Deformations in high fill foundations comprising soil–stone mixtures must be accurately predicted to ensure construction quality and long-term operational safety.However,existing computational and analytical methods ...Deformations in high fill foundations comprising soil–stone mixtures must be accurately predicted to ensure construction quality and long-term operational safety.However,existing computational and analytical methods inadequately capture their complex mechanical behavior.We conducted a series of triaxial tests on unsaturated soil samples collected from a high fill project site in northwestern China under three stress paths.The incremental nonlinear and elastoplastic constitutive models for unsaturated soils were modified,and a calculation method was developed for the vertical and lateral deformations of high fill foundations using the layered summation approach.The results indicate that for soil samples with the same mixing ratio(m)and compaction coefficient(n),the strength of the sample and its tendency to exhibit shear dilation increase with the net confining pressure or matric suction.Additionally,the stress–strain curve of the soil sample gradually changes from the strain-hardening type to the ideal elastoplastic type as the compaction coefficient increases.Moreover,the compaction coefficient is an important factor influencing the magnitude of yield stress and yield suction in soil samples,and the yield points of both are similar in shape to the loadingcollapse(LC)and suction increase(SI)yield lines obtained using the Barcelona model in the net mean stress-generalized shear stress(p-s)plane,respectively.The modified incremental nonlinear instantaneous model simultaneously considers the effects of the compaction coefficient,suction and mixing ratio,and the model parameter can be simplified to the tangential modulus expression in the Duncan-Chang model when the suction is zero.Furthermore,the modified elastoplastic constitutive model,which considers the effects of the net mean stress,suction and partial stress,can be simplified to the elastoplastic constitutive relationship of saturated soil when the suction is zero.The proposed deformation calculation method,based on the layered summation theory,is applicable to both elastic and elastoplastic foundation states,as confirmed through numerical simulations.Our work can be used as a reference for the calculation of foundation deformation in similar mixed material high fill projects.展开更多
Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Re...Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.展开更多
With the emergence of general foundational models,such as Chat Generative Pre-trained Transformer(ChatGPT),researchers have shown considerable interest in the potential applications of foundation models in the process...With the emergence of general foundational models,such as Chat Generative Pre-trained Transformer(ChatGPT),researchers have shown considerable interest in the potential applications of foundation models in the process industry.This paper provides a comprehensive overview of the challenges and opportunities presented by the use of foundation models in the process industry,including the frameworks,core applications,and future prospects.First,this paper proposes a framework for foundation models for the process industry.Second,it summarizes the key capabilities of industrial foundation models and their practical applications.Finally,it highlights future research directions and identifies unresolved open issues related to the use of foundation models in the process industry.展开更多
With the continuous development of the offshore wind industry,the design concept of composite foundation has been given attention in the past decade.This paper presents an accurate method for investigating the horizon...With the continuous development of the offshore wind industry,the design concept of composite foundation has been given attention in the past decade.This paper presents an accurate method for investigating the horizontal vibration of monopile-friction wheel composite foundations in layered saturated soil.Firstly,the three-dimensional continuum mechanics theory with the range of linear elasticity is introduced to calculate the frictional resistance distributed on the upper soil surface.Then,the resistances of multilayered soils and inviscid seawater to the pile shaft under horizontal harmonic excitation are obtained using Novak's plane strain model,Biot's porous media theory and radiationwave theory.Thirdly,the expressions for the deformation,bending moment and internal force of the Euler-Bernoulli pile are derived using the boundary conditions with definitephysical meaning and transfer matrix method.By comparing with the results of 1g laboratory test and the idealized formula reported by the literature,the rationality and accuracy of the developed dynamical model can be verified.Finally,this paper conducts a series of worked examples to investigate the influencesof the elastic modulus and thickness of three-layer saturated soil and the location of interlayer soil on the horizontal dynamic vibration of composite foundation.The results show that an increase in elastic modulus of the surface soil is an effective way to improve the dynamic stability of the composite foundation in service conditions.The conclusions drawn from the numerical examples can develop some guidelines for the current foundation design of offshore wind turbines.展开更多
Gastrointestinal(GI)cancers represent a major global health concern due to their high incidence and mortality rates.Foundation models(FMs),also referred to as large models,represent a novel class of artificial intelli...Gastrointestinal(GI)cancers represent a major global health concern due to their high incidence and mortality rates.Foundation models(FMs),also referred to as large models,represent a novel class of artificial intelligence technologies that have demonstrated considerable potential in addressing these challenges.These models encompass large language models(LLMs),vision FMs(VFMs),and multimodal LLMs(MLLMs),all of which utilize transformer architectures and self-supervised pre-training on extensive unlabeled datasets to achieve robust cross-domain generalization.This review delineates the principal applications of these models:LLMs facilitate the structuring of clinical narratives,extraction of insights from medical records,and enhancement of physician-patient communication;VFMs are employed in the analysis of endoscopic,radiological,and pathological images for lesion detection and staging;MLLMs integrate heterogeneous data modalities,including imaging,textual information,and genomic data,to support diagnostic processes,treatment prediction,and prognostic evaluation.Despite these promising developments,several challenges remain,such as the need for data standardization,limited diversity within training datasets,substantial computational resource requirements,and ethical-legal concerns.In conclusion,FMs exhibit significant potential to advance research and clinical management of GI cancers.Future research efforts should prioritize the refinement of these models,promote international collaborations,and adopt interdisciplinary approaches.Such a comprehensive strategy is essential to fully harness the capabilities of FMs,driving substantial progress in the fight against GI malignancies.展开更多
