Medicago sativa L.),a key factor limiting its productivity,remains poorly understood.To dissect this complex trait,we integrate genome-wide association studie...The genetic basis of early-stage salt tolerance in alfalfa(Medicago sativa L.),a key factor limiting its productivity,remains poorly understood.To dissect this complex trait,we integrate genome-wide association studies(GWAS)and transcriptomics from 176 accessions within a machine learning based genomic prediction framework.Analysis reveals weak genetic correlations among four salt-tolerance traits and a gradual decline in performance under increasing salt stress.GWAS identify 60 significant associated SNPs,with the highest number detected under 100 mM salt stress.Salt tolerance exhibits an additive effect from favorable haplotypes,which are most abundant in Chinese accessions.GWAS-associated genes are related to key regulators of hormone signaling and osmotic adjustment,while transcriptome analysis indicates a global repression of stress-responsive transcription factors.Integrating these multi-omics datasets allows us to identify 14 candidate genes,including MsHSD1(seed dormancy)and MsMTATP6(energy metabolism).Crucially,incorporating these markers into genomic prediction models improve cross-population predictive accuracy to an average of 54.4%.This study provides insights into the genetic architecture of salt tolerance in alfalfa and offers valuable markers to facilitate molecular breeding.展开更多
This study reveals the critical role of multiscale interaction within the westerly wind bursts(WWBs)west of the MJO convection in modulating the prediction skill for the November MJO event during the DYNAMO(Dynamics o...This study reveals the critical role of multiscale interaction within the westerly wind bursts(WWBs)west of the MJO convection in modulating the prediction skill for the November MJO event during the DYNAMO(Dynamics of the Madden–Julian Oscillation)field campaign.The characteristics of the MJO convection envelope are obtained by the largescale precipitation tracking method,and a novel metric is introduced to quantify the prediction skill for the MJO convection in the ECMWF reforecast.The ECMWF forecast exhibits approximately 17 days in skillful prediction for the MJO convection—significantly lower than that derived from the global measure.The reforecast ensembles are further classified into high and low skill catalogs based on the mean prediction skill during the observed WWBs period.High-skill ensembles exhibit significantly enhanced low-level westerlies,amplified MJO convection,and reduced spatial separation between the low-level westerlies and MJO convection during the WWBs period,indicating stronger coupling between the large-scale circulation and the convection.Mechanistic analysis reveals that enhanced westerlies in high-skill ensembles can transfer more high-frequency energy to the MJO convection through the flux convergence of interaction energy for MJO convection development,resulting in better prediction skill.展开更多
With the widespread deployment of assembly robots in smart manufacturing,efficiently offloading tasks and allocating resources in highly dynamic industrial environments has become a critical challenge for Mobile Edge ...With the widespread deployment of assembly robots in smart manufacturing,efficiently offloading tasks and allocating resources in highly dynamic industrial environments has become a critical challenge for Mobile Edge Computing(MEC).To address this challenge,this paper constructs a cloud-edge-end collaborative MEC system that enables assembly robots to offload complex workflow tasks via multiple paths(horizontal,vertical,and hybrid collaboration).Tomitigate uncertainties arising frommobility,the location predictionmodule is employed.This enables proactive channel-quality estimation,providing forward-looking insights for offloading decisions.Furthermore,we propose a fairness-aware joint optimization framework.Utilizing an improved Multi-Agent Deep Reinforcement Learning(MADRL)algorithm whose reward function incorporates total system cost,positional reliability,and timeout penalties,the framework aims to balance resource distribution among assembly robots while maximizing system utility.Simulation results demonstrate that the proposed framework outperforms traditional offloading strategies.By integrating predictive mobility management with fairness-aware optimization,the framework offers a robust solution for dynamic industrial MEC environments.展开更多
Satellite clock bias(SCB)prediction is essential for enhancing the accuracy and reliability of real-time precise point positioning(RT-PPP)in Global Navigation Satellite Systems(GNSS).To address the nonlinearity,non-st...Satellite clock bias(SCB)prediction is essential for enhancing the accuracy and reliability of real-time precise point positioning(RT-PPP)in Global Navigation Satellite Systems(GNSS).To address the nonlinearity,non-stationarity,and short-term interruptions of SCB data under complex environments,this paper proposes an enhanced SCB prediction model combining Temporal Convolutional Networks(TCN)and Transformers.Experimental results indicate that,in a 24-h prediction task,the proposed model reduces root mean square error(RMSE)and range error(RE)by 95.6%,86.0%,and 61.3%,and93.7%,86.3%,and 58.8%,respectively,compared with LSTM,Transformer,and CNN-BiGRU-Attention models,while improving computational efficiency by 48.6%over the Transformer.Moreover,although the clock bias products generated by the proposed method result in slightly higher static PPP positioning errors than the International GNSS Service(IGS)rapid clock products,the error differences are generally at the millimeter level,demonstrating the feasibility of using predicted clock bias products to replace rapid clock products in the short term.This method addresses the PPP positioning issue during short-term network service interruptions from the perspective of time series prediction and provides potential solutions for engineering applications such as landslide,earthquake,and subsidence monitoring.展开更多
Deep learning has undeniably sharpened our ability to forecast risk in neuropsychiatry[1].Yet the very success of prediction has exposed a deeper limitation:we are still remarkably uncertain about which levers to pull...Deep learning has undeniably sharpened our ability to forecast risk in neuropsychiatry[1].Yet the very success of prediction has exposed a deeper limitation:we are still remarkably uncertain about which levers to pull to change patient trajectories[2].Accurate risk scores that cannot be translated into credible actions leave clinicians where they began,testing symptomatic fixes and hoping for the best.If we want to move beyond this impasse,the next step is not simply to train larger models,but to rethink what we ask of them.展开更多
