In order to address the issue of overly conservative offline reinforcement learning(RL) methods that limit the generalization of policy in the out-of-distribution(OOD) region,this article designs a surrogate target fo...In order to address the issue of overly conservative offline reinforcement learning(RL) methods that limit the generalization of policy in the out-of-distribution(OOD) region,this article designs a surrogate target for OOD value function based on dataset distance and proposes a novel generalized Q-learning mechanism with distance regularization(GQDR).In theory,we not only prove the convergence of GQDR,but also ensure that the difference between the Q-value learned by GQDR and its true value is bounded.Furthermore,an offline generalized actor-critic method with distance regularization(OGACDR) is proposed by combining GQDR with actor-critic learning framework.Two implementations of OGACDR,OGACDR-EXP and OGACDRSQR,are introduced according to exponential(EXP) and opensquare(SQR) distance weight functions,and it has been theoretically proved that OGACDR provides a safe policy improvement.Experimental results on Gym-MuJoCo continuous control tasks show that OGACDR can not only alleviate the overestimation and overconservatism of Q-value function,but also outperform conservative offline RL baselines.展开更多
Offline policy evaluation,evaluating and selecting complex policies for decision-making by only using offline datasets is important in reinforcement learning.At present,the model-based offline policy evaluation(MBOPE)...Offline policy evaluation,evaluating and selecting complex policies for decision-making by only using offline datasets is important in reinforcement learning.At present,the model-based offline policy evaluation(MBOPE)is widely welcomed because of its easy to implement and good performance.MBOPE directly approximates the unknown value of a given policy using the Monte Carlo method given the estimated transition and reward functions of the environment.Usually,multiple models are trained,and then one of them is selected to be used.However,a challenge remains in selecting an appropriate model from those trained for further use.The authors first analyse the upper bound of the difference between the approximated value and the unknown true value.Theoretical results show that this difference is related to the trajectories generated by the given policy on the learnt model and the prediction error of the transition and reward functions at these generated data points.Based on the theoretical results,a new criterion is proposed to tell which trained model is better suited for evaluating the given policy.At last,the effectiveness of the proposed criterion is demonstrated on both benchmark and synthetic offline datasets.展开更多
This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mi...This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mission is disturbed by the measurement noise and the target adversarial behavior.To address these problems,a model-free Combined Adaptive-length Datadriven Predictive Controller(CADPC)is proposed.It consists of a separated subsystem identification method and a combined predictive control strategy.The subsystem identification method is composed of an adaptive data length,thereby reducing sensitivity to undetermined measurement noises and disturbances.Based on the subsystem identification,the combined predictive controller is established,reducing calculating resource.The stability of the CADPC is rigorously proven using the Input-to-State Stable(ISS)theorem and the small-gain theorem.Simulations demonstrate that CADPC effectively handles the model-free space robot post operation in the presence of significant disturbances,state measurement noise,and control input errors.It achieves improved steady-state accuracy,reduced steady-state control consumption,and minimized control input chattering.展开更多
When estimating the capacity of lithium-ion batteries offline or online,it is essential to extract a health feature(HF)that can effectively characterize capacity degradation under both conventional ideal and complex d...When estimating the capacity of lithium-ion batteries offline or online,it is essential to extract a health feature(HF)that can effectively characterize capacity degradation under both conventional ideal and complex dynamic operating conditions.However,the extraction of most HFs relies on complete charge-discharge cycle data,making them less adaptable to complex dynamic operating conditions.Existing mechanism HFs,while capable of characterizing capacity degradation from a mechanism perspective,suffer from limitations such as insufficient physical model expressiveness,high dimension,and redundancy of the mechanism HF.These issues increase the complexity of subsequent modeling of the relationship between HFs and capacity,thereby restricting their promotion in engineering practice.To meet this gap,this paper proposes a novel mechanism-based HF.Firstly,a multi-physical fields coupling model is developed to describe the interactions between electrochemical,thermal,and aging behaviors of the battery.Secondly,based on the aging mechanism,the accumulated charge of lithium lost during the formation of the solid electrolyte interphase(SEI)film is extracted as HF to provide a more intuitive representation of capacity degradation.Then,to reduce estimation errors caused by considering only a single aging mechanism,multiple representative regression models are employed to establish the mapping relationship between the mechanism HF and capacity,further enhancing the accuracy of final results.Finally,the proposed method is implemented and validated using real battery data under three different types of operating conditions.Experimental results demonstrate that,compared to other commonly used HFs,the proposed HF exhibits significant competitive advantages in handling incomplete cycle data,unknown operating conditions,and capacity estimation models.The minimum estimation error under ideal conditions is 0.0074,and the minimum estimation error under complex dynamic conditions is 0.0268.展开更多