The existing analytical models for umbrella arch method(UAM)based on elastic foundation beams often overlook the influence of the surrounding soil beyond the beam edges on the shear stresses acting on the beam.Consequ...The existing analytical models for umbrella arch method(UAM)based on elastic foundation beams often overlook the influence of the surrounding soil beyond the beam edges on the shear stresses acting on the beam.Consequently,such models fail to adequately reflect the continuity characteristics of soil deformation.Leveraging the Pasternak foundation-Euler beam model,this study considers the generalized shear force on the beam to account for the influence of soil outside the beam ends on the shear stress.An analytical model for the deformation and internal forces of finite-length beams subjected to arbitrary loads is derived based on the initial parameter method under various conditions.The mechanical model of the elastic foundation beam for advanced umbrella arch under typical tunnel excavation cycles is established,yielding analytical solutions for the longitudinal response of the umbrella arch.The reliability of the analytical model is verified with the existing test data.The improved model addresses anomalies in existing models,such as abnormal upward deformation in the loosened segment and maximum deflection occurring within the soil mass.Additionally,dimensionless characteristic parameters reflecting the relative stiffness between the umbrella arch structure and the foundation soil are proposed.Results indicate that the magnitude of soil characteristic parameters significantly influences the deformation and internal forces of the umbrella arch.Within common ranges of soil values,the maximum deformation and internal forces of the umbrella arch under semi-logarithmic coordinates exhibit nearly linear decay with decreasing soil characteristic parameters.The impact of tunnel excavation height on the stress of unsupported sections of the umbrella arch is minor,but it is more significant for umbrella arch buried within the soil mass.Conversely,the influence of tunnel excavation advance on the umbrella arch is opposite.展开更多
Foundation models(FMs)have rapidly evolved and have achieved signicant accomplishments in computer vision tasks.Specically,the prompt mechanism conveniently allows users to integrate image prior information into the m...Foundation models(FMs)have rapidly evolved and have achieved signicant accomplishments in computer vision tasks.Specically,the prompt mechanism conveniently allows users to integrate image prior information into the model,making it possible to apply models without any training.Therefore,we proposed a workflow based on foundation models and zero training to solve the tasks of photoacoustic(PA)image processing.We employed the Segment Anything Model(SAM)by setting simple prompts and integrating the model's outputs with prior knowledge of the imaged objects to accomplish various tasks,including:(1)removing the skin signal in three-dimensional PA image rendering;(2)dual speed-of-sound reconstruction,and(3)segmentation ofnger blood vessels.Through these demonstrations,we have concluded that FMs can be directly applied in PA imaging without the requirement for network design and training.This potentially allows for a hands-on,convenient approach to achieving efficient and accurate segmentation of PA images.This paper serves as a comprehensive tutorial,facilitating the mastery of the technique through the provision of code and sample datasets.展开更多
Predictive maintenance often involves imbalanced multivariate time series datasets with scarce failure events,posing challenges for model training due to the high dimensionality of the data and the need for domain-spe...Predictive maintenance often involves imbalanced multivariate time series datasets with scarce failure events,posing challenges for model training due to the high dimensionality of the data and the need for domain-specific preprocessing,which frequently leads to the development of large and complex models.Inspired by the success of Large Language Models(LLMs),transformer-based foundation models have been developed for time series(TSFM).These models have been proven to reconstruct time series in a zero-shot manner,being able to capture different patterns that effectively characterize time series.This paper proposes the use of TSFM to generate embeddings of the input data space,making them more interpretable for machine learning models.To evaluate the effectiveness of our approach,we trained three classical machine learning algorithms and one neural network using the embeddings generated by the TSFM called Moment for predicting the remaining useful life of aircraft engines.We test the models trained with both the full training dataset and only 10%of the training samples.Our results show that training simple models,such as support vector regressors or neural networks,with embeddings generated by Moment not only accelerates the training process but also enhances performance in few-shot learning scenarios,where data is scarce.This suggests a promising alternative to complex deep learning architectures,particularly in industrial contexts with limited labeled data.展开更多
To thoroughly examine the complex relationships between tire and pavement vibrations,a sophisticated vehicle-pavement coupled system is proposed,incorporating a non-uniform dynamic friction force between the tire and ...To thoroughly examine the complex relationships between tire and pavement vibrations,a sophisticated vehicle-pavement coupled system is proposed,incorporating a non-uniform dynamic friction force between the tire and the pavement.According to the Timoshenko beam theory,a dynamic model of pavement structure with a finite length beam was formulated on a nonlinear Pasternak foundation.To more accurately describe the coupling relationship between the tire and the pavement,and to take into account the vibration state under vehicle-pavement interaction,the load distribution between the tire and the pavement is modeled as a dynamic non-uniform contact.Combined with the classic LuGre tire model,the adhesion between the tire and the pavement is calculated.The Galerkin truncation method is employed to transform the pavement vibration partial differential equation into a finite ordinary differential equation,and the integral expression of the nonlinear foundation beam term is derived using the product to sum formula.By using the Runge-Kutta method,the tire-road coupled system can be numerically calculated,thus determining tire adhesion.This research demonstrates that compared with tire force under the traditional static load distribution,load distribution has a significant influence on adhesion.This study offers valuable insights for pavement structure design and vehicle performance control.展开更多
