Effective partitioning is crucial for enabling parallel restoration of power systems after blackouts.This paper proposes a novel partitioning method based on deep reinforcement learning.First,the partitioning decision...Effective partitioning is crucial for enabling parallel restoration of power systems after blackouts.This paper proposes a novel partitioning method based on deep reinforcement learning.First,the partitioning decision process is formulated as a Markov decision process(MDP)model to maximize the modularity.Corresponding key partitioning constraints on parallel restoration are considered.Second,based on the partitioning objective and constraints,the reward function of the partitioning MDP model is set by adopting a relative deviation normalization scheme to reduce mutual interference between the reward and penalty in the reward function.The soft bonus scaling mechanism is introduced to mitigate overestimation caused by abrupt jumps in the reward.Then,the deep Q network method is applied to solve the partitioning MDP model and generate partitioning schemes.Two experience replay buffers are employed to speed up the training process of the method.Finally,case studies on the IEEE 39-bus test system demonstrate that the proposed method can generate a high-modularity partitioning result that meets all key partitioning constraints,thereby improving the parallelism and reliability of the restoration process.Moreover,simulation results demonstrate that an appropriate discount factor is crucial for ensuring both the convergence speed and the stability of the partitioning training.展开更多
This study develops a surrogate super-resolution(SR)framework that accelerates finite element method(FEM)-based computational fluid dynamics(CFD)using deep learning.High-resolution(HR)FEM-based CFDremains computationa...This study develops a surrogate super-resolution(SR)framework that accelerates finite element method(FEM)-based computational fluid dynamics(CFD)using deep learning.High-resolution(HR)FEM-based CFDremains computationally prohibitive for time-sensitive applications,including patient-specific aneurysm hemodynamics where rapid turnaround is valuable.The proposed pipeline learns to reconstruct HR velocity-magnitude fields fromlow-resolution(LR)FEM solutions generated under the same governing equations and boundary conditions.It consistsof three modules:(i)offline pre-training of a residual network on representative vascular geometries;(ii)lightweightfine-tuning to adapt the pretrained model to geometric variability,including patient-specific aneurysm morphologies;and(iii)an unstructured-to-structured sampling strategy with region-of-interest upsampling that concentrates resolution in flow-critical zones(e.g.,the aneurysm sac)rather than the full domain.This targeted reconstruction substantiallyreduces inference and post-processing cost while preserving key HR flow features.Experiments on cerebral aneurysmmodels show that HR velocity-magnitude fields can be recovered with accuracy comparable to direct HR simulationsat less than 1%of the direct HR simulation cost per analysis(LR simulation and SR inference),while adaptation to newgeometries requires only lightweight fine-tuning with limited target-specific HR data.While clinical endpoints andadditional variables(e.g.,pressure or wall-based metrics)are left for future work,the results indicate that the proposedsurrogate SR approach can streamline FEM-based CFD workflows toward near real-time hemodynamic analysis acrossmorphologically similar vascular models.展开更多
This study aimed to integrate Monte Carlo(MC)simulation with deep learning(DL)-based denoising techniques to achieve fast and accurate prediction of high-quality electronic portal imaging device(EPID)transmission dose...This study aimed to integrate Monte Carlo(MC)simulation with deep learning(DL)-based denoising techniques to achieve fast and accurate prediction of high-quality electronic portal imaging device(EPID)transmission dose(TD)for patientspecific quality assurance(PSQA).A total of 100 lung cases were used to obtain the noisy EPID TD by the ARCHER MC code under four kinds of particle numbers(1×10^(6),1×10^(7),1×10^(8)and 1×10^(9)),and the original EPID TD was denoised by the SUNet neural network.The denoised EPID TD was assessed both qualitatively and quantitatively using the structural similarity(SSIM),peak signal-to-noise ratio(PSNR),and gamma passing rate(GPR)with respect to 1×10^(9)as a reference.The computation times for both the MC simulation and DL-based denoising were recorded.As the number of particles increased,both the quality of the noisy EPID TD and computation time increased significantly(1×10^(6):1.12 s,1×10^(7):1.72 s,1×10^(8):8.62 s,and 1×10^(9):73.89 s).In contrast,the DL-based denoising time remained at 0.13-0.16 s.The denoised EPID TD shows a smoother visual appearance and profile curves,but differences between 1×10^(6)and 1×10^(9)still remain.SSIM improves from 0.61 to 0.95 for 1×10^(6),0.70 to 0.96 for 1×10^(7),and 0.90 to 0.97 for 1×10^(8).PSNR increases by>20%for 1×10^(6)and 1×10^(7),and>10%for 1×10^(8).GPR improves from 48.47%to 89.10%for 1×10^(6),61.04%to 94.35%for 1×10^(7),and 91.88%to 99.55%for 1×10^(8).The method that combines MC simulation with DL-based denoising for EPID TD generation can accelerate TD prediction and maintain high accuracy,offering a promising solution for efficient PSQA.展开更多