Non-learning based motion and path planning of an Unmanned Aerial Vehicle(UAV)is faced with low computation efficiency,mapping memory occupation and local optimization problems.This article investigates the challenge ...Non-learning based motion and path planning of an Unmanned Aerial Vehicle(UAV)is faced with low computation efficiency,mapping memory occupation and local optimization problems.This article investigates the challenge of quadrotor control using offline reinforcement learning.By establishing a data-driven learning paradigm that operates without real-environment interaction,the proposed workflow offers a safer approach than traditional reinforcement learning,making it particularly suited for UAV control in industrial scenarios.The introduced algorithm evaluates dataset uncertainty and employs a pessimistic estimation to foster offline deep reinforcement learning.Experiments highlight the algorithm's superiority over traditional online reinforcement learning methods,especially when learning from offline datasets.Furthermore,the article emphasizes the importance of a more general behavior policy.In evaluations,the trained policy demonstrated versatility by adeptly navigating diverse obstacles,underscoring its real-world applicability.展开更多
This study integrates multiple sources of data(transaction data,policy text,public opinion data)with visualization techniques(such as heat maps,time-series trend charts,3D building brochures)to construct an analysis f...This study integrates multiple sources of data(transaction data,policy text,public opinion data)with visualization techniques(such as heat maps,time-series trend charts,3D building brochures)to construct an analysis framework for the Chengdu real estate market.By using the Adaptive Neuro-Fuzzy Inference System(ANFIS)prediction model,spatial GIS(Geographic Information System analysis)analysis,and interactive dashboards,this study reveals market differentiation,policy impacts,and changes in demand structure,thereby providing decision support for the government,enterprises,and homebuyers.展开更多
To address the issue of instability or even imbalance in the orientation and attitude control of quadrotor unmanned aerial vehicles(QUAVs)under random disturbances,this paper proposes a distributed antidisturbance dat...To address the issue of instability or even imbalance in the orientation and attitude control of quadrotor unmanned aerial vehicles(QUAVs)under random disturbances,this paper proposes a distributed antidisturbance data-driven event-triggered fusion control method,which achieves efficient fault diagnosis while suppressing random disturbances and mitigating communication conflicts within the QUAV swarm.First,the impact of random disturbances on the UAV swarm is analyzed,and a model for orientation and attitude control of QUAVs under stochastic perturbations is established,with the disturbance gain threshold determined.Second,a fault diagnosis system based on a high-gain observer is designed,constructing a fault gain criterion by integrating orientation and attitude information from QUAVs.Subsequently,a model-free dynamic linearization-based data modeling(MFDLDM)framework is developed using model-free adaptive control,which efficiently fits the nonlinear control model of the QUAV swarm while reducing temporal constraints on control data.On this basis,this paper constructs a distributed data-driven event-triggered controller based on the staggered communication mechanism,which consists of an equivalent QUAV controller and an event-triggered controller,and is able to reduce the communication conflicts while suppressing the influence of random interference.Finally,by incorporating random disturbances into the controller,comparative experiments and physical validations are conducted on the QUAV platforms,fully demonstrating the strong adaptability and robustness of the proposed distributed event-triggered fault-tolerant control system.展开更多
Wetting deformation in earth-rockfill dams is a critical factor influencingdam safety.Although numerous mathematical models have been developed to describe this phenomenon,most of them rely on empirical formulations a...Wetting deformation in earth-rockfill dams is a critical factor influencingdam safety.Although numerous mathematical models have been developed to describe this phenomenon,most of them rely on empirical formulations and lack prior knowledge of model parameters,which is essential for Bayesian parameter inversion to enhance accuracy and reduce uncertainty.This study introduces a datadriven approach to establishing prior knowledge of earth-rockfill dams.Driving factors are utilized to determine the potential range of model parameters,and settlement changes within this range are calculated.The results are iteratively compared with actual monitoring data until the calculated range encompasses the observed data,thereby providing prior knowledge of the model parameters.The proposed method is applied to the right-bank earth-rockfilldam of Danjiangkou.Employing a Gibbs sample size of 30,000,the proposed method effectively calibrates the prior knowledge of the wetting model parameters,achieving a root mean square error(RMSE)of 5.18 mm for the settlement predictions.By comparison,the use of non-informative priors with sample sizes of 30,000 and 50,000 results in significantly larger RMSE values of 11.97 mm and 16.07 mm,respectively.Furthermore,the computational efficiencyof the proposed method is demonstrated by an inversion computation time of 902 s for 30,000 samples,which is notably shorter than the 1026 s and 1558 s required for noninformative priors with 30,000 and 50,000 samples,respectively.These findingsunderscore the superior performance of the proposed approach in terms of both prediction accuracy and computational efficiency.These results demonstrate that the proposed method not only improves the predictive accuracy but also enhances the computational efficiency,enabling optimal parameter identificationwith reduced computational effort.This approach provides a robust and efficientframework for advancing dam safety assessments.展开更多