The Euler-Bernoulli(E-B)beam theory is combined with Green-Lindsay's(G-L)generalized thermoelasticity theory to analyze the vibration of microbeams.The frequency control equation,based on the two-parameter Winkler...The Euler-Bernoulli(E-B)beam theory is combined with Green-Lindsay's(G-L)generalized thermoelasticity theory to analyze the vibration of microbeams.The frequency control equation,based on the two-parameter Winkler-Pasternak elastic foundation for simply-supported microbeams,is presented.This study investigates the effects of the side-to-thickness ratio and relaxation time parameters on the vibrational natural frequency of thermoelastic microbeam resonators.The frequencies derived from the present model are compared with those from Lord and Shulman's(L-S)theory.The fourthorder solutions for natural vibration frequencies are graphically displayed for comparison.Therefore,attention should be paid to the use of effective foundations to prevent microbeam damage caused by contraction and expansion problems caused by high temperatures.展开更多
Asia is the largest and most populous continent in the world;its security and stability play a crucial role in global peace and development.As the world is undergoing unprecedented transformations and entering a new p...Asia is the largest and most populous continent in the world;its security and stability play a crucial role in global peace and development.As the world is undergoing unprecedented transformations and entering a new period of turbulence and changes,Asia’s ascension is tempered by complex security challenges,both traditional and non-traditional.Recent geopolitical upheavals,from the Ukraine crisis to the Middle East turmoil,have spurred Asian countries to reassess international security dilemmas,exploring how to draw insights from the essence of Asian civilizational values to contribute positively to global security and stability.展开更多
Since the 1950s,when the Turing Test was introduced,there has been notable progress in machine language intelligence.Language modeling,crucial for AI development,has evolved from statistical to neural models over the ...Since the 1950s,when the Turing Test was introduced,there has been notable progress in machine language intelligence.Language modeling,crucial for AI development,has evolved from statistical to neural models over the last two decades.Recently,transformer-based Pre-trained Language Models(PLM)have excelled in Natural Language Processing(NLP)tasks by leveraging large-scale training corpora.Increasing the scale of these models enhances performance significantly,introducing abilities like context learning that smaller models lack.The advancement in Large Language Models,exemplified by the development of ChatGPT,has made significant impacts both academically and industrially,capturing widespread societal interest.This survey provides an overview of the development and prospects from Large Language Models(LLM)to Large Multimodal Models(LMM).It first discusses the contributions and technological advancements of LLMs in the field of natural language processing,especially in text generation and language understanding.Then,it turns to the discussion of LMMs,which integrates various data modalities such as text,images,and sound,demonstrating advanced capabilities in understanding and generating cross-modal content,paving new pathways for the adaptability and flexibility of AI systems.Finally,the survey highlights the prospects of LMMs in terms of technological development and application potential,while also pointing out challenges in data integration,cross-modal understanding accuracy,providing a comprehensive perspective on the latest developments in this field.展开更多
To reveal the bearing capacity of the X-section concrete piles pile raft foundation in silica sand,a series of vertical load tests are carried out.The X-section concrete piles are compared with circular section pile w...To reveal the bearing capacity of the X-section concrete piles pile raft foundation in silica sand,a series of vertical load tests are carried out.The X-section concrete piles are compared with circular section pile with the same section area.The load−settlement curves,axial force and skin friction,strain on concave and convex edge of the pile,pile-sand stress ratio,distributions of side and tip resistance are presented.The results show that bearing capacity of the X section concrete pile raft foundation is much larger than that of the circular pile raft foundation.Besides,compared with the circular pile,the peak axial force of X-section piles under raft is deeper and smaller while the neutral point of X-section concrete pile is deeper.Moreover,the strain on the concave edge is much larger than that on the convex edge of the pile,and the convex edge has more potential in bearing capacity as the vertical load increases.The X-section pile has higher pile-sand stress ratios and load-sharing between side resistance and tip resistance.Above all,the X-section concrete pile can significantly increase the bearing capacity of pile-raft foundations in silica sand.展开更多
To obtain the vertical earth pressure on a soft foundation box culvert and investigate the interaction of the soil-culvert-foundation system, both a centrifugal model test and a numerical simulation were conducted and...To obtain the vertical earth pressure on a soft foundation box culvert and investigate the interaction of the soil-culvert-foundation system, both a centrifugal model test and a numerical simulation were conducted and the comparisons with the current methods to determine the load on a culvert were completed. The results of the model test and numerical analysis are in satisfactory agreement, which shows that the direction of the shear stress between the culvert and the adjacent embankment depends on the differential settlement between them. A vertical earth pressure concentration appears on the culvert with a rigid piles foundation because of a downward shear stress. The ratio of the load on a soft foundation culvert and the overburden pressure above the culvert raises first and then decreases as the backfill height increases. In order to reduce the load on a culvert, it is suggested to limit the stiffness difference of the foundations under the culvert and embankment and to use a light backfill over the culvert.展开更多