In the era of materials genome engineering,data-driven machine learning has become a powerful tool for accelerating the re-search and development of metallic materials.However,the predictive accuracy and generalizatio...In the era of materials genome engineering,data-driven machine learning has become a powerful tool for accelerating the re-search and development of metallic materials.However,the predictive accuracy and generalization ability of traditional machine learning models are often limited by the scarcity and heterogeneity of available data,especially in small-sample scenarios.To address these chal-lenges,transfer learning has emerged as an effective strategy to leverage knowledge from related domains,thereby enhancing model per-formance with limited target data.This review systematically summarizes the fundamental concepts,methodologies,and representative applications of transfer learning in the prediction of metallic materials'properties.Transfer learning can be categorized into feature-based,instance-based,parameter-based,and knowledge-based methods.This work discusses their respective mechanisms,advantages,and limit-ations.Case studies demonstrate that transfer learning can significantly improve prediction accuracy,data efficiency,and model inter-pretability in tasks such as mechanical property prediction and alloy design.Furthermore,this work highlights emerging trends including hybrid,multi-task,meta,and adaptive transfer learning,which further expand the applicability of these techniques.Finally,this work out-lines future research directions,emphasizing the need for data standardization,algorithmic innovation,multimodal data fusion,and the in-tegration of physical principles to achieve robust,interpretable,and generalizable models.The perspectives presented aim to advance the intelligent design and discovery of metallic materials,promoting efficient knowledge transfer and collaborative innovation in materials science.展开更多
The viscosity of refining slags plays a critical role in metallurgical processes.However,obtaining accurate viscosity data remains challenging due to the complexities of high-temperature experiments,often relying on e...The viscosity of refining slags plays a critical role in metallurgical processes.However,obtaining accurate viscosity data remains challenging due to the complexities of high-temperature experiments,often relying on empirical models with limited predictive capabilities.This study focuses on the influence of optical basicity on viscosity in CaO-Al_(2)O_(3)-based refining slags,leveraging machine learning to address data scarcity and improve prediction accuracy.An automated framework for algorithm integration,parameter tuning,and evaluation ranking framework(Auto-APE)is employed to develop customized data-driven models for various slag systems,including CaO-Al_(2)O_(3)-SiO_(2),CaO-Al_(2)O_(3)-CaF_(2),CaO-Al_(2)O_(3)-SiO_(2)-MgO,and CaO-Al_(2)O_(3)-SiO_(2)-MgO-CaF_(2).By incorporating optical basicity as a key feature,the models achieve an average validation error of 8.0%to 15.1%,significantly outperforming traditional empirical models.Additionally,symbolic regression is introduced to rapidly construct domain-specific features,such as optical basicity-like descriptors,offering a potential breakthrough in performance prediction for small datasets.This work highlights the critical role of domain-specific knowledge in understanding and predicting viscosity,providing a robust machine learning-based approach for optimizing refining slag properties.展开更多
Thermal power plants are the main contributors to greenhouse gas emissions.The prediction of the emission supports the decision makers and environmental sustainability.The objective of this study is to enhance the acc...Thermal power plants are the main contributors to greenhouse gas emissions.The prediction of the emission supports the decision makers and environmental sustainability.The objective of this study is to enhance the accuracy of emission prediction models,supporting more effective real-time monitoring and enabling informed operational decisions that align with environmental compliance efforts.This paper presents a data-driven approach for the accurate prediction of gas emissions,specifically nitrogen oxides(NOx)and carbon monoxide(CO),in natural gas power plants using an optimized hybrid machine learning framework.The proposed model integrates a Feedforward Neural Network(FFNN)trained using Particle Swarm Optimization to capture the nonlinear emission dynamics under varying gas turbine operating conditions.To further enhance predictive performance,the K-Nearest Neighbor(K-NN)algorithm serves as a post-processing method to enhance IPSO-FFNN predictions through adjustment and refinement,improving overall prediction accuracy,while Neighbor Component Analysis is used to identify and rank the most influential operational variables.The study makes a significant contribution through the combination of NCA feature selection with PSO global optimization,FFNN nonlinear modelling,and K-NN error correction into one unified system,which delivers precise emission predictions.The model was developed and tested using a real-world dataset collected from gas-fired turbine operations,with validated results demonstrating robust accuracy,achieving Root Mean Square Error values of 0.355 for CO and 0.368 for NOx.When benchmarked against conventional models such as standard FFNN,Support Vector Regression,and Long Short-Term Memory networks,the hybrid model achieved substantial improvements,up to 97.8%in Mean Squared Error,95%in Mean Absolute Error(MAE),and 85.19%in RMSE for CO;and 97.16%in MSE,93.4%in MAE,and 83.15%in RMSE for NOx.These results underscore the model’s potential for improving emission prediction,thereby supporting enhanced operational efficiency and adherence to environmental standards.展开更多