Storm-enhanced density(SED)and the tongue of ionization(TOI)are key ionospheric storm-time structures whose rapid evolution and fine-scale variability remain challenging to capture with conventional empirical high-lat...Storm-enhanced density(SED)and the tongue of ionization(TOI)are key ionospheric storm-time structures whose rapid evolution and fine-scale variability remain challenging to capture with conventional empirical high-latitude drivers.In this study,we examine the May 10–11,2024,superstorm using the Thermosphere–Ionosphere–Electrodynamics General Circulation Model(TIEGCM)with observation-constrained high-latitude forcing.Auroral precipitation parameters(energy flux and mean energy)are assimilated from a Defense Meteorological Satellite Program(DMSP)Special Sensor Ultraviolet Spectrographic Imager(SSUSI)using a multi-resolution Gaussian process(Lattice Kriging)approach,whereas high-latitude convection potentials are derived by assimilating Super Dual Auroral Radar Network(SuperDARN)observations with the Thomas and Shepherd(2018)model(TS18).For comparison,an additional simulation is performed using empirical models for both convection and auroral forcing.The results show that during the main phase of the May 10 storm,the data-driven simulation provides a more realistic depiction of the SED source region than does the empirical model run by capturing its rapid intensification more clearly and reproducing its spatial location and structural features with higher fidelity.These improvements lead to a more accurate representation of its poleward extension into the polar cap that develops into the TOI.Above the ionospheric F2 peak over the SED source region,SuperDARN-constrained potentials generate stronger and more localized E×B drifts that dominate plasma uplift and drive its transport into the polar cap,although neutral winds and downward ambipolar diffusion partially offset these effects.Below the F2 peak,neutral winds and photochemical processes play a major role in shaping the spatial extent and intensity of the SED and TOI.These results highlight the role of observation-constrained high-latitude drivers in representing ionosphere–thermosphere responses during extreme storms and suggest their relevance for improving physical interpretation and model performance.展开更多
The Wufeng–Longmaxi Formation derives its name from the Upper Ordovician Wufeng Formation and the Lower Silurian Longmaxi Formation,found in sequence in the Sichuan Basin.This formation hosts rich shale gas reservoir...The Wufeng–Longmaxi Formation derives its name from the Upper Ordovician Wufeng Formation and the Lower Silurian Longmaxi Formation,found in sequence in the Sichuan Basin.This formation hosts rich shale gas reservoirs,and its shale gas enrichment patterns are examined in this study using data from 1197 shale samples collected from 14 wells.Five basic and three key parameters,eight in all,are assessed for each sample.The five basic parameters include burial depth and the contents of four mineral types—quartz,clay,carbonate,and other minerals;the three key parameters,representing shale gas enrichment,are total organic carbon(TOC)content,porosity,and gas content.The SHapley Additive exPlanations(SHAP)analysis originated in game theory is used here in an interpretable machine learning framework,to address issues of heterogeneous data structure,noisy relationships,and multi-objective optimization.An evaluation of the ranking,contribution values,and conditions of changes for these parameters offers new quantitative insights into shale gas enrichment patterns.A quantitative analysis of the relationship between data-sets identifies the primary factors controlling TOC,porosity,and gas content of shale gas reservoirs.The results show that TOC and porosity jointly influence gas content;mineral content has a significant impact on both,TOC and porosity;and the burial depth governs porosity which,in turn,affects the conditions under which shale gas is preserved.Input parameter thresholds are also determined and provide a basis for the establishment of quantitative criteria to evaluate shale gas enrichment.The predictive accuracy of the model used in this study is significantly improved by the step-wise addition of two input parameters,namely TOC and porosity,separately and together.Thus,the game theory method in big data-driven analysis uses a combination of TOC and porosity to evaluate the gas content with encouraging results—suggesting that these are the key parameters that indicate source rock and reservoir properties.展开更多
The key challenge in the preparation of perovskite solar cells is to enhance the reproducibility of PSC manufacturing,particularly by better controlling multiple high-dimensional process parameters.This study proposes...The key challenge in the preparation of perovskite solar cells is to enhance the reproducibility of PSC manufacturing,particularly by better controlling multiple high-dimensional process parameters.This study proposes a machine learning(ML)approach to efficiently predict and analyze perovskite film fabrication processes.By evaluating five classic ML algorithms on 130 experimental data sets from blade-coating parameters,the Random Forest(RF)model was identified as the most effective,enabling rapid prediction of over 100,000 parameter sets in just 10 min-equivalent to 3 years of manual experimentation.The RF model demonstrated strong predictive accuracy,with an R^(2) close to 0.8.This approach led to the identification of optimal process parameter combinations,significantly improving the reproducibility of PSCs and reducing performance variance by approximately threefold,thereby advancing the development of scalable manufacturing processes.展开更多
以高职航海技术专业为例,分析现代学徒制培养模式下的瓶颈问题,探究引入O2O(Online to Offline)教学方式,在线上线下形成学生、学校和企业共同参与的工学结合、产教融合的教学生态,提高航海技术专业学徒制学生的自主学习能力、专业技能...以高职航海技术专业为例,分析现代学徒制培养模式下的瓶颈问题,探究引入O2O(Online to Offline)教学方式,在线上线下形成学生、学校和企业共同参与的工学结合、产教融合的教学生态,提高航海技术专业学徒制学生的自主学习能力、专业技能和综合素质,培养满足社会和航运企业需求的航海人才。展开更多
Under the Kyoto Protocol,Japanwas supposed to reduce six percent of the green house gas (GHG) emission in 2012. However, until the year 2010, the statistics suggested that the GHG emission increased 4.2%. What is more...Under the Kyoto Protocol,Japanwas supposed to reduce six percent of the green house gas (GHG) emission in 2012. However, until the year 2010, the statistics suggested that the GHG emission increased 4.2%. What is more challenge is, afterFukushimacrisis, without the nuclear energy,Japanmay produce about 15 percent more GHG emissions than1990 inthis fiscal year. It still has to struggle to meet the target set by Kyoto Protocol. The demonstration area of “smart community” suggests Japanese exploration for new low carbon strategies. The study proposed a demand side response energy system, a dynamic tree-like hierarchical model for smart community. The model not only conveyed the concept of smart grid, but also built up a smart heat energy supply chain by offline heat transport system. Further, this model promoted a collaborative energy utilization mode between the industrial sector and the civil sector. In addition, the research chose the smart community inKitakyushuas case study and executed the model. The simulation and the analysis of the model not only evaluate the environmental effect of different technologies but also suggest that the smart community inJapanhas the potential but not easy to achieve the target, cut down 50% of the CO2 emission.展开更多