The stator of the maglev track plays a crucial role in the operation of the maglev system.Currently,the efficiency of maglev track inspection is limited by several factors,including the large span of elevated structur...The stator of the maglev track plays a crucial role in the operation of the maglev system.Currently,the efficiency of maglev track inspection is limited by several factors,including the large span of elevated structures,manual visual inspection,short inspection window times,and limited GPS positioning accuracy.To address these issues,this paper proposes a deep learning-based method for detecting and locating stator surface damage.This study establishes a maglev track stator surface image dataset,trains different object detection models,and compares their performance.Ultimately,YOLO and ByteTrack object tracking algorithms were chosen as the basic framework and enhanced to achieve automatic identification of high-speed maglev track stator surface damage images and track and count stator surface localization feature images.By matching the identified damaged images with their corresponding stator segment and beam segment sequence numbers,the location of the damage is pinpointed to the corresponding stator segment,enabling rapid and accurate identification and localization of complex damage to the maglev track stator surface.展开更多
Deep learning and fatigue life prediction remain focal research areas in rail vehicle engineering.This study addresses the vibration fatigue of wheelset lifting lug in Chengdu Metro Line 1 bogies,aiming to develop a f...Deep learning and fatigue life prediction remain focal research areas in rail vehicle engineering.This study addresses the vibration fatigue of wheelset lifting lug in Chengdu Metro Line 1 bogies,aiming to develop a fatigue life prediction method for critical bogie components using deep learning models and measured track load spectra.Extensive field tests on Chengdu Metro Line 1 were conducted to acquire acceleration and stress response data of the wheelset lifting lug,generating training samples for the neural network system.Component stress responses were calculated via time-domain track acceleration and validated against in-situ stress measurements.Results show that neural network-fitted dynamic stress values exhibit excellent consistency with measured data,with errors constrained within 5%.This study validates the proposed small-sample deep learning approach as an effective and accurate solution for fatigue life prediction of critical bogie components under operational load conditions.展开更多
Deep learning methods have achieved significant progress in solving partial differential equations.However,when applied to the widely used anisotropic scattering neutron transport equations in reactor engineering,thes...Deep learning methods have achieved significant progress in solving partial differential equations.However,when applied to the widely used anisotropic scattering neutron transport equations in reactor engineering,these encounter significant challenges.To address this issue,this study introduces a multi-antiderivative transformation alternating iterative deep learning method(M-AIM).This method transforms the integral terms of the scattering and fission sources in the transport equation into multiple antiderivative functions corresponding to the integrand,converts the differential-integral form of the transport equation into an exact differential equation,and establishes the necessary constraints for a unique solution.The M-AIM uses multiple deep neural networks to map the unknown angular flux density of transport equations and represents various forms of antiderivative functions.It constructs the corresponding weighted loss functions.By alternating iterative training with deep learning methods applied to these neural networks,the loss is reduced gradually.When the loss decreases to a preset minimum,the neural network approaches a numerical solution for both angular flux density and antiderivative functions.This paper presents a numerical verification of geometries such as flat plates and spheres.It verifies the validity of the theoretical framework and associated methods.The study contributes to the development of novel technical approaches for applying deep learning to solve anisotropic scattering neutron transport equations in reactor engineering.展开更多
With the continuous development of artificial intelligence and computer vision technology,numerous deep learning-based lane line detection methods have emerged.DeepLabv3+,as a classic semantic segmentation model,has f...With the continuous development of artificial intelligence and computer vision technology,numerous deep learning-based lane line detection methods have emerged.DeepLabv3+,as a classic semantic segmentation model,has found widespread application in the field of lane line detection.However,the accuracy of lane line segmentation is often compromised by factors such as changes in lighting conditions,occlusions,and wear and tear on the lane lines.Additionally,DeepLabv3+suffers from high memory consumption and challenges in deployment on embedded platforms.To address these issues,this paper proposes a lane line detection method for complex road scenes based on DeepLabv3+and MobileNetV4(MNv4).First,the lightweight MNv4 is adopted as the backbone network,and the standard convolutions in ASPP are replaced with depthwise separable convolutions.Second,a polarization attention mechanism is introduced after the ASPP module to enhance the model’s generalization capability.Finally,the Simple Linear Iterative Clustering(SLIC)superpixel segmentation algorithmis employed to preserve lane line edge information.MNv4-DeepLabv3+was tested on the TuSimple and CULane datasets.On the TuSimple dataset,theMean Intersection over Union(MIoU)and Mean Pixel Accuracy(mPA)improved by 1.01%and 7.49%,respectively.On the CULane dataset,MIoU andmPA increased by 3.33%and 7.74%,respectively.Thenumber of parameters decreased from 54.84 to 3.19 M.Experimental results demonstrate that MNv4-DeepLabv3+significantly optimizes model parameter count and enhances segmentation accuracy.展开更多