Precise forecasts of wildfire danger are crucial for proactive fuel management and emergency responses,yet they pose a challenge at the subseasonal scale due to limitations in prediction capabilities and a gap between...Precise forecasts of wildfire danger are crucial for proactive fuel management and emergency responses,yet they pose a challenge at the subseasonal scale due to limitations in prediction capabilities and a gap between forecast outputs and the needs of decision-makers.This study introduces an innovative hybrid modeling framework that integrates artificial intelligence(AI)with climate dynamic prediction systems to accurately forecast High Fire-Danger Days(HFDDs)for the following month.These HFDDs are derived from historical satellite fire data and the optimum fire danger index,with a particular focus on Southwest China as a case study.The AI module,based on the ResNet-18 neural network model,integrates observational and physically constrained analysis to establish links between HFDDs and optimal predictors of atmospheric circulation from both the concurrent and preceding months.Leveraging climate dynamical forecasting,this hybrid model provides more reliable deterministic predictions for monthly HFDDs than conventional methods that rely solely on terrestrial variables such as precipitation.More importantly,the integration of dynamical ensemble prediction enhances the model’s capability for skillful probabilistic predictions of HFDDs,facilitating the creation of customized fire danger outlooks and emergency action maps tailored to stakeholders’needs.The model’s added economic value was also evaluated,demonstrating its potential to improve decision-making in disaster management and bridge the“last-mile gap”in climate service delivery.This work contributes to the Seamless Prediction and Services for Sustainable Natural and Built Environment(SEPRESS)Program(2025–32),under the United Nations Educational Scientific and Cultural Organization(UNESCO)International Decade of Sciences for Sustainable Development(2024–33).展开更多
Anthropogenic ammonia emissions primarily originate from agriculture,especially field fertilization.These emissions represent nitrogen loss for farmers and contribute to air pollution,posing risks to human health and ...Anthropogenic ammonia emissions primarily originate from agriculture,especially field fertilization.These emissions represent nitrogen loss for farmers and contribute to air pollution,posing risks to human health and the environment.Estimating ammonia emissions is crucial for national inventories and policy-making.Various models exist for predicting emissions,including mechanistic,empirical,and semi-empirical approaches.While machine learning(ML)is widely used in environmental science,its application to ammonia emissions remains limited.In this study,we used 5939 ammonia emission data from 538 trials,extracted from the ALFAM2 database,to train three machine learning methods-random forest,gradient boosting,and lasso-for predicting cumulative ammonia emissions 72 h after manure application.These methods were compared to the semi-empirical ALFAM2 model using an independent test dataset.Random forest(RMSE=4.51,r=0.94,MAE=3.28,Bias=0.92)and gradient boosting(RMSE=6.19,r=0.89,MAE=4.10,Bias=0.51)showed the best performance,while the lasso log-linear model(RMSE=7.30,r=0.84,MAE=5.57,Bias=-1.38)performed worst.Both random forest and gradient boosting outperformed the semi-empirical ALFAM2 model,which showed performance comparable to the lasso model.We then used these models and the ALFAM2 model to compare five slurry management techniques,varying in application method(trailing hoses,trailing shoes,and open slot)and post-application incorporation,across 128 scenarios with different manure types and weather conditions.Compared to broadcast application,alternative techniques reduced emissions by a median of-13.6%to-61.7%.This study highlights the promise of ML models in assessing ammonia emission reduction methods,while emphasizing the importance of evaluating model sensitivity to algorithm choice.展开更多
Customer churn is the rate at which customers discontinue doing business with a company over a given time period.It is an essential measure for businesses to monitor high churn rates,as they often indicate underlying ...Customer churn is the rate at which customers discontinue doing business with a company over a given time period.It is an essential measure for businesses to monitor high churn rates,as they often indicate underlying issues with services,products,or customer experience,resulting in considerable income loss.Prediction of customer churn is a crucial task aimed at retaining customers and maintaining revenue growth.Traditional machine learning(ML)models often struggle to capture complex temporal dependencies in client behavior data.To address this,an optimized deep learning(DL)approach using a Regularized Bidirectional Long Short-Term Memory(RBiLSTM)model is proposed to mitigate overfitting and improve generalization error.The model integrates dropout,L2-regularization,and early stopping to enhance predictive accuracy while preventing over-reliance on specific patterns.Moreover,this study investigates the effect of optimization techniques on boosting the training efficiency of the developed model.Experimental results on a recent public customer churn dataset demonstrate that the trained model outperforms the traditional ML models and some other DL models,such as Long Short-Term Memory(LSTM)and Deep Neural Network(DNN),in churn prediction performance and stability.The proposed approach achieves 96.1%accuracy,compared with LSTM and DNN,which attain 94.5%and 94.1%accuracy,respectively.These results confirm that the proposed approach can be used as a valuable tool for businesses to identify at-risk consumers proactively and implement targeted retention strategies.展开更多
基金supported by the National Natural Science Foundation of China(Grant Nos.62472149,62376089,62202147)Hubei Provincial Science and Technology Plan Project(2023BCB04100).
文摘Accurate traffic flow prediction has a profound impact on modern traffic management. Traffic flow has complex spatial-temporal correlations and periodicity, which poses difficulties for precise prediction. To address this problem, a Multi-head Self-attention and Spatial-Temporal Graph Convolutional Network (MSSTGCN) for multiscale traffic flow prediction is proposed. Firstly, to capture the hidden traffic periodicity of traffic flow, traffic flow is divided into three kinds of periods, including hourly, daily, and weekly data. Secondly, a graph attention residual layer is constructed to learn the global spatial features across regions. Local spatial-temporal dependence is captured by using a T-GCN module. Thirdly, a transformer layer is introduced to learn the long-term dependence in time. A position embedding mechanism is introduced to label position information for all traffic sequences. Thus, this multi-head self-attention mechanism can recognize the sequence order and allocate weights for different time nodes. Experimental results on four real-world datasets show that the MSSTGCN performs better than the baseline methods and can be successfully adapted to traffic prediction tasks.