Offline Urdu Nastaleeq text recognition has long been a serious problem due to its very cursive nature. In order to get rid of the character segmentation problems, many researchers are shifting focus towards segmentat...Offline Urdu Nastaleeq text recognition has long been a serious problem due to its very cursive nature. In order to get rid of the character segmentation problems, many researchers are shifting focus towards segmentation free ligature based recognition approaches. Majority of the prevalent ligature based recognition systems heavily rely on hand-engineered feature extraction techniques. However, such techniques are more error prone and may often lead to a loss of useful information that might hardly be captured later by any manual features. Most of the prevalent Urdu Nastaleeq test recognition was trained and tested on small sets. This paper proposes the use of stacked denoising autoencoder for automatic feature extraction directly from raw pixel values of ligature images. Such deep learning networks have not been applied for the recognition of Urdu text thus far. Different stacked denoising autoencoders have been trained on 178573 ligatures with 3732 classes from un-degraded(noise free) UPTI(Urdu Printed Text Image) data set. Subsequently, trained networks are validated and tested on degraded versions of UPTI data set. The experimental results demonstrate accuracies in range of 93% to 96% which are better than the existing Urdu OCR systems for such large dataset of ligatures.展开更多
Inherent complexity of plant metabolites necessitates the use of multi-dimensional information to accomplish comprehensive profiling and confirmative identification.A dimension-enhanced strategy,by offline two-dimensi...Inherent complexity of plant metabolites necessitates the use of multi-dimensional information to accomplish comprehensive profiling and confirmative identification.A dimension-enhanced strategy,by offline two-dimensional liquid chromatography/ion mobility-quadrupole time-of-flight mass spectrometry(2 D-LC/IM-QTOF-MS)enabling four-dimensional separations(2 D-LC,IM,and MS),is proposed.In combination with in-house database-driven automated peak annotation,this strategy was utilized to characterize ginsenosides simultaneously from white ginseng(WG)and red ginseng(RG).An offline 2 DLC system configuring an Xbridge Amide column and an HSS T3 column showed orthogonality 0.76 in the resolution of ginsenosides.Ginsenoside analysis was performed by data-independent high-definition MSE(HDMSE)in the negative ESI mode on a Vion?IMS-QTOF hybrid high-resolution mass spectrometer,which could better resolve ginsenosides than MSEand directly give the CCS information.An in-house ginsenoside database recording 504 known ginsenosides and 58 reference compounds,was established to assist the identification of ginsenosides.Streamlined workflows,by applying UNIFI?to automatedly annotate the HDMSEdata,were proposed.We could separate and characterize 323 ginsenosides(including 286 from WG and 306 from RG),and 125 thereof may have not been isolated from the Panax genus.The established 2 D-LC/IM-QTOF-HDMSEapproach could also act as a magnifier to probe differentiated components between WG and RG.Compared with conventional approaches,this dimensionenhanced strategy could better resolve coeluting herbal components and more efficiently,more reliably identify the multicomponents,which,we believe,offers more possibilities for the systematic exposure and confirmative identification of plant metabolites.展开更多
An improved generalized predictive control algorithm is presented in thispaper by incorporating offline identification into online identification. Unlike the existinggeneralized predictive control algorithms, the prop...An improved generalized predictive control algorithm is presented in thispaper by incorporating offline identification into online identification. Unlike the existinggeneralized predictive control algorithms, the proposed approach divides parameters of a predictivemodel into the time invariant and time-varying ones, which are treated respectively by offline andonline identification algorithms. Therefore, both the reliability and accuracy of the predictivemodel are improved. Two simulation examples of control of a fixed bed reactor show that this newalgorithm is not only reliable and stable in the case of uncertainties and abnormal disturbances,but also adaptable to slow time varying processes.展开更多
基金supported by the National Natural Science Foundation of China(62373364,62176259)the Key Research and Development Program of Jiangsu Province(BE2022095)。
文摘In order to address the issue of overly conservative offline reinforcement learning(RL) methods that limit the generalization of policy in the out-of-distribution(OOD) region,this article designs a surrogate target for OOD value function based on dataset distance and proposes a novel generalized Q-learning mechanism with distance regularization(GQDR).In theory,we not only prove the convergence of GQDR,but also ensure that the difference between the Q-value learned by GQDR and its true value is bounded.Furthermore,an offline generalized actor-critic method with distance regularization(OGACDR) is proposed by combining GQDR with actor-critic learning framework.Two implementations of OGACDR,OGACDR-EXP and OGACDRSQR,are introduced according to exponential(EXP) and opensquare(SQR) distance weight functions,and it has been theoretically proved that OGACDR provides a safe policy improvement.Experimental results on Gym-MuJoCo continuous control tasks show that OGACDR can not only alleviate the overestimation and overconservatism of Q-value function,but also outperform conservative offline RL baselines.