Structural Health Monitoring(SHM)plays a critical role in ensuring the safety,integrity,longevity and economic efficiency of civil infrastructures.The field has undergone a profound transformation over the last few de...Structural Health Monitoring(SHM)plays a critical role in ensuring the safety,integrity,longevity and economic efficiency of civil infrastructures.The field has undergone a profound transformation over the last few decades,evolving from traditional methods—often reliant on visual inspections—to data-driven intelligent systems.This review paper analyzes this historical trajectory,beginning with the approaches that relied on modal parameters as primary damage indicators.The advent of advanced sensor technologies and increased computational power brings a significant change,making Machine Learning(ML)a viable and powerful tool for damage assessment.More recently,Deep Learning(DL)has emerged as a paradigm shift,allowing for more automated processing of large data sets(such as the structural vibration signals and other types of sensors)with excellent performance and accuracy,often surpassing previous methods.This paper systematically reviews these technological milestones—from traditional vibration-based methods to the current state-of-the-art in deep learning.Finally,it critically examines emerging trends—such as Digital Twins and Transformer-based architectures—and discusses future research directions that will shape the next generation of SHM systems for civil engineering.展开更多
This systematic review aims to comprehensively examine and compare deep learning methods for brain tumor segmentation and classification using MRI and other imaging modalities,focusing on recent trends from 2022 to 20...This systematic review aims to comprehensively examine and compare deep learning methods for brain tumor segmentation and classification using MRI and other imaging modalities,focusing on recent trends from 2022 to 2025.The primary objective is to evaluate methodological advancements,model performance,dataset usage,and existing challenges in developing clinically robust AI systems.We included peer-reviewed journal articles and highimpact conference papers published between 2022 and 2025,written in English,that proposed or evaluated deep learning methods for brain tumor segmentation and/or classification.Excluded were non-open-access publications,books,and non-English articles.A structured search was conducted across Scopus,Google Scholar,Wiley,and Taylor&Francis,with the last search performed in August 2025.Risk of bias was not formally quantified but considered during full-text screening based on dataset diversity,validation methods,and availability of performance metrics.We used narrative synthesis and tabular benchmarking to compare performance metrics(e.g.,accuracy,Dice score)across model types(CNN,Transformer,Hybrid),imaging modalities,and datasets.A total of 49 studies were included(43 journal articles and 6 conference papers).These studies spanned over 9 public datasets(e.g.,BraTS,Figshare,REMBRANDT,MOLAB)and utilized a range of imaging modalities,predominantly MRI.Hybrid models,especially ResViT and UNetFormer,consistently achieved high performance,with classification accuracy exceeding 98%and segmentation Dice scores above 0.90 across multiple studies.Transformers and hybrid architectures showed increasing adoption post2023.Many studies lacked external validation and were evaluated only on a few benchmark datasets,raising concerns about generalizability and dataset bias.Few studies addressed clinical interpretability or uncertainty quantification.Despite promising results,particularly for hybrid deep learning models,widespread clinical adoption remains limited due to lack of validation,interpretability concerns,and real-world deployment barriers.展开更多
An image processing and deep learning method for identifying different types of rock images was proposed.Preprocessing,such as rock image acquisition,gray scaling,Gaussian blurring,and feature dimensionality reduction...An image processing and deep learning method for identifying different types of rock images was proposed.Preprocessing,such as rock image acquisition,gray scaling,Gaussian blurring,and feature dimensionality reduction,was conducted to extract useful feature information and recognize and classify rock images using Tensor Flow-based convolutional neural network(CNN)and Py Qt5.A rock image dataset was established and separated into workouts,confirmation sets,and test sets.The framework was subsequently compiled and trained.The categorization approach was evaluated using image data from the validation and test datasets,and key metrics,such as accuracy,precision,and recall,were analyzed.Finally,the classification model conducted a probabilistic analysis of the measured data to determine the equivalent lithological type for each image.The experimental results indicated that the method combining deep learning,Tensor Flow-based CNN,and Py Qt5 to recognize and classify rock images has an accuracy rate of up to 98.8%,and can be successfully utilized for rock image recognition.The system can be extended to geological exploration,mine engineering,and other rock and mineral resource development to more efficiently and accurately recognize rock samples.Moreover,it can match them with the intelligent support design system to effectively improve the reliability and economy of the support scheme.The system can serve as a reference for supporting the design of other mining and underground space projects.展开更多