基金supported by the Beijing Natural Science Foundation(Certificate Number:L234025).
文摘Spatial-temporal traffic prediction technology is crucial for network planning,resource allocation optimizing,and user experience improving.With the development of virtual network operators,multi-operator collaborations,and edge computing,spatial-temporal traffic data has taken on a distributed nature.Consequently,noncentralized spatial-temporal traffic prediction solutions have emerged as a recent research focus.Currently,the majority of research typically adopts federated learning methods to train traffic prediction models distributed on each base station.This method reduces additional burden on communication systems.However,this method has a drawback:it cannot handle irregular traffic data.Due to unstable wireless network environments,device failures,insufficient storage resources,etc.,data missing inevitably occurs during the process of collecting traffic data.This results in the irregular nature of distributed traffic data.Yet,commonly used traffic prediction models such as Recurrent Neural Networks(RNN)and Long Short-Term Memory(LSTM)typically assume that the data is complete and regular.To address the challenge of handling irregular traffic data,this paper transforms irregular traffic prediction into problems of estimating latent variables and generating future traffic.To solve the aforementioned problems,this paper introduces split learning to design a structured distributed learning framework.The framework comprises a Global-level Spatial structure mining Model(GSM)and several Nodelevel Generative Models(NGMs).NGM and GSM represent Seq2Seq models deployed on the base station and graph neural network models deployed on the cloud or central controller.Firstly,the time embedding layer in NGM establishes the mapping relationship between irregular traffic data and regular latent temporal feature variables.Secondly,GSM collects statistical feature parameters of latent temporal feature variables from various nodes and executes graph embedding for spatial-temporal traffic data.Finally,NGM generates future traffic based on latent temporal and spatial feature variables.The introduction of the time attention mechanism enhances the framework’s capability to handle irregular traffic data.Graph attention network introduces spatially correlated base station traffic feature information into local traffic prediction,which compensates for missing information in local irregular traffic data.The proposed framework effectively addresses the distributed prediction issues of irregular traffic data.By testing on real world datasets,the proposed framework improves traffic prediction accuracy by 35%compared to other commonly used distributed traffic prediction methods.
基金supported by the National Natural Science Foundation of China(No.42061065)the Third Xinjiang Comprehensive Scientific Expedition,China(No.2022xjkk03010102).
文摘Root zone soil moisture(RZSM)plays a critical role in land-atmosphere hydrological cycles and serves as the primary water source for vegetation growth.However,the correlations between RZSM and its associated variables,including surface soil moisture(SSM),often exhibit nonlinearities that are challenging to identify and quantify using conventional statistical techniques.Therefore,this study presents a hybrid convolutional neural network(CNN)-long short-term memory neural network(LSTM)-attention(CLA)model for predicting RZSM.Owing to the scarcity of soil moisture(SM)observation data,the physical model Hydrus-1D was employed to simulate a comprehensive dataset of spatial-temporal SM.Meteorological data and moderate resolution imaging spectroradiometer vegetation characterization parameters were used as predictor variables for the training and validation of the CLA model.The results of the CLA model for SM prediction in the root zone were significantly enhanced compared with those of the traditional LSTM and CNN-LSTM models.This was particularly notable at the depth of 80–100 cm,where the fitness(R^(2))reached nearly 0.9298.Moreover,the root mean square error of the CLA model was reduced by 49%and 57%compared with those of the LSTM and CNN-LSTM models,respectively.This study demonstrates that the integration of physical modeling and deep learning methods provides a more comprehensive and accurate understanding of spatial-temporal SM variations in the root zone.
文摘Titanium-based semiconductors are known for their high chemical stability and suitable band gap widths.However,the conventional experimental screening methods are inefficient due to the wide variety of materials.To speed up the selection process,this work focuses on interpretable feature learning and band gap prediction for titanium-based semiconductors.First,titanium compounds were selected from the Materials Project database by machine learning,and elemental features were extracted using the Magpie descriptors.Then,principal component analysis(PCA)was applied to reduce the data dimensionality,creating a representative dataset.Meantime,heatmaps and SHAP(SHapley Additive exPlanations)methods were used to demonstrate the influence of key features such as electronegativity,covalent radius,period number,and unit cell volume on the bandgap,understanding the relationship between the material’s properties and performance.After comparing different machine learning models,including Random Forest(RF),Support Vector Machines(SVM),Linear Regression(LR),and Gradient Boosting Regression(GBR),the RF was found to be the most accurate for band gap prediction.Finally,the model performance was improved through parameter tuning,showing high accuracy.These findings provide strong data support and design guidance for the development of materials in fields like photocatalysis and solar cells.
基金supported by the National Natural Science Foundation of China(Grant No.U2342208)support from NSF/Climate Dynamics Award#2025057。
文摘Predicting monsoon climate is one of the major endeavors in climate science and is becoming increasingly challenging due to global warming. The accuracy of monsoon seasonal predictions significantly impacts the lives of billions who depend on or are affected by monsoons, as it is essential for the water cycle, food security, ecology, disaster prevention, and the economy of monsoon regions. Given the extensive literature on Asian monsoon climate prediction, we limit our focus to reviewing the seasonal prediction and predictability of the Asian Summer Monsoon (ASM). However, much of this review is also relevant to monsoon predictions in other seasons and regions. Over the past two decades, considerable progress has been made in the seasonal forecasting of the ASM, driven by an enhanced understanding of the sources of predictability and the dynamics of seasonal variability, along with advanced development in sophisticated models and technologies. This review centers on advances in understanding the physical foundation for monsoon climate prediction (section 2), significant findings and insights into the primary and regional sources of predictability arising from feedback processes among various climate components (sections 3 and 4), the effects of global warming and external forcings on predictability (section 5), developments in seasonal prediction models and techniques (section 6), the challenges and limitations of monsoon climate prediction (section 7), and emerging research trends with suggestions for future directions (section 8). We hope this review will stimulate creative activities to enhance monsoon climate prediction.