As the top of the pile foundation in high-pile wharf is connected to the superstructure and most of the pile bodies are located below the water surface, traditional damage detection methods are greatly limited in thei...As the top of the pile foundation in high-pile wharf is connected to the superstructure and most of the pile bodies are located below the water surface, traditional damage detection methods are greatly limited in their application to pile foundation in service. In the present study, a new method for pile foundation damage detection is developed based on the curve shape of the curvature mode difference(CMD) before and after damage. In the method, the influence at each node on the overall CMD curve shape is analyzed through a data deletion model, statistical characteristic indexes are established to reflect the difference between damaged and undamaged units, and structural damage is accurately detected. The effectiveness and robustness of the method are verified by a finite element model(FEM) of high-pile wharf under different damage conditions and different intensities of Gaussian white noise. The applicability of the method is then experimentally validated by a physical model of high-pile wharf. Both the FEM and the experimental results show that the method is capable of detecting pile foundation damage in noisy curvature mode and has strong application potential.展开更多
The traditional ground direct current method is not suitable for leakage detection of underground diaphragm walls in foundation pits because of its low accuracy and poor anti-noise ability.Here,we propose a joint surf...The traditional ground direct current method is not suitable for leakage detection of underground diaphragm walls in foundation pits because of its low accuracy and poor anti-noise ability.Here,we propose a joint surface-borehole observation device for leakage electric fi eld detection to achieve rapid measurement of the electric fi eld distribution characteristics at ground level in the foundation pit,thus enabling rapid localization of leakage points.We first establish the mechanism and basic equation of the leakage electric field response by combining the electric field formed by electrokinetic effect(EK)and the stable electric fi eld formed by conduction current in a combined leakage channel.Then,the fi nite–infi nite element coupling method is used to solve the electric fi eld equation to simulate the responses of a three-dimensional foundation pit leakage model.Furthermore,we conduct numerical simulations of diff erent pit models to investigate the infl uencing factors of the detection device and response characteristics of the change in the properties of the leakage channel.The results demonstrate that the proposed joint surface-borehole observation device can effi ciently reveal anomalous potential caused by leakage,and the amplitude of the electric fi eld generated by EK can eff ectively strengthen the leakage electric fi eld signal at the leakage,thus improving detection accuracy and effi ciency.展开更多
The influence of non-coaxial constitutive model on predictions of dense sand behavior is investigated in this paper. The non-coaxial model with strain softening plasticity is applied into finite-element program ABAQUS...The influence of non-coaxial constitutive model on predictions of dense sand behavior is investigated in this paper. The non-coaxial model with strain softening plasticity is applied into finite-element program ABAQUS, which is first used to predict the stress-strain behavior and the non-coaxial characteristic between the orientations of the principal stress and principal plastic strain rate in simple shear tests. The model is also used to predict load settlement responses and bearing capacity factors of shallow foundations. A series of centrifuge tests for shallow foundations on saturated dense sand are performed under drained conditions and the test results are compared with the corresponding numerical results. Various footing dimensions, depths of embedment, and footing shapes are considered in these tests. In view of the load settlement relationships, the stiffness of the load-displacement curves is significantly affected by the non-coaxial model compared with those predicted by the coaxial model, and a lower value of non-coaxial modulus gives a softer response. Considering the soil behavior at failure, the coaxial model predictions of bearing capacity factors are more advanced than those of centrifuge test results and the non-coaxial model results;besides, the non-coaxial model gives better predictions. The non-coaxial model predictions are closer to those of the centrifuge results when a proper non-coaxial plastic modulus is chosen.展开更多
To ensure the safety of buildings surrounding foundation pits, a study was made on a settlement monitoring and trend prediction method. A statistical testing method for analyzing the stability of a settlement monitori...To ensure the safety of buildings surrounding foundation pits, a study was made on a settlement monitoring and trend prediction method. A statistical testing method for analyzing the stability of a settlement monitoring datum has been discussed. According to a comprehensive survey, data of 16 stages at operating control point, were verified by a standard t test to determine the stability of the operating control point. A stationary auto-regression model, AR(p), used for the observation point settlement prediction has been investigated. Given the 16 stages of the settlement data at an observation point, the applicability of this model was analyzed. Settlement of last four stages was predicted using the stationary auto-regression model AR (1); the maximum difference between predicted and measured values was 0.6 mm, indicating good prediction results of the model. Hence, this model can be applied to settlement predictions for buildings surrounding foundation pits.展开更多
The need for simplified physical models representing frequency dependent soil impedances has been the motivation behind many researches throughout history. Generally, such models are generated to capture impedance fun...The need for simplified physical models representing frequency dependent soil impedances has been the motivation behind many researches throughout history. Generally, such models are generated to capture impedance functions in a wide range of excitation frequencies, which leads to relatively complex models. That is while there is just a limited range of frequencies that really influence the response of the structure. Here, a new methodology based on the response-matching concept is proposed, which can lead to the development of simpler discrete models. The idea is then used to upgrade an existing simple model of surface foundations to the case of embedded foundations. The applicability of the model in both frequency domain and time domain analyses of soil-structure systems with embedded foundations is discussed. Moreover, the accuracy of the results is compared with another existing discrete model for embedded foundations.展开更多
基金funded by the National Natural Science Foundation of China(Grant Nos.52368049,52168051,and 42462028)Lanzhou Young Scientific and Technological Talents Innovation Project(Grant Nos.2023-QN-27 and 2023-QN-52)Major Project of the Joint Scientific Research Fund of Gansu Province(Grant No.25JRRL007)。