As joint operations have become a key trend in modern military development,unmanned aerial vehicles(UAVs)play an increasingly important role in enhancing the intelligence and responsiveness of combat systems.However,t...As joint operations have become a key trend in modern military development,unmanned aerial vehicles(UAVs)play an increasingly important role in enhancing the intelligence and responsiveness of combat systems.However,the heterogeneity of aircraft,partial observability,and dynamic uncertainty in operational airspace pose significant challenges to autonomous collision avoidance using traditional methods.To address these issues,this paper proposes an adaptive collision avoidance approach for UAVs based on deep reinforcement learning.First,a unified uncertainty model incorporating dynamic wind fields is constructed to capture the complexity of joint operational environments.Then,to effectively handle the heterogeneity between manned and unmanned aircraft and the limitations of dynamic observations,a sector-based partial observation mechanism is designed.A Dynamic Threat Prioritization Assessment algorithm is also proposed to evaluate potential collision threats from multiple dimensions,including time to closest approach,minimum separation distance,and aircraft type.Furthermore,a Hierarchical Prioritized Experience Replay(HPER)mechanism is introduced,which classifies experience samples into high,medium,and low priority levels to preferentially sample critical experiences,thereby improving learning efficiency and accelerating policy convergence.Simulation results show that the proposed HPER-D3QN algorithm outperforms existing methods in terms of learning speed,environmental adaptability,and robustness,significantly enhancing collision avoidance performance and convergence rate.Finally,transfer experiments on a high-fidelity battlefield airspace simulation platform validate the proposed method's deployment potential and practical applicability in complex,real-world joint operational scenarios.展开更多
Bearing is an indispensable key component in mechanical equipment,and its working state is directly related to the stability and safety of the whole equipment.In recent years,with the rapid development of artificial i...Bearing is an indispensable key component in mechanical equipment,and its working state is directly related to the stability and safety of the whole equipment.In recent years,with the rapid development of artificial intelligence technology,especially the breakthrough of deep learning technology,it provides a new idea for bearing fault diagnosis.Deep learning can automatically learn features from a large amount of data,has a strong nonlinear modeling ability,and can effectively solve the problems existing in traditional methods.Aiming at the key problems in bearing fault diagnosis,this paper studies the fault diagnosis method based on deep learning,which not only provides a new solution for bearing fault diagnosis but also provides a reference for the application of deep learning in other mechanical fault diagnosis fields.展开更多
Natural gas hydrate(NGH)has a bright future as a clean energy source with huge reserves.Coring is one of the most direct methods for NGH exploration and research.Preserving the in-situ properties of the core as much a...Natural gas hydrate(NGH)has a bright future as a clean energy source with huge reserves.Coring is one of the most direct methods for NGH exploration and research.Preserving the in-situ properties of the core as much as possible during the coring process is crucial for the assessment of NGH resources.However,most existing NGH coring techniques cannot preserve the in-situ temperature of NGH,leading to distortion of the physical properties of the obtained core,which makes it difficult to effectively guide NGH exploration and development.To overcome this limitation,this study introduces an innovative active temperature-preserved coring method for NGH utilizing phase change materials(PCM).An active temperature-preserved corer(ATPC)is designed and developed,and an indoor experimental system is established to investigate the heat transfer during the coring process.Based on the experimental results under different environment temperatures,a heat transfer model for the entire ATPC coring process has been established.The indoor experimental results are consistent with the theoretical predictions of the heat transfer model,confirming its validity.This model has reconstructed the temperature changes of the NGH core during the coring process,demonstrating that compared to the traditional coring method with only passive temperature-preserved measures,ATPC can effectively reduce the core temperature by more than 5.25℃.With ATPC,at environment temperatures of 15,20,25,and 30℃,the duration of low-temperature state for the NGH core is 53.85,32.87,20.32,and 11.83 min,respectively.These findings provide new perspectives on temperature-preserving core sampling in NGH and provide technical support for exploration and development in NGH.展开更多
To improve the accuracy of thermal response estimation and overcome the limitations of the linear regression model and Artificial Neural Network(ANN)model,this study introduces a deep learning estimation method specif...To improve the accuracy of thermal response estimation and overcome the limitations of the linear regression model and Artificial Neural Network(ANN)model,this study introduces a deep learning estimation method specifically based on the Long Short-Term Memory(LSTM)network,to predict temperature-induced girder end displacements of the Dasha Waterway Bridge,a suspension bridge in China.First,to enhance data quality and select target sensors,preprocessing based on the sigma rule and nearest neighbor interpolation is applied to the raw data.Furthermore,to eliminate the high-frequency components from the displacement signal,the wavelet transform is conducted.Subsequently,the linear regression model and ANN model are established,whose results do not meet the requirements and fail to address the time lag effect between temperature and displacements.The study proceeds to develop the LSTM network model and determine the optimal parameters through hyperparameter sensitivity analysis.Finally,the results of the LSTM network model are discussed by a comparative analysis against the linear regression model and ANN model,which indicates a higher accuracy in predicting temperatureinduced girder end displacements and the ability to mitigate the time-lag effect.To be more specific,in comparison between the linear regression model and LSTM network,the mean square error decreases from 6.5937 to 1.6808 and R^(2) increases from 0.683 to 0.930,which corresponds to a 74.51%decrease in MSE and a 36.14%improvement in R^(2).Compared to ANN,with an MSE of 4.6371 and an R^(2) of 0.807,LSTM shows a decrease in MSE of 63.75%and an increase in R^(2) of 13.23%,demonstrating a significant enhancement in predictive performance.展开更多