基金financially supported by the National Natural Science Foundation of China(Nos.U2244225 and 42020104005)the Ministry of Education of China(111 Project)the Fundamental Research Funds for the Central Universities,China University of Geosciences(Wuhan)and China Geological Survey(No.DD20211391)。
文摘The goal of this study was to determine the spatiotemporal characteristics of mangrove distribution and fragmentation patterns from 1988 through 2019 in Dongzhaigang.Land cover datasets were generated for Dongzhaigang for multiple years via a decision tree method based on a classification and regression tree(CART)algorithm using Landsat time series images.Spatiotemporal transform and fragmentation patterns of mangrove distribution were separately assessed with a transfer matrix of land cover types and a landscape pattern index.The classification method combined with multi-band images showed good accuracy,with overall accuracy higher than 90%.Mangrove areas in 1988,1999,2009,and 2019 were 2050,1875,1818,and 1750 ha,respectively,with decreases mainly due to conversion to aquaculture ponds and farmland.A mangrove growth index(MGI)was proposed,reflecting the water-mangrove relationship,showing positive mangrove growth from 1988–2009 and negative growth from 2009–2019.Study results indicated anthropogenic factors play a leading role in the extent and scale of mangrove effects over the past 30 years.According to the analysis results,corresponding management and protection measures are proposed to provide reference for the sustainable development of Dongzhaigang Mangrove Wetland ecosystem.
文摘The objective of the current study is to investigate an adaptive predictive observer-based autopilot for a skid-to-turn(STT)missile model with uncertainties and unknown dynamic equations.A predictive control for the STT missile is designed based on nonlinear model predictive control(NMPC)using Taylor series expansion,after which,via a neural network(NN),unknown functions are approximated.The present study also evaluates an adaptive optimal observer of a new strategy-based nonlinear system.Specifically,to estimate the missile states such as normal acceleration and its derivatives for the future,originally the Taylor series states expansion was gained to any specified order,based on their receding horizons.To address the problem of prediction error,an analytic solution was prepared that led to a closed form regarding the nonlinear optimal observer.Out of the gains resulting from the analytic solution,as developed for the problem of prediction error,the selection of the proposed observer gain was optimally conducted to meet the stability condition.Thus,combining the adaptive predictive autopilot and the adaptive optimal observer scheme was implemented to secure the performance,which needed only estimated normal acceleration and its derivatives.Meanwhile,no angular velocity measurement or wind angle estimation was required.Ultimately,the proposed technique was found effective,as confirmed by the qualitative simulation results.
文摘Transit managers can use Intelligent Transportation System technologies to access large amounts of data to monitor network status.However,the presentation of the data lacks structural information.Existing single-network description technologies are ineffective in representing the temporal and spatial characteristics simultaneously.Therefore,there is a need for complementary methods to address these deficiencies.To address these limitations,this paper proposes an approach that combines Network Snapshots and Temporal Paths for the scheduled system.A dual information network is constructed to assess the degree of operational deviation considering the planning tasks.To validate the effectiveness,discussions are conducted through a modified cosine similarity calculation on theoretical analysis,delay level description,and the ability to identify abnormal dates.Compared to some state-of-the-art methods,the proposed method achieves an average Spearman delay correlation of 0.847 and a relative distance of 3.477.Furthermore,case analyses are invested in regions of China's Mainland,Europe,and the United States,investigating both the overall and sub-regional network fluctuations.To represent the impact of network fluctuations in sub-regions,a response loss value was developed.The times that are prone to fluctuations are also discussed through the classification of time series data.The research can offer a novel approach to system monitoring,providing a research direction that utilizes individual data combined to represent macroscopic states.Our code will be released at https://github.com/daozhong/STPN.git.
基金funded by Scion's Strategic Science Investment Fund(SSIF)the Forest Growers Levy Trust(FGLT)through the Resilient Forests Programme(Task No.A89220)。
文摘Pinus radiata(D.Don)dominates New Zealand's forestry industry,constituting 91%of plantations,and is among the world's most important plantation species.Given the socio-economic and environmental importance of this species,it is important to have accurate and precise projections over time to make efficient decisions for forest management and greenfield investments in afforestation projects,especially for permanent carbon forests.Future projections of any natural resource systems rely on modeling;however,the acceleration of climate change makes future projections of yield less certain.These challenges also impact national expectations of the contribution planted forests will provide to address climate change and meet international commitments under the Paris Agreement.Using a large national-scale set of contemporary ground-measured data(2013–2023),this study investigates the performance of two growth models developed over 30 years ago that are widely used by NZ plantation growers:1)the Pumice Plateau Model 1988(PPM88)and 2)the 300-index(including a model variant of regional drift).Model simulations were made using the FORECASTER modeling suite with geographic boundaries to adjust for drift in space and time.Basal area(BA,m^(2)⋅ha^(-1))and volume(m^(3)⋅ha^(-1))were simulated,and standard errors and goodness-of-fit metrics calculated up to a typical rotation age of 30 years.Model residuals were then separated and analysed for the main plantation growing regions.The models overpredicted observed growth by between 6.8%and 16.2%,but model predictions and errors varied significantly between regions.The results of this study provided clear evidence of divergence between the outputs of both models and the measured data.Finally,this study suggests future measures to address challenges posed by these discrepancies that will provide better information for forest management and investment decisions in a changing climate.