文摘Offline policy evaluation,evaluating and selecting complex policies for decision-making by only using offline datasets is important in reinforcement learning.At present,the model-based offline policy evaluation(MBOPE)is widely welcomed because of its easy to implement and good performance.MBOPE directly approximates the unknown value of a given policy using the Monte Carlo method given the estimated transition and reward functions of the environment.Usually,multiple models are trained,and then one of them is selected to be used.However,a challenge remains in selecting an appropriate model from those trained for further use.The authors first analyse the upper bound of the difference between the approximated value and the unknown true value.Theoretical results show that this difference is related to the trajectories generated by the given policy on the learnt model and the prediction error of the transition and reward functions at these generated data points.Based on the theoretical results,a new criterion is proposed to tell which trained model is better suited for evaluating the given policy.At last,the effectiveness of the proposed criterion is demonstrated on both benchmark and synthetic offline datasets.
基金supported by the National Natural Science Foundation of China(No.12372045)the National Key Research and the Development Program of China(Nos.2023YFC2205900,2023YFC2205901)。
文摘This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mission is disturbed by the measurement noise and the target adversarial behavior.To address these problems,a model-free Combined Adaptive-length Datadriven Predictive Controller(CADPC)is proposed.It consists of a separated subsystem identification method and a combined predictive control strategy.The subsystem identification method is composed of an adaptive data length,thereby reducing sensitivity to undetermined measurement noises and disturbances.Based on the subsystem identification,the combined predictive controller is established,reducing calculating resource.The stability of the CADPC is rigorously proven using the Input-to-State Stable(ISS)theorem and the small-gain theorem.Simulations demonstrate that CADPC effectively handles the model-free space robot post operation in the presence of significant disturbances,state measurement noise,and control input errors.It achieves improved steady-state accuracy,reduced steady-state control consumption,and minimized control input chattering.
基金supported by the National Natural Science Foundation of China(NSFC,No.62303031)the Fundamental Research Funds for the Central Universities。
文摘When estimating the capacity of lithium-ion batteries offline or online,it is essential to extract a health feature(HF)that can effectively characterize capacity degradation under both conventional ideal and complex dynamic operating conditions.However,the extraction of most HFs relies on complete charge-discharge cycle data,making them less adaptable to complex dynamic operating conditions.Existing mechanism HFs,while capable of characterizing capacity degradation from a mechanism perspective,suffer from limitations such as insufficient physical model expressiveness,high dimension,and redundancy of the mechanism HF.These issues increase the complexity of subsequent modeling of the relationship between HFs and capacity,thereby restricting their promotion in engineering practice.To meet this gap,this paper proposes a novel mechanism-based HF.Firstly,a multi-physical fields coupling model is developed to describe the interactions between electrochemical,thermal,and aging behaviors of the battery.Secondly,based on the aging mechanism,the accumulated charge of lithium lost during the formation of the solid electrolyte interphase(SEI)film is extracted as HF to provide a more intuitive representation of capacity degradation.Then,to reduce estimation errors caused by considering only a single aging mechanism,multiple representative regression models are employed to establish the mapping relationship between the mechanism HF and capacity,further enhancing the accuracy of final results.Finally,the proposed method is implemented and validated using real battery data under three different types of operating conditions.Experimental results demonstrate that,compared to other commonly used HFs,the proposed HF exhibits significant competitive advantages in handling incomplete cycle data,unknown operating conditions,and capacity estimation models.The minimum estimation error under ideal conditions is 0.0074,and the minimum estimation error under complex dynamic conditions is 0.0268.