Seasonal prediction of summer rainfall in China plays a crucial role in decision-making,environmental protection,and socio-economic development,while it currently has a low prediction skill.We developed a deep learnin...Seasonal prediction of summer rainfall in China plays a crucial role in decision-making,environmental protection,and socio-economic development,while it currently has a low prediction skill.We developed a deep learning-based seasonal prediction bias correction method for summer rainfall in China.Based on prediction fields from the flexible Global Ocean-Atmosphere-Land System Model finite volume version 2(FGOALS-f2),we optimized the loss function of U-Net,trained with different hyperparameters,and selected the optimum model.U-Net model can extract multi-scale feature information and preserve spatial information,making it suitable for processing meteorological data.With this endto-end model,the precipitation distribution can be obtained directly without using the traditional method of data dimensionality reduction(e.g.,Empirical Orthogonal Function),which could maximize the retention of spatio-temporal information of the input data.Optimization of the loss function enhances the prediction results and mitigates model overfitting.The independent prediction shows a significant skill improvement measured by the anomalous correlation coefficient score.The skill has an average value of 0.679 in China(0°–63°N,73°–133°E)and 0.691 in the region of the Chinese mainland,which significantly improves the dynamical prediction skill by 1357%and 4836%.This study suggests that the deep learning(U-Net)-based seasonal prediction bias correction method is a promising approach for improving rainfall prediction of the dynamical model.展开更多
Deep learning-based object detection has revolutionized various fields,including agriculture.This paper presents a systematic review based on the PRISMA 2020 approach for object detection techniques in agriculture by ...Deep learning-based object detection has revolutionized various fields,including agriculture.This paper presents a systematic review based on the PRISMA 2020 approach for object detection techniques in agriculture by exploring the evolution of different methods and applications over the past three years,highlighting the shift from conventional computer vision to deep learning-based methodologies owing to their enhanced efficacy in real time.The review emphasizes the integration of advanced models,such as You Only Look Once(YOLO)v9,v10,EfficientDet,Transformer-based models,and hybrid frameworks that improve the precision,accuracy,and scalability for crop monitoring and disease detection.The review also highlights benchmark datasets and evaluation metrics.It addresses limitations,like domain adaptation challenges,dataset heterogeneity,and occlusion,while offering insights into prospective research avenues,such as multimodal learning,explainable AI,and federated learning.Furthermore,the main aim of this paper is to serve as a thorough resource guide for scientists,researchers,and stakeholders for implementing deep learning-based object detection methods for the development of intelligent,robust,and sustainable agricultural systems.展开更多
Currently,regression prediction methods based on logging data is one of the main methods for analyzing gas content of coal seams.However,the complexity of logging parameters for deep coal seams and the scarcity of mea...Currently,regression prediction methods based on logging data is one of the main methods for analyzing gas content of coal seams.However,the complexity of logging parameters for deep coal seams and the scarcity of measured gas content data signifcantly afects the accuracy and generalizability of data regression models.Accurately predicting the gas content of coal seams under small-sample condition become a difcult point in deep coalbed methane(CBM)exploration.The ModelAgnostic Meta-Learning(MAML)and Support Vector Regression(SVR)algorithms are among the few suitable for smallsample learning,exhibiting strong adaptability under limited sample conditions.In this study,logging parameters are used as input variables to construct MAML and SVR models,and their performance in predicting gas content of deep coal seams across diferent regions and layers is compared.The results demonstrate that the MAML algorithm efectively addresses the complex relationships between gas content of deep coal seam and logging parameters.The prediction errors for test dataset and new samples are merely 3.61%and 4.52%respectively,indicating exceptional adaptability,robust generalization capability,and stable model performance.In contrast,the dependency of SVR model on input parameters restricts its accuracy and generalizability in predicting gas content in deep coal seams with varying geological conditions.Although achieving a test dataset error of 4.71%,the SVR model demonstrates substantially degraded performance when applied to novel samples,with prediction errors escalating to 12.46%.Therefore,the MAML model is selected to predict gas content in the unknown areas of the Baijiahai region.The prediction results reveal that the gas content of coal seams in the Xishanyao formation(J2x)ranges from 1.32 m^(3)/t to 16.11 m^(3)/t,while that in the Badaowan Formation(J1b)varies between 1.73 m^(3)/t and 11.27 m^(3)/t.Notably,the gas enrichment areas are predominantly distributed in well blocks adjacent to fault systems,such as wells C31 and BJ8,etc.,which align with the favorable geological conditions for deep CBM accumulation in the Baijiahai region.These spatial distribution patterns not only corroborate existing geological insights but also further validate the reliability of the MAML model in predicting gas content within deep coal seams.展开更多