基金supported by the National Key Research and Development Program of China(2022YFF1003203)Biological Breeding-National Science and Technology Major Project(2022ZDo4011)+2 种基金the Central Public-interest Scientific Institution Basal Research Fund(Y2025YC44)the Central Public-interest Scientific Institution Basal Research Fund(2025-YWF-ZYSQ-04)the China Postdoctoral Science Foundation(2023M733832).
文摘The genetic basis of early-stage salt tolerance in alfalfa(Medicago sativa L.),a key factor limiting its productivity,remains poorly understood.To dissect this complex trait,we integrate genome-wide association studies(GWAS)and transcriptomics from 176 accessions within a machine learning based genomic prediction framework.Analysis reveals weak genetic correlations among four salt-tolerance traits and a gradual decline in performance under increasing salt stress.GWAS identify 60 significant associated SNPs,with the highest number detected under 100 mM salt stress.Salt tolerance exhibits an additive effect from favorable haplotypes,which are most abundant in Chinese accessions.GWAS-associated genes are related to key regulators of hormone signaling and osmotic adjustment,while transcriptome analysis indicates a global repression of stress-responsive transcription factors.Integrating these multi-omics datasets allows us to identify 14 candidate genes,including MsHSD1(seed dormancy)and MsMTATP6(energy metabolism).Crucially,incorporating these markers into genomic prediction models improve cross-population predictive accuracy to an average of 54.4%.This study provides insights into the genetic architecture of salt tolerance in alfalfa and offers valuable markers to facilitate molecular breeding.
基金sponsored by the National Natural Science Foundation of China(Grant Nos.U2442206,42205067,and 41922035)the National Key R&D Program of China(Grant No.2024YFC3013100)the Key Research Program of Frontier Sciences of CAS(Grant No.QYZDB-SSW-DQC017).
文摘This study reveals the critical role of multiscale interaction within the westerly wind bursts(WWBs)west of the MJO convection in modulating the prediction skill for the November MJO event during the DYNAMO(Dynamics of the Madden–Julian Oscillation)field campaign.The characteristics of the MJO convection envelope are obtained by the largescale precipitation tracking method,and a novel metric is introduced to quantify the prediction skill for the MJO convection in the ECMWF reforecast.The ECMWF forecast exhibits approximately 17 days in skillful prediction for the MJO convection—significantly lower than that derived from the global measure.The reforecast ensembles are further classified into high and low skill catalogs based on the mean prediction skill during the observed WWBs period.High-skill ensembles exhibit significantly enhanced low-level westerlies,amplified MJO convection,and reduced spatial separation between the low-level westerlies and MJO convection during the WWBs period,indicating stronger coupling between the large-scale circulation and the convection.Mechanistic analysis reveals that enhanced westerlies in high-skill ensembles can transfer more high-frequency energy to the MJO convection through the flux convergence of interaction energy for MJO convection development,resulting in better prediction skill.
基金supported by the National Key R&D Program of China under Grant Nos.2024YFD2400200 and 2024YFD2400204supported in part by the Science and Technology Development Program for the Two Zones under Grant No.2023LQ02004.
文摘With the widespread deployment of assembly robots in smart manufacturing,efficiently offloading tasks and allocating resources in highly dynamic industrial environments has become a critical challenge for Mobile Edge Computing(MEC).To address this challenge,this paper constructs a cloud-edge-end collaborative MEC system that enables assembly robots to offload complex workflow tasks via multiple paths(horizontal,vertical,and hybrid collaboration).Tomitigate uncertainties arising frommobility,the location predictionmodule is employed.This enables proactive channel-quality estimation,providing forward-looking insights for offloading decisions.Furthermore,we propose a fairness-aware joint optimization framework.Utilizing an improved Multi-Agent Deep Reinforcement Learning(MADRL)algorithm whose reward function incorporates total system cost,positional reliability,and timeout penalties,the framework aims to balance resource distribution among assembly robots while maximizing system utility.Simulation results demonstrate that the proposed framework outperforms traditional offloading strategies.By integrating predictive mobility management with fairness-aware optimization,the framework offers a robust solution for dynamic industrial MEC environments.
基金supported by the National Natural Science Foundation of China(42304050)Major Science and Technology Projects in Anhui Province,grant number(202103a05020026)+1 种基金Open Foundation of the Key Laboratory of Universities in Anhui Province for Prevention of Mine Geological Disasters(2022-MGDP-08)University Natural Science Research Project of Anhui Province(2023AH051190)。
文摘Satellite clock bias(SCB)prediction is essential for enhancing the accuracy and reliability of real-time precise point positioning(RT-PPP)in Global Navigation Satellite Systems(GNSS).To address the nonlinearity,non-stationarity,and short-term interruptions of SCB data under complex environments,this paper proposes an enhanced SCB prediction model combining Temporal Convolutional Networks(TCN)and Transformers.Experimental results indicate that,in a 24-h prediction task,the proposed model reduces root mean square error(RMSE)and range error(RE)by 95.6%,86.0%,and 61.3%,and93.7%,86.3%,and 58.8%,respectively,compared with LSTM,Transformer,and CNN-BiGRU-Attention models,while improving computational efficiency by 48.6%over the Transformer.Moreover,although the clock bias products generated by the proposed method result in slightly higher static PPP positioning errors than the International GNSS Service(IGS)rapid clock products,the error differences are generally at the millimeter level,demonstrating the feasibility of using predicted clock bias products to replace rapid clock products in the short term.This method addresses the PPP positioning issue during short-term network service interruptions from the perspective of time series prediction and provides potential solutions for engineering applications such as landslide,earthquake,and subsidence monitoring.