基金supported by the National Natural Science Foundation of China(No.52272382)the Aeronautical Science Foundation of China(No.20200017051001)the Fundamental Research Funds for the Central Universities,China。
文摘Non-learning based motion and path planning of an Unmanned Aerial Vehicle(UAV)is faced with low computation efficiency,mapping memory occupation and local optimization problems.This article investigates the challenge of quadrotor control using offline reinforcement learning.By establishing a data-driven learning paradigm that operates without real-environment interaction,the proposed workflow offers a safer approach than traditional reinforcement learning,making it particularly suited for UAV control in industrial scenarios.The introduced algorithm evaluates dataset uncertainty and employs a pessimistic estimation to foster offline deep reinforcement learning.Experiments highlight the algorithm's superiority over traditional online reinforcement learning methods,especially when learning from offline datasets.Furthermore,the article emphasizes the importance of a more general behavior policy.In evaluations,the trained policy demonstrated versatility by adeptly navigating diverse obstacles,underscoring its real-world applicability.
基金Chengdu City Philosophy and Social Sciences Research Center“artificial intelligence+urban communication”theory and Application Research Center Project“Chengdu real estate vertical market public opinion data visualization research”(Project No.RZCC2025017).
文摘This study integrates multiple sources of data(transaction data,policy text,public opinion data)with visualization techniques(such as heat maps,time-series trend charts,3D building brochures)to construct an analysis framework for the Chengdu real estate market.By using the Adaptive Neuro-Fuzzy Inference System(ANFIS)prediction model,spatial GIS(Geographic Information System analysis)analysis,and interactive dashboards,this study reveals market differentiation,policy impacts,and changes in demand structure,thereby providing decision support for the government,enterprises,and homebuyers.
基金supported in part by the National Natural Science Foundation of China,Grant/Award Number:62003267the Key Research and Development Program of Shaanxi Province,Grant/Award Number:2023-GHZD-33Open Project of the State Key Laboratory of Intelligent Game,Grant/Award Number:ZBKF-23-05。
文摘To address the issue of instability or even imbalance in the orientation and attitude control of quadrotor unmanned aerial vehicles(QUAVs)under random disturbances,this paper proposes a distributed antidisturbance data-driven event-triggered fusion control method,which achieves efficient fault diagnosis while suppressing random disturbances and mitigating communication conflicts within the QUAV swarm.First,the impact of random disturbances on the UAV swarm is analyzed,and a model for orientation and attitude control of QUAVs under stochastic perturbations is established,with the disturbance gain threshold determined.Second,a fault diagnosis system based on a high-gain observer is designed,constructing a fault gain criterion by integrating orientation and attitude information from QUAVs.Subsequently,a model-free dynamic linearization-based data modeling(MFDLDM)framework is developed using model-free adaptive control,which efficiently fits the nonlinear control model of the QUAV swarm while reducing temporal constraints on control data.On this basis,this paper constructs a distributed data-driven event-triggered controller based on the staggered communication mechanism,which consists of an equivalent QUAV controller and an event-triggered controller,and is able to reduce the communication conflicts while suppressing the influence of random interference.Finally,by incorporating random disturbances into the controller,comparative experiments and physical validations are conducted on the QUAV platforms,fully demonstrating the strong adaptability and robustness of the proposed distributed event-triggered fault-tolerant control system.
基金supported by the National Key R&D Program of China(Grant No.2023YFC3209504)Natural Science Foundation of Wuhan(Grant No.2024040801020271)the Fundamental Research Funds for Central Public Welfare Research Institutes(Grant No.CKSF2025718/YT).
文摘Wetting deformation in earth-rockfill dams is a critical factor influencingdam safety.Although numerous mathematical models have been developed to describe this phenomenon,most of them rely on empirical formulations and lack prior knowledge of model parameters,which is essential for Bayesian parameter inversion to enhance accuracy and reduce uncertainty.This study introduces a datadriven approach to establishing prior knowledge of earth-rockfill dams.Driving factors are utilized to determine the potential range of model parameters,and settlement changes within this range are calculated.The results are iteratively compared with actual monitoring data until the calculated range encompasses the observed data,thereby providing prior knowledge of the model parameters.The proposed method is applied to the right-bank earth-rockfilldam of Danjiangkou.Employing a Gibbs sample size of 30,000,the proposed method effectively calibrates the prior knowledge of the wetting model parameters,achieving a root mean square error(RMSE)of 5.18 mm for the settlement predictions.By comparison,the use of non-informative priors with sample sizes of 30,000 and 50,000 results in significantly larger RMSE values of 11.97 mm and 16.07 mm,respectively.Furthermore,the computational efficiencyof the proposed method is demonstrated by an inversion computation time of 902 s for 30,000 samples,which is notably shorter than the 1026 s and 1558 s required for noninformative priors with 30,000 and 50,000 samples,respectively.These findingsunderscore the superior performance of the proposed approach in terms of both prediction accuracy and computational efficiency.These results demonstrate that the proposed method not only improves the predictive accuracy but also enhances the computational efficiency,enabling optimal parameter identificationwith reduced computational effort.This approach provides a robust and efficientframework for advancing dam safety assessments.