文摘Deformations in high fill foundations comprising soil–stone mixtures must be accurately predicted to ensure construction quality and long-term operational safety.However,existing computational and analytical methods inadequately capture their complex mechanical behavior.We conducted a series of triaxial tests on unsaturated soil samples collected from a high fill project site in northwestern China under three stress paths.The incremental nonlinear and elastoplastic constitutive models for unsaturated soils were modified,and a calculation method was developed for the vertical and lateral deformations of high fill foundations using the layered summation approach.The results indicate that for soil samples with the same mixing ratio(m)and compaction coefficient(n),the strength of the sample and its tendency to exhibit shear dilation increase with the net confining pressure or matric suction.Additionally,the stress–strain curve of the soil sample gradually changes from the strain-hardening type to the ideal elastoplastic type as the compaction coefficient increases.Moreover,the compaction coefficient is an important factor influencing the magnitude of yield stress and yield suction in soil samples,and the yield points of both are similar in shape to the loadingcollapse(LC)and suction increase(SI)yield lines obtained using the Barcelona model in the net mean stress-generalized shear stress(p-s)plane,respectively.The modified incremental nonlinear instantaneous model simultaneously considers the effects of the compaction coefficient,suction and mixing ratio,and the model parameter can be simplified to the tangential modulus expression in the Duncan-Chang model when the suction is zero.Furthermore,the modified elastoplastic constitutive model,which considers the effects of the net mean stress,suction and partial stress,can be simplified to the elastoplastic constitutive relationship of saturated soil when the suction is zero.The proposed deformation calculation method,based on the layered summation theory,is applicable to both elastic and elastoplastic foundation states,as confirmed through numerical simulations.Our work can be used as a reference for the calculation of foundation deformation in similar mixed material high fill projects.
文摘Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.
基金supported by the National Natural Science Foundation of China(62225302,623B2014,and 62173023).
文摘With the emergence of general foundational models,such as Chat Generative Pre-trained Transformer(ChatGPT),researchers have shown considerable interest in the potential applications of foundation models in the process industry.This paper provides a comprehensive overview of the challenges and opportunities presented by the use of foundation models in the process industry,including the frameworks,core applications,and future prospects.First,this paper proposes a framework for foundation models for the process industry.Second,it summarizes the key capabilities of industrial foundation models and their practical applications.Finally,it highlights future research directions and identifies unresolved open issues related to the use of foundation models in the process industry.
基金supported by the National Natural Science Foundation of China(Grant No.52178329),the China Scholarship Council(Grant No.202306130155)the Postgraduate Scientific Research Innovation Project of Hunan Province,China(Grant No.CX20230442).
文摘With the continuous development of the offshore wind industry,the design concept of composite foundation has been given attention in the past decade.This paper presents an accurate method for investigating the horizontal vibration of monopile-friction wheel composite foundations in layered saturated soil.Firstly,the three-dimensional continuum mechanics theory with the range of linear elasticity is introduced to calculate the frictional resistance distributed on the upper soil surface.Then,the resistances of multilayered soils and inviscid seawater to the pile shaft under horizontal harmonic excitation are obtained using Novak's plane strain model,Biot's porous media theory and radiationwave theory.Thirdly,the expressions for the deformation,bending moment and internal force of the Euler-Bernoulli pile are derived using the boundary conditions with definitephysical meaning and transfer matrix method.By comparing with the results of 1g laboratory test and the idealized formula reported by the literature,the rationality and accuracy of the developed dynamical model can be verified.Finally,this paper conducts a series of worked examples to investigate the influencesof the elastic modulus and thickness of three-layer saturated soil and the location of interlayer soil on the horizontal dynamic vibration of composite foundation.The results show that an increase in elastic modulus of the surface soil is an effective way to improve the dynamic stability of the composite foundation in service conditions.The conclusions drawn from the numerical examples can develop some guidelines for the current foundation design of offshore wind turbines.
基金Supported by the Open Project Program of Panxi Crops Research and Utilization Key Laboratory of Sichuan Province,No.SZKF202302the Fundamental Research Funds for the Central Universities No.2019CDYGYB024.
文摘Gastrointestinal(GI)cancers represent a major global health concern due to their high incidence and mortality rates.Foundation models(FMs),also referred to as large models,represent a novel class of artificial intelligence technologies that have demonstrated considerable potential in addressing these challenges.These models encompass large language models(LLMs),vision FMs(VFMs),and multimodal LLMs(MLLMs),all of which utilize transformer architectures and self-supervised pre-training on extensive unlabeled datasets to achieve robust cross-domain generalization.This review delineates the principal applications of these models:LLMs facilitate the structuring of clinical narratives,extraction of insights from medical records,and enhancement of physician-patient communication;VFMs are employed in the analysis of endoscopic,radiological,and pathological images for lesion detection and staging;MLLMs integrate heterogeneous data modalities,including imaging,textual information,and genomic data,to support diagnostic processes,treatment prediction,and prognostic evaluation.Despite these promising developments,several challenges remain,such as the need for data standardization,limited diversity within training datasets,substantial computational resource requirements,and ethical-legal concerns.In conclusion,FMs exhibit significant potential to advance research and clinical management of GI cancers.Future research efforts should prioritize the refinement of these models,promote international collaborations,and adopt interdisciplinary approaches.Such a comprehensive strategy is essential to fully harness the capabilities of FMs,driving substantial progress in the fight against GI malignancies.