文摘Deep learning has undeniably sharpened our ability to forecast risk in neuropsychiatry[1].Yet the very success of prediction has exposed a deeper limitation:we are still remarkably uncertain about which levers to pull to change patient trajectories[2].Accurate risk scores that cannot be translated into credible actions leave clinicians where they began,testing symptomatic fixes and hoping for the best.If we want to move beyond this impasse,the next step is not simply to train larger models,but to rethink what we ask of them.
基金supported by the National NaturalScience Foundation of China(Nos.52301029 and 52274359)the Fundamental Research Funds for the CentralUniversities,China(No.06500165)+2 种基金the Guangdong Basicand Applied Basic Research Foundation,China(No.2022A1515140006)Young Elite Scientists Sponsorship Program by CAST(No.2023QNRC001)Beijing Young Elite Scientists Sponsorship Program by BMES,China.
文摘In the era of materials genome engineering,data-driven machine learning has become a powerful tool for accelerating the re-search and development of metallic materials.However,the predictive accuracy and generalization ability of traditional machine learning models are often limited by the scarcity and heterogeneity of available data,especially in small-sample scenarios.To address these chal-lenges,transfer learning has emerged as an effective strategy to leverage knowledge from related domains,thereby enhancing model per-formance with limited target data.This review systematically summarizes the fundamental concepts,methodologies,and representative applications of transfer learning in the prediction of metallic materials'properties.Transfer learning can be categorized into feature-based,instance-based,parameter-based,and knowledge-based methods.This work discusses their respective mechanisms,advantages,and limit-ations.Case studies demonstrate that transfer learning can significantly improve prediction accuracy,data efficiency,and model inter-pretability in tasks such as mechanical property prediction and alloy design.Furthermore,this work highlights emerging trends including hybrid,multi-task,meta,and adaptive transfer learning,which further expand the applicability of these techniques.Finally,this work out-lines future research directions,emphasizing the need for data standardization,algorithmic innovation,multimodal data fusion,and the in-tegration of physical principles to achieve robust,interpretable,and generalizable models.The perspectives presented aim to advance the intelligent design and discovery of metallic materials,promoting efficient knowledge transfer and collaborative innovation in materials science.
基金supported by the National Key Research and Development Program of China(No.2023YFB3712401),the National Natural Science Foundation of China(No.52274301)the Aeronautical Science Foundation of China(No.2023Z0530S6005)the Ningbo Yongjiang Talent-Introduction Programme(No.2022A-023-C).
文摘The viscosity of refining slags plays a critical role in metallurgical processes.However,obtaining accurate viscosity data remains challenging due to the complexities of high-temperature experiments,often relying on empirical models with limited predictive capabilities.This study focuses on the influence of optical basicity on viscosity in CaO-Al_(2)O_(3)-based refining slags,leveraging machine learning to address data scarcity and improve prediction accuracy.An automated framework for algorithm integration,parameter tuning,and evaluation ranking framework(Auto-APE)is employed to develop customized data-driven models for various slag systems,including CaO-Al_(2)O_(3)-SiO_(2),CaO-Al_(2)O_(3)-CaF_(2),CaO-Al_(2)O_(3)-SiO_(2)-MgO,and CaO-Al_(2)O_(3)-SiO_(2)-MgO-CaF_(2).By incorporating optical basicity as a key feature,the models achieve an average validation error of 8.0%to 15.1%,significantly outperforming traditional empirical models.Additionally,symbolic regression is introduced to rapidly construct domain-specific features,such as optical basicity-like descriptors,offering a potential breakthrough in performance prediction for small datasets.This work highlights the critical role of domain-specific knowledge in understanding and predicting viscosity,providing a robust machine learning-based approach for optimizing refining slag properties.
文摘Thermal power plants are the main contributors to greenhouse gas emissions.The prediction of the emission supports the decision makers and environmental sustainability.The objective of this study is to enhance the accuracy of emission prediction models,supporting more effective real-time monitoring and enabling informed operational decisions that align with environmental compliance efforts.This paper presents a data-driven approach for the accurate prediction of gas emissions,specifically nitrogen oxides(NOx)and carbon monoxide(CO),in natural gas power plants using an optimized hybrid machine learning framework.The proposed model integrates a Feedforward Neural Network(FFNN)trained using Particle Swarm Optimization to capture the nonlinear emission dynamics under varying gas turbine operating conditions.To further enhance predictive performance,the K-Nearest Neighbor(K-NN)algorithm serves as a post-processing method to enhance IPSO-FFNN predictions through adjustment and refinement,improving overall prediction accuracy,while Neighbor Component Analysis is used to identify and rank the most influential operational variables.The study makes a significant contribution through the combination of NCA feature selection with PSO global optimization,FFNN nonlinear modelling,and K-NN error correction into one unified system,which delivers precise emission predictions.The model was developed and tested using a real-world dataset collected from gas-fired turbine operations,with validated results demonstrating robust accuracy,achieving Root Mean Square Error values of 0.355 for CO and 0.368 for NOx.When benchmarked against conventional models such as standard FFNN,Support Vector Regression,and Long Short-Term Memory networks,the hybrid model achieved substantial improvements,up to 97.8%in Mean Squared Error,95%in Mean Absolute Error(MAE),and 85.19%in RMSE for CO;and 97.16%in MSE,93.4%in MAE,and 83.15%in RMSE for NOx.These results underscore the model’s potential for improving emission prediction,thereby supporting enhanced operational efficiency and adherence to environmental standards.