基金The Shandong Provincial Natural Science Foundation(Grant No.ZR2022JQ18)supported this worksupported by the National Natural Science Foundation of China(NNFSC)Youth Program(Grant No.42304168)+1 种基金supported by the National Key R&D Program of China(Grant No.2022YFF0504400)the NNSFC(Grant Nos.42188101 and 42174210)。
文摘Storm-enhanced density(SED)and the tongue of ionization(TOI)are key ionospheric storm-time structures whose rapid evolution and fine-scale variability remain challenging to capture with conventional empirical high-latitude drivers.In this study,we examine the May 10–11,2024,superstorm using the Thermosphere–Ionosphere–Electrodynamics General Circulation Model(TIEGCM)with observation-constrained high-latitude forcing.Auroral precipitation parameters(energy flux and mean energy)are assimilated from a Defense Meteorological Satellite Program(DMSP)Special Sensor Ultraviolet Spectrographic Imager(SSUSI)using a multi-resolution Gaussian process(Lattice Kriging)approach,whereas high-latitude convection potentials are derived by assimilating Super Dual Auroral Radar Network(SuperDARN)observations with the Thomas and Shepherd(2018)model(TS18).For comparison,an additional simulation is performed using empirical models for both convection and auroral forcing.The results show that during the main phase of the May 10 storm,the data-driven simulation provides a more realistic depiction of the SED source region than does the empirical model run by capturing its rapid intensification more clearly and reproducing its spatial location and structural features with higher fidelity.These improvements lead to a more accurate representation of its poleward extension into the polar cap that develops into the TOI.Above the ionospheric F2 peak over the SED source region,SuperDARN-constrained potentials generate stronger and more localized E×B drifts that dominate plasma uplift and drive its transport into the polar cap,although neutral winds and downward ambipolar diffusion partially offset these effects.Below the F2 peak,neutral winds and photochemical processes play a major role in shaping the spatial extent and intensity of the SED and TOI.These results highlight the role of observation-constrained high-latitude drivers in representing ionosphere–thermosphere responses during extreme storms and suggest their relevance for improving physical interpretation and model performance.
基金funded by the Technical Development(Entrusted)Project of Science and Department of SINOPEC(Grant No.P23240-4)the National Natural Science Foundation of China(Grant Nos.42172165,42272143 and 2025ZD1403901-05)。
文摘The Wufeng–Longmaxi Formation derives its name from the Upper Ordovician Wufeng Formation and the Lower Silurian Longmaxi Formation,found in sequence in the Sichuan Basin.This formation hosts rich shale gas reservoirs,and its shale gas enrichment patterns are examined in this study using data from 1197 shale samples collected from 14 wells.Five basic and three key parameters,eight in all,are assessed for each sample.The five basic parameters include burial depth and the contents of four mineral types—quartz,clay,carbonate,and other minerals;the three key parameters,representing shale gas enrichment,are total organic carbon(TOC)content,porosity,and gas content.The SHapley Additive exPlanations(SHAP)analysis originated in game theory is used here in an interpretable machine learning framework,to address issues of heterogeneous data structure,noisy relationships,and multi-objective optimization.An evaluation of the ranking,contribution values,and conditions of changes for these parameters offers new quantitative insights into shale gas enrichment patterns.A quantitative analysis of the relationship between data-sets identifies the primary factors controlling TOC,porosity,and gas content of shale gas reservoirs.The results show that TOC and porosity jointly influence gas content;mineral content has a significant impact on both,TOC and porosity;and the burial depth governs porosity which,in turn,affects the conditions under which shale gas is preserved.Input parameter thresholds are also determined and provide a basis for the establishment of quantitative criteria to evaluate shale gas enrichment.The predictive accuracy of the model used in this study is significantly improved by the step-wise addition of two input parameters,namely TOC and porosity,separately and together.Thus,the game theory method in big data-driven analysis uses a combination of TOC and porosity to evaluate the gas content with encouraging results—suggesting that these are the key parameters that indicate source rock and reservoir properties.
基金Key Research and Development Program of Hubei Province,China(Grant No.2022BAA096)Zhejiang Provincial Natural Science Foundation of China(This material is based upon work funded by Zhejiang Provincial Natural Science Foundation of China under Grant No.LR25A020002)support of the Center for Materials Analysis and Characterization,Material Characterization Lab,and Nanofabrication Lab at Hubei University。
文摘The key challenge in the preparation of perovskite solar cells is to enhance the reproducibility of PSC manufacturing,particularly by better controlling multiple high-dimensional process parameters.This study proposes a machine learning(ML)approach to efficiently predict and analyze perovskite film fabrication processes.By evaluating five classic ML algorithms on 130 experimental data sets from blade-coating parameters,the Random Forest(RF)model was identified as the most effective,enabling rapid prediction of over 100,000 parameter sets in just 10 min-equivalent to 3 years of manual experimentation.The RF model demonstrated strong predictive accuracy,with an R^(2) close to 0.8.This approach led to the identification of optimal process parameter combinations,significantly improving the reproducibility of PSCs and reducing performance variance by approximately threefold,thereby advancing the development of scalable manufacturing processes.