Porosity is an important attribute for evaluating the petrophysical properties of reservoirs, and has guiding significance for the exploration and development of oil and gas. The seismic inversion is a key method for ...Porosity is an important attribute for evaluating the petrophysical properties of reservoirs, and has guiding significance for the exploration and development of oil and gas. The seismic inversion is a key method for comprehensively obtaining the porosity. Deep learning methods provide an intelligent approach to suppress the ambiguity of the conventional inversion method. However, under the trace-bytrace inversion strategy, there is a lack of constraints from geological structural information, resulting in poor lateral continuity of prediction results. In addition, the heterogeneity and the sedimentary variability of subsurface media also lead to uncertainty in intelligent prediction. To achieve fine prediction of porosity, we consider the lateral continuity and variability and propose an improved structural modeling deep learning porosity prediction method. First, we combine well data, waveform attributes, and structural information as constraints to model geophysical parameters, constructing a high-quality training dataset with sedimentary facies-controlled significance. Subsequently, we introduce a gated axial attention mechanism to enhance the features of dataset and design a bidirectional closed-loop network system constrained by inversion and forward processes. The constraint coefficient is adaptively adjusted by the petrophysical information contained between the porosity and impedance in the study area. We demonstrate the effectiveness of the adaptive coefficient through numerical experiments.Finally, we compare the performance differences between the proposed method and conventional deep learning methods using data from two study areas. The proposed method achieves better consistency with the logging porosity, demonstrating the superiority of the proposed method.展开更多
The accurate identification of microporosity is crucial for the characterization of hydrocarbon reservoir permeability and production.Scanning electron microscopy(SEM)is among the limited number of methods available t...The accurate identification of microporosity is crucial for the characterization of hydrocarbon reservoir permeability and production.Scanning electron microscopy(SEM)is among the limited number of methods available to directly observe the microscopic structure of the hydrocarbon reservoir rocks.Nevertheless,precise segmentation of microscopic pores at different depths in SEM images remains an unsolved challenge,known as the‘depth-related resolution loss'problem.Therefore,in this study,a 3D reconstruction technique for regions of interest(ROI)was developed for in-depth pixel analysis and differentiation among various depths of SEM images.The processed SEM images,together with the processing outcomes of this technique,were used as the input database to train a stochastic depth with multi-channel residual pathways(SdstMcrp)deep learning model programmed in Python to develop a tool for segmenting the microscopic pore spaces in SEM images obtained from the Beibuwan Basin.The more accurate segmentation helped to detect an average of 1.2 times more microporosity in SEM images,accounting for about 1.6 times more pixels and 1.2 times more pore surface area.Finally,the impact of the accurate segmentation on the calculation of permeability,a significant reservoir production property,was investigated using fractal geometry models and sensitivity analysis.The results showed that the obtained permeability values would vary by a factor of 6,which represents a considerable difference.These findings demonstrate that the proposed models can effectively identify features across a wide range of grayscale values in SEM images.展开更多
基金funded by the Beijing Engineering Research Center of Electric Rail Transportation.
文摘Effective partitioning is crucial for enabling parallel restoration of power systems after blackouts.This paper proposes a novel partitioning method based on deep reinforcement learning.First,the partitioning decision process is formulated as a Markov decision process(MDP)model to maximize the modularity.Corresponding key partitioning constraints on parallel restoration are considered.Second,based on the partitioning objective and constraints,the reward function of the partitioning MDP model is set by adopting a relative deviation normalization scheme to reduce mutual interference between the reward and penalty in the reward function.The soft bonus scaling mechanism is introduced to mitigate overestimation caused by abrupt jumps in the reward.Then,the deep Q network method is applied to solve the partitioning MDP model and generate partitioning schemes.Two experience replay buffers are employed to speed up the training process of the method.Finally,case studies on the IEEE 39-bus test system demonstrate that the proposed method can generate a high-modularity partitioning result that meets all key partitioning constraints,thereby improving the parallelism and reliability of the restoration process.Moreover,simulation results demonstrate that an appropriate discount factor is crucial for ensuring both the convergence speed and the stability of the partitioning training.