基金Projects(52008403,52378421)supported by the National Natural Science Foundation of ChinaProject(2022-Key-10)supported by the Science and Technology Research and Development Program Project of China Railway Group LimitedProject(202207)supported by the Hunan Provincial Transportation Science and Technology,China。
文摘The existing analytical models for umbrella arch method(UAM)based on elastic foundation beams often overlook the influence of the surrounding soil beyond the beam edges on the shear stresses acting on the beam.Consequently,such models fail to adequately reflect the continuity characteristics of soil deformation.Leveraging the Pasternak foundation-Euler beam model,this study considers the generalized shear force on the beam to account for the influence of soil outside the beam ends on the shear stress.An analytical model for the deformation and internal forces of finite-length beams subjected to arbitrary loads is derived based on the initial parameter method under various conditions.The mechanical model of the elastic foundation beam for advanced umbrella arch under typical tunnel excavation cycles is established,yielding analytical solutions for the longitudinal response of the umbrella arch.The reliability of the analytical model is verified with the existing test data.The improved model addresses anomalies in existing models,such as abnormal upward deformation in the loosened segment and maximum deflection occurring within the soil mass.Additionally,dimensionless characteristic parameters reflecting the relative stiffness between the umbrella arch structure and the foundation soil are proposed.Results indicate that the magnitude of soil characteristic parameters significantly influences the deformation and internal forces of the umbrella arch.Within common ranges of soil values,the maximum deformation and internal forces of the umbrella arch under semi-logarithmic coordinates exhibit nearly linear decay with decreasing soil characteristic parameters.The impact of tunnel excavation height on the stress of unsupported sections of the umbrella arch is minor,but it is more significant for umbrella arch buried within the soil mass.Conversely,the influence of tunnel excavation advance on the umbrella arch is opposite.
基金support from Strategic Project of Precision Surgery,Tsinghua UniversityInitiative Scientific Research Program,Institute for Intelligent Healthcare,Tsinghua University+5 种基金Tsinghua-Foshan Institute of Advanced ManufacturingNational Natural Science Foundation of China(61735016)Beijing Nova Program(20230484308)Young Elite Scientists Sponsorship Program by CAST(2023QNRC001)Youth Elite Program of Beijing Friendship Hospital(YYQCJH2022-9)Science and Technology Program of Beijing Tongzhou District(KJ2023CX012).
文摘Foundation models(FMs)have rapidly evolved and have achieved signicant accomplishments in computer vision tasks.Specically,the prompt mechanism conveniently allows users to integrate image prior information into the model,making it possible to apply models without any training.Therefore,we proposed a workflow based on foundation models and zero training to solve the tasks of photoacoustic(PA)image processing.We employed the Segment Anything Model(SAM)by setting simple prompts and integrating the model's outputs with prior knowledge of the imaged objects to accomplish various tasks,including:(1)removing the skin signal in three-dimensional PA image rendering;(2)dual speed-of-sound reconstruction,and(3)segmentation ofnger blood vessels.Through these demonstrations,we have concluded that FMs can be directly applied in PA imaging without the requirement for network design and training.This potentially allows for a hands-on,convenient approach to achieving efficient and accurate segmentation of PA images.This paper serves as a comprehensive tutorial,facilitating the mastery of the technique through the provision of code and sample datasets.
基金Funded by the Spanish Government and FEDER funds(AEI/FEDER,UE)under grant PID2021-124502OB-C42(PRESECREL)the predoctoral program“Concepción Arenal del Programa de Personal Investigador en formación Predoctoral”funded by Universidad de Cantabria and Cantabria’s Government(BOC 18-10-2021).
文摘Predictive maintenance often involves imbalanced multivariate time series datasets with scarce failure events,posing challenges for model training due to the high dimensionality of the data and the need for domain-specific preprocessing,which frequently leads to the development of large and complex models.Inspired by the success of Large Language Models(LLMs),transformer-based foundation models have been developed for time series(TSFM).These models have been proven to reconstruct time series in a zero-shot manner,being able to capture different patterns that effectively characterize time series.This paper proposes the use of TSFM to generate embeddings of the input data space,making them more interpretable for machine learning models.To evaluate the effectiveness of our approach,we trained three classical machine learning algorithms and one neural network using the embeddings generated by the TSFM called Moment for predicting the remaining useful life of aircraft engines.We test the models trained with both the full training dataset and only 10%of the training samples.Our results show that training simple models,such as support vector regressors or neural networks,with embeddings generated by Moment not only accelerates the training process but also enhances performance in few-shot learning scenarios,where data is scarce.This suggests a promising alternative to complex deep learning architectures,particularly in industrial contexts with limited labeled data.
基金financially supported by the National Natural Science Foundation of China(Grant No.12072204).