文摘Under the Kyoto Protocol,Japanwas supposed to reduce six percent of the green house gas (GHG) emission in 2012. However, until the year 2010, the statistics suggested that the GHG emission increased 4.2%. What is more challenge is, afterFukushimacrisis, without the nuclear energy,Japanmay produce about 15 percent more GHG emissions than1990 inthis fiscal year. It still has to struggle to meet the target set by Kyoto Protocol. The demonstration area of “smart community” suggests Japanese exploration for new low carbon strategies. The study proposed a demand side response energy system, a dynamic tree-like hierarchical model for smart community. The model not only conveyed the concept of smart grid, but also built up a smart heat energy supply chain by offline heat transport system. Further, this model promoted a collaborative energy utilization mode between the industrial sector and the civil sector. In addition, the research chose the smart community inKitakyushuas case study and executed the model. The simulation and the analysis of the model not only evaluate the environmental effect of different technologies but also suggest that the smart community inJapanhas the potential but not easy to achieve the target, cut down 50% of the CO2 emission.
基金National Natural Science Foundation of China (Project No. 61273365)111 Project (No. B08004) are gratefully acknowledged
文摘Offline Urdu Nastaleeq text recognition has long been a serious problem due to its very cursive nature. In order to get rid of the character segmentation problems, many researchers are shifting focus towards segmentation free ligature based recognition approaches. Majority of the prevalent ligature based recognition systems heavily rely on hand-engineered feature extraction techniques. However, such techniques are more error prone and may often lead to a loss of useful information that might hardly be captured later by any manual features. Most of the prevalent Urdu Nastaleeq test recognition was trained and tested on small sets. This paper proposes the use of stacked denoising autoencoder for automatic feature extraction directly from raw pixel values of ligature images. Such deep learning networks have not been applied for the recognition of Urdu text thus far. Different stacked denoising autoencoders have been trained on 178573 ligatures with 3732 classes from un-degraded(noise free) UPTI(Urdu Printed Text Image) data set. Subsequently, trained networks are validated and tested on degraded versions of UPTI data set. The experimental results demonstrate accuracies in range of 93% to 96% which are better than the existing Urdu OCR systems for such large dataset of ligatures.
基金the National Natural Science Foundation of China(Grant No.81872996)the State Key Research and Development Project(Grant No.2017YFC1702104)+1 种基金the State Key Project for the Creation of Major New Drugs(2018ZX09711001-009-010)the Tianjin Municipal Education Commission Research Project(Grant No.2017ZD07)。
文摘Inherent complexity of plant metabolites necessitates the use of multi-dimensional information to accomplish comprehensive profiling and confirmative identification.A dimension-enhanced strategy,by offline two-dimensional liquid chromatography/ion mobility-quadrupole time-of-flight mass spectrometry(2 D-LC/IM-QTOF-MS)enabling four-dimensional separations(2 D-LC,IM,and MS),is proposed.In combination with in-house database-driven automated peak annotation,this strategy was utilized to characterize ginsenosides simultaneously from white ginseng(WG)and red ginseng(RG).An offline 2 DLC system configuring an Xbridge Amide column and an HSS T3 column showed orthogonality 0.76 in the resolution of ginsenosides.Ginsenoside analysis was performed by data-independent high-definition MSE(HDMSE)in the negative ESI mode on a Vion?IMS-QTOF hybrid high-resolution mass spectrometer,which could better resolve ginsenosides than MSEand directly give the CCS information.An in-house ginsenoside database recording 504 known ginsenosides and 58 reference compounds,was established to assist the identification of ginsenosides.Streamlined workflows,by applying UNIFI?to automatedly annotate the HDMSEdata,were proposed.We could separate and characterize 323 ginsenosides(including 286 from WG and 306 from RG),and 125 thereof may have not been isolated from the Panax genus.The established 2 D-LC/IM-QTOF-HDMSEapproach could also act as a magnifier to probe differentiated components between WG and RG.Compared with conventional approaches,this dimensionenhanced strategy could better resolve coeluting herbal components and more efficiently,more reliably identify the multicomponents,which,we believe,offers more possibilities for the systematic exposure and confirmative identification of plant metabolites.
基金Supported by the National Natural Science Foundation of China (No. 20206028) and the Qingdao Municipal Major Lab of Industry Information Technology.
文摘An improved generalized predictive control algorithm is presented in thispaper by incorporating offline identification into online identification. Unlike the existinggeneralized predictive control algorithms, the proposed approach divides parameters of a predictivemodel into the time invariant and time-varying ones, which are treated respectively by offline andonline identification algorithms. Therefore, both the reliability and accuracy of the predictivemodel are improved. Two simulation examples of control of a fixed bed reactor show that this newalgorithm is not only reliable and stable in the case of uncertainties and abnormal disturbances,but also adaptable to slow time varying processes.