基金J.YANG was supported by funding from the National Natural Science Foundation of China(Grant Nos.42475022,42261144671)the National Key R&D Program of China(Project No.2024YFC3013100)+2 种基金the Fundamental Research Funds for the Central UniversitiesM.LU was supported by the Otto Poon Centre of Climate Resilience and Sustainability at HKUST and the Hong Kong Research Grant Committee(Project No.16300424)Data processing and storage were supported by the National Key Scientific and Technological Infrastructure project“Earth System Numerical Simulation Facility”(EarthLab).
文摘Precise forecasts of wildfire danger are crucial for proactive fuel management and emergency responses,yet they pose a challenge at the subseasonal scale due to limitations in prediction capabilities and a gap between forecast outputs and the needs of decision-makers.This study introduces an innovative hybrid modeling framework that integrates artificial intelligence(AI)with climate dynamic prediction systems to accurately forecast High Fire-Danger Days(HFDDs)for the following month.These HFDDs are derived from historical satellite fire data and the optimum fire danger index,with a particular focus on Southwest China as a case study.The AI module,based on the ResNet-18 neural network model,integrates observational and physically constrained analysis to establish links between HFDDs and optimal predictors of atmospheric circulation from both the concurrent and preceding months.Leveraging climate dynamical forecasting,this hybrid model provides more reliable deterministic predictions for monthly HFDDs than conventional methods that rely solely on terrestrial variables such as precipitation.More importantly,the integration of dynamical ensemble prediction enhances the model’s capability for skillful probabilistic predictions of HFDDs,facilitating the creation of customized fire danger outlooks and emergency action maps tailored to stakeholders’needs.The model’s added economic value was also evaluated,demonstrating its potential to improve decision-making in disaster management and bridge the“last-mile gap”in climate service delivery.This work contributes to the Seamless Prediction and Services for Sustainable Natural and Built Environment(SEPRESS)Program(2025–32),under the United Nations Educational Scientific and Cultural Organization(UNESCO)International Decade of Sciences for Sustainable Development(2024–33).
基金the French state aid managed by the ANR under the“Investissements d’avenir”programme with the reference ANR-16-CONV-0003from the AgroEcoSystem department of INRAE.We are grateful to the INRAE MIGALE bioinformatics facility(MIGALE,INRAE,2020.Migale bioinformatics Facility,doi:10.15454/1.5572390655343293E12)for providing help and/or computing and/or storage resources.We are also grateful to Sasha Hafner for his help in reproducing some of the results of Hafner et al.(2019).
文摘Anthropogenic ammonia emissions primarily originate from agriculture,especially field fertilization.These emissions represent nitrogen loss for farmers and contribute to air pollution,posing risks to human health and the environment.Estimating ammonia emissions is crucial for national inventories and policy-making.Various models exist for predicting emissions,including mechanistic,empirical,and semi-empirical approaches.While machine learning(ML)is widely used in environmental science,its application to ammonia emissions remains limited.In this study,we used 5939 ammonia emission data from 538 trials,extracted from the ALFAM2 database,to train three machine learning methods-random forest,gradient boosting,and lasso-for predicting cumulative ammonia emissions 72 h after manure application.These methods were compared to the semi-empirical ALFAM2 model using an independent test dataset.Random forest(RMSE=4.51,r=0.94,MAE=3.28,Bias=0.92)and gradient boosting(RMSE=6.19,r=0.89,MAE=4.10,Bias=0.51)showed the best performance,while the lasso log-linear model(RMSE=7.30,r=0.84,MAE=5.57,Bias=-1.38)performed worst.Both random forest and gradient boosting outperformed the semi-empirical ALFAM2 model,which showed performance comparable to the lasso model.We then used these models and the ALFAM2 model to compare five slurry management techniques,varying in application method(trailing hoses,trailing shoes,and open slot)and post-application incorporation,across 128 scenarios with different manure types and weather conditions.Compared to broadcast application,alternative techniques reduced emissions by a median of-13.6%to-61.7%.This study highlights the promise of ML models in assessing ammonia emission reduction methods,while emphasizing the importance of evaluating model sensitivity to algorithm choice.
文摘Customer churn is the rate at which customers discontinue doing business with a company over a given time period.It is an essential measure for businesses to monitor high churn rates,as they often indicate underlying issues with services,products,or customer experience,resulting in considerable income loss.Prediction of customer churn is a crucial task aimed at retaining customers and maintaining revenue growth.Traditional machine learning(ML)models often struggle to capture complex temporal dependencies in client behavior data.To address this,an optimized deep learning(DL)approach using a Regularized Bidirectional Long Short-Term Memory(RBiLSTM)model is proposed to mitigate overfitting and improve generalization error.The model integrates dropout,L2-regularization,and early stopping to enhance predictive accuracy while preventing over-reliance on specific patterns.Moreover,this study investigates the effect of optimization techniques on boosting the training efficiency of the developed model.Experimental results on a recent public customer churn dataset demonstrate that the trained model outperforms the traditional ML models and some other DL models,such as Long Short-Term Memory(LSTM)and Deep Neural Network(DNN),in churn prediction performance and stability.The proposed approach achieves 96.1%accuracy,compared with LSTM and DNN,which attain 94.5%and 94.1%accuracy,respectively.These results confirm that the proposed approach can be used as a valuable tool for businesses to identify at-risk consumers proactively and implement targeted retention strategies.