文摘To thoroughly examine the complex relationships between tire and pavement vibrations,a sophisticated vehicle-pavement coupled system is proposed,incorporating a non-uniform dynamic friction force between the tire and the pavement.According to the Timoshenko beam theory,a dynamic model of pavement structure with a finite length beam was formulated on a nonlinear Pasternak foundation.To more accurately describe the coupling relationship between the tire and the pavement,and to take into account the vibration state under vehicle-pavement interaction,the load distribution between the tire and the pavement is modeled as a dynamic non-uniform contact.Combined with the classic LuGre tire model,the adhesion between the tire and the pavement is calculated.The Galerkin truncation method is employed to transform the pavement vibration partial differential equation into a finite ordinary differential equation,and the integral expression of the nonlinear foundation beam term is derived using the product to sum formula.By using the Runge-Kutta method,the tire-road coupled system can be numerically calculated,thus determining tire adhesion.This research demonstrates that compared with tire force under the traditional static load distribution,load distribution has a significant influence on adhesion.This study offers valuable insights for pavement structure design and vehicle performance control.
基金the Deanship of Research and Graduate Studies at King Khalid University for funding this work through a large research project(No.RGP2/80/45)。
文摘The Euler-Bernoulli(E-B)beam theory is combined with Green-Lindsay's(G-L)generalized thermoelasticity theory to analyze the vibration of microbeams.The frequency control equation,based on the two-parameter Winkler-Pasternak elastic foundation for simply-supported microbeams,is presented.This study investigates the effects of the side-to-thickness ratio and relaxation time parameters on the vibrational natural frequency of thermoelastic microbeam resonators.The frequencies derived from the present model are compared with those from Lord and Shulman's(L-S)theory.The fourthorder solutions for natural vibration frequencies are graphically displayed for comparison.Therefore,attention should be paid to the use of effective foundations to prevent microbeam damage caused by contraction and expansion problems caused by high temperatures.
文摘Asia is the largest and most populous continent in the world;its security and stability play a crucial role in global peace and development.As the world is undergoing unprecedented transformations and entering a new period of turbulence and changes,Asia’s ascension is tempered by complex security challenges,both traditional and non-traditional.Recent geopolitical upheavals,from the Ukraine crisis to the Middle East turmoil,have spurred Asian countries to reassess international security dilemmas,exploring how to draw insights from the essence of Asian civilizational values to contribute positively to global security and stability.
基金We acknowledge funding from NSFC Grant 62306283.
文摘Since the 1950s,when the Turing Test was introduced,there has been notable progress in machine language intelligence.Language modeling,crucial for AI development,has evolved from statistical to neural models over the last two decades.Recently,transformer-based Pre-trained Language Models(PLM)have excelled in Natural Language Processing(NLP)tasks by leveraging large-scale training corpora.Increasing the scale of these models enhances performance significantly,introducing abilities like context learning that smaller models lack.The advancement in Large Language Models,exemplified by the development of ChatGPT,has made significant impacts both academically and industrially,capturing widespread societal interest.This survey provides an overview of the development and prospects from Large Language Models(LLM)to Large Multimodal Models(LMM).It first discusses the contributions and technological advancements of LLMs in the field of natural language processing,especially in text generation and language understanding.Then,it turns to the discussion of LMMs,which integrates various data modalities such as text,images,and sound,demonstrating advanced capabilities in understanding and generating cross-modal content,paving new pathways for the adaptability and flexibility of AI systems.Finally,the survey highlights the prospects of LMMs in terms of technological development and application potential,while also pointing out challenges in data integration,cross-modal understanding accuracy,providing a comprehensive perspective on the latest developments in this field.
基金Project(51878103)supported by the National Natural Science Foundation of ChinaProject(2016YFE0200100)supported by the National Key Research and Development Program of China。
文摘To reveal the bearing capacity of the X-section concrete piles pile raft foundation in silica sand,a series of vertical load tests are carried out.The X-section concrete piles are compared with circular section pile with the same section area.The load−settlement curves,axial force and skin friction,strain on concave and convex edge of the pile,pile-sand stress ratio,distributions of side and tip resistance are presented.The results show that bearing capacity of the X section concrete pile raft foundation is much larger than that of the circular pile raft foundation.Besides,compared with the circular pile,the peak axial force of X-section piles under raft is deeper and smaller while the neutral point of X-section concrete pile is deeper.Moreover,the strain on the concave edge is much larger than that on the convex edge of the pile,and the convex edge has more potential in bearing capacity as the vertical load increases.The X-section pile has higher pile-sand stress ratios and load-sharing between side resistance and tip resistance.Above all,the X-section concrete pile can significantly increase the bearing capacity of pile-raft foundations in silica sand.
基金Project(2012AA112504) supported by the National High Technology Research and Development Program of ChinaProjects(51108048,51478054) supported by the National Natural Science Foundation of China
文摘To obtain the vertical earth pressure on a soft foundation box culvert and investigate the interaction of the soil-culvert-foundation system, both a centrifugal model test and a numerical simulation were conducted and the comparisons with the current methods to determine the load on a culvert were completed. The results of the model test and numerical analysis are in satisfactory agreement, which shows that the direction of the shear stress between the culvert and the adjacent embankment depends on the differential settlement between them. A vertical earth pressure concentration appears on the culvert with a rigid piles foundation because of a downward shear stress. The ratio of the load on a soft foundation culvert and the overburden pressure above the culvert raises first and then decreases as the backfill height increases. In order to reduce the load on a culvert, it is suggested to limit the stiffness difference of the foundations under the culvert and embankment and to use a light backfill over the culvert.