文摘This study develops a surrogate super-resolution(SR)framework that accelerates finite element method(FEM)-based computational fluid dynamics(CFD)using deep learning.High-resolution(HR)FEM-based CFDremains computationally prohibitive for time-sensitive applications,including patient-specific aneurysm hemodynamics where rapid turnaround is valuable.The proposed pipeline learns to reconstruct HR velocity-magnitude fields fromlow-resolution(LR)FEM solutions generated under the same governing equations and boundary conditions.It consistsof three modules:(i)offline pre-training of a residual network on representative vascular geometries;(ii)lightweightfine-tuning to adapt the pretrained model to geometric variability,including patient-specific aneurysm morphologies;and(iii)an unstructured-to-structured sampling strategy with region-of-interest upsampling that concentrates resolution in flow-critical zones(e.g.,the aneurysm sac)rather than the full domain.This targeted reconstruction substantiallyreduces inference and post-processing cost while preserving key HR flow features.Experiments on cerebral aneurysmmodels show that HR velocity-magnitude fields can be recovered with accuracy comparable to direct HR simulationsat less than 1%of the direct HR simulation cost per analysis(LR simulation and SR inference),while adaptation to newgeometries requires only lightweight fine-tuning with limited target-specific HR data.While clinical endpoints andadditional variables(e.g.,pressure or wall-based metrics)are left for future work,the results indicate that the proposedsurrogate SR approach can streamline FEM-based CFD workflows toward near real-time hemodynamic analysis acrossmorphologically similar vascular models.
基金supported by National Key R&D Program of China(No.2022YFC2404604)Chongqing Research Institution Performance Incentive Guidance Special Project(No.CSTB2023JXJL-YFX0080)Chongqing Medical Scientific Research Project(Joint project of Chongqing Health Commission and Science and Technology Bureau)(No.2022DBXM005)。
文摘This study aimed to integrate Monte Carlo(MC)simulation with deep learning(DL)-based denoising techniques to achieve fast and accurate prediction of high-quality electronic portal imaging device(EPID)transmission dose(TD)for patientspecific quality assurance(PSQA).A total of 100 lung cases were used to obtain the noisy EPID TD by the ARCHER MC code under four kinds of particle numbers(1×10^(6),1×10^(7),1×10^(8)and 1×10^(9)),and the original EPID TD was denoised by the SUNet neural network.The denoised EPID TD was assessed both qualitatively and quantitatively using the structural similarity(SSIM),peak signal-to-noise ratio(PSNR),and gamma passing rate(GPR)with respect to 1×10^(9)as a reference.The computation times for both the MC simulation and DL-based denoising were recorded.As the number of particles increased,both the quality of the noisy EPID TD and computation time increased significantly(1×10^(6):1.12 s,1×10^(7):1.72 s,1×10^(8):8.62 s,and 1×10^(9):73.89 s).In contrast,the DL-based denoising time remained at 0.13-0.16 s.The denoised EPID TD shows a smoother visual appearance and profile curves,but differences between 1×10^(6)and 1×10^(9)still remain.SSIM improves from 0.61 to 0.95 for 1×10^(6),0.70 to 0.96 for 1×10^(7),and 0.90 to 0.97 for 1×10^(8).PSNR increases by>20%for 1×10^(6)and 1×10^(7),and>10%for 1×10^(8).GPR improves from 48.47%to 89.10%for 1×10^(6),61.04%to 94.35%for 1×10^(7),and 91.88%to 99.55%for 1×10^(8).The method that combines MC simulation with DL-based denoising for EPID TD generation can accelerate TD prediction and maintain high accuracy,offering a promising solution for efficient PSQA.
基金supported in part by the National Natural Science Foundation of China under Grant 52432012in part by the Shanghai Science and Technology Project with 25ZR1402508。
文摘The stator of the maglev track plays a crucial role in the operation of the maglev system.Currently,the efficiency of maglev track inspection is limited by several factors,including the large span of elevated structures,manual visual inspection,short inspection window times,and limited GPS positioning accuracy.To address these issues,this paper proposes a deep learning-based method for detecting and locating stator surface damage.This study establishes a maglev track stator surface image dataset,trains different object detection models,and compares their performance.Ultimately,YOLO and ByteTrack object tracking algorithms were chosen as the basic framework and enhanced to achieve automatic identification of high-speed maglev track stator surface damage images and track and count stator surface localization feature images.By matching the identified damaged images with their corresponding stator segment and beam segment sequence numbers,the location of the damage is pinpointed to the corresponding stator segment,enabling rapid and accurate identification and localization of complex damage to the maglev track stator surface.