基金financially supported by the National Natural Science Foundation of China (Grant Nos. 51709093 and 51679081)Fujian Provincial Department of Transportation Science and Technology Development Project (Grant No. 201708)Hohai University Student Innovation and Entrepreneurship Training Project (Grant No. 201910294014Z)。
文摘As the top of the pile foundation in high-pile wharf is connected to the superstructure and most of the pile bodies are located below the water surface, traditional damage detection methods are greatly limited in their application to pile foundation in service. In the present study, a new method for pile foundation damage detection is developed based on the curve shape of the curvature mode difference(CMD) before and after damage. In the method, the influence at each node on the overall CMD curve shape is analyzed through a data deletion model, statistical characteristic indexes are established to reflect the difference between damaged and undamaged units, and structural damage is accurately detected. The effectiveness and robustness of the method are verified by a finite element model(FEM) of high-pile wharf under different damage conditions and different intensities of Gaussian white noise. The applicability of the method is then experimentally validated by a physical model of high-pile wharf. Both the FEM and the experimental results show that the method is capable of detecting pile foundation damage in noisy curvature mode and has strong application potential.
基金partially supported by the National Natural Science Foundation of China (Nos. 41864004 and 41674077)Jiangxi Provincial Academic Leaders (Youth) Training Program (No. 20204BCJL23058)Open Fund from Engineering Research Center for Seismic Disaster Prevention and Engineering Geological Disaster Detection of Jiangxi Province (SDGD202102)。
文摘The traditional ground direct current method is not suitable for leakage detection of underground diaphragm walls in foundation pits because of its low accuracy and poor anti-noise ability.Here,we propose a joint surface-borehole observation device for leakage electric fi eld detection to achieve rapid measurement of the electric fi eld distribution characteristics at ground level in the foundation pit,thus enabling rapid localization of leakage points.We first establish the mechanism and basic equation of the leakage electric field response by combining the electric field formed by electrokinetic effect(EK)and the stable electric fi eld formed by conduction current in a combined leakage channel.Then,the fi nite–infi nite element coupling method is used to solve the electric fi eld equation to simulate the responses of a three-dimensional foundation pit leakage model.Furthermore,we conduct numerical simulations of diff erent pit models to investigate the infl uencing factors of the detection device and response characteristics of the change in the properties of the leakage channel.The results demonstrate that the proposed joint surface-borehole observation device can effi ciently reveal anomalous potential caused by leakage,and the amplitude of the electric fi eld generated by EK can eff ectively strengthen the leakage electric fi eld signal at the leakage,thus improving detection accuracy and effi ciency.
基金financially supported by the National Natural Science Foundation of China(Grant Nos.51079018 and 11202109)
文摘The influence of non-coaxial constitutive model on predictions of dense sand behavior is investigated in this paper. The non-coaxial model with strain softening plasticity is applied into finite-element program ABAQUS, which is first used to predict the stress-strain behavior and the non-coaxial characteristic between the orientations of the principal stress and principal plastic strain rate in simple shear tests. The model is also used to predict load settlement responses and bearing capacity factors of shallow foundations. A series of centrifuge tests for shallow foundations on saturated dense sand are performed under drained conditions and the test results are compared with the corresponding numerical results. Various footing dimensions, depths of embedment, and footing shapes are considered in these tests. In view of the load settlement relationships, the stiffness of the load-displacement curves is significantly affected by the non-coaxial model compared with those predicted by the coaxial model, and a lower value of non-coaxial modulus gives a softer response. Considering the soil behavior at failure, the coaxial model predictions of bearing capacity factors are more advanced than those of centrifuge test results and the non-coaxial model results;besides, the non-coaxial model gives better predictions. The non-coaxial model predictions are closer to those of the centrifuge results when a proper non-coaxial plastic modulus is chosen.
基金Project 50279005 supported by the National Natural Science Foundation of China
文摘To ensure the safety of buildings surrounding foundation pits, a study was made on a settlement monitoring and trend prediction method. A statistical testing method for analyzing the stability of a settlement monitoring datum has been discussed. According to a comprehensive survey, data of 16 stages at operating control point, were verified by a standard t test to determine the stability of the operating control point. A stationary auto-regression model, AR(p), used for the observation point settlement prediction has been investigated. Given the 16 stages of the settlement data at an observation point, the applicability of this model was analyzed. Settlement of last four stages was predicted using the stationary auto-regression model AR (1); the maximum difference between predicted and measured values was 0.6 mm, indicating good prediction results of the model. Hence, this model can be applied to settlement predictions for buildings surrounding foundation pits.
文摘The need for simplified physical models representing frequency dependent soil impedances has been the motivation behind many researches throughout history. Generally, such models are generated to capture impedance functions in a wide range of excitation frequencies, which leads to relatively complex models. That is while there is just a limited range of frequencies that really influence the response of the structure. Here, a new methodology based on the response-matching concept is proposed, which can lead to the development of simpler discrete models. The idea is then used to upgrade an existing simple model of surface foundations to the case of embedded foundations. The applicability of the model in both frequency domain and time domain analyses of soil-structure systems with embedded foundations is discussed. Moreover, the accuracy of the results is compared with another existing discrete model for embedded foundations.