基金supported by the CRRC Original Technology TenYear Cultivation Program(Grant No.2022CYY007)。
文摘Deep learning and fatigue life prediction remain focal research areas in rail vehicle engineering.This study addresses the vibration fatigue of wheelset lifting lug in Chengdu Metro Line 1 bogies,aiming to develop a fatigue life prediction method for critical bogie components using deep learning models and measured track load spectra.Extensive field tests on Chengdu Metro Line 1 were conducted to acquire acceleration and stress response data of the wheelset lifting lug,generating training samples for the neural network system.Component stress responses were calculated via time-domain track acceleration and validated against in-situ stress measurements.Results show that neural network-fitted dynamic stress values exhibit excellent consistency with measured data,with errors constrained within 5%.This study validates the proposed small-sample deep learning approach as an effective and accurate solution for fatigue life prediction of critical bogie components under operational load conditions.
基金supported by the National Natural Science Foundation of China(No.12575189)。
文摘Deep learning methods have achieved significant progress in solving partial differential equations.However,when applied to the widely used anisotropic scattering neutron transport equations in reactor engineering,these encounter significant challenges.To address this issue,this study introduces a multi-antiderivative transformation alternating iterative deep learning method(M-AIM).This method transforms the integral terms of the scattering and fission sources in the transport equation into multiple antiderivative functions corresponding to the integrand,converts the differential-integral form of the transport equation into an exact differential equation,and establishes the necessary constraints for a unique solution.The M-AIM uses multiple deep neural networks to map the unknown angular flux density of transport equations and represents various forms of antiderivative functions.It constructs the corresponding weighted loss functions.By alternating iterative training with deep learning methods applied to these neural networks,the loss is reduced gradually.When the loss decreases to a preset minimum,the neural network approaches a numerical solution for both angular flux density and antiderivative functions.This paper presents a numerical verification of geometries such as flat plates and spheres.It verifies the validity of the theoretical framework and associated methods.The study contributes to the development of novel technical approaches for applying deep learning to solve anisotropic scattering neutron transport equations in reactor engineering.
文摘With the continuous development of artificial intelligence and computer vision technology,numerous deep learning-based lane line detection methods have emerged.DeepLabv3+,as a classic semantic segmentation model,has found widespread application in the field of lane line detection.However,the accuracy of lane line segmentation is often compromised by factors such as changes in lighting conditions,occlusions,and wear and tear on the lane lines.Additionally,DeepLabv3+suffers from high memory consumption and challenges in deployment on embedded platforms.To address these issues,this paper proposes a lane line detection method for complex road scenes based on DeepLabv3+and MobileNetV4(MNv4).First,the lightweight MNv4 is adopted as the backbone network,and the standard convolutions in ASPP are replaced with depthwise separable convolutions.Second,a polarization attention mechanism is introduced after the ASPP module to enhance the model’s generalization capability.Finally,the Simple Linear Iterative Clustering(SLIC)superpixel segmentation algorithmis employed to preserve lane line edge information.MNv4-DeepLabv3+was tested on the TuSimple and CULane datasets.On the TuSimple dataset,theMean Intersection over Union(MIoU)and Mean Pixel Accuracy(mPA)improved by 1.01%and 7.49%,respectively.On the CULane dataset,MIoU andmPA increased by 3.33%and 7.74%,respectively.Thenumber of parameters decreased from 54.84 to 3.19 M.Experimental results demonstrate that MNv4-DeepLabv3+significantly optimizes model parameter count and enhances segmentation accuracy.
基金The authors would like to thank CNPq(Conselho Nacional de Desenvolvimento Científico e Tecnológico)—grants 407256/2022-9,303550/2025-2,402533/2023-2 and 303982/2022-5FAPEMIG(Fundação de AmparoàPesquisa do Estado de Minas Gerais)—grants APQ-00032-24 and APD-01113-25 for their financial support.
文摘Structural Health Monitoring(SHM)plays a critical role in ensuring the safety,integrity,longevity and economic efficiency of civil infrastructures.The field has undergone a profound transformation over the last few decades,evolving from traditional methods—often reliant on visual inspections—to data-driven intelligent systems.This review paper analyzes this historical trajectory,beginning with the approaches that relied on modal parameters as primary damage indicators.The advent of advanced sensor technologies and increased computational power brings a significant change,making Machine Learning(ML)a viable and powerful tool for damage assessment.More recently,Deep Learning(DL)has emerged as a paradigm shift,allowing for more automated processing of large data sets(such as the structural vibration signals and other types of sensors)with excellent performance and accuracy,often surpassing previous methods.This paper systematically reviews these technological milestones—from traditional vibration-based methods to the current state-of-the-art in deep learning.Finally,it critically examines emerging trends—such as Digital Twins and Transformer-based architectures—and discusses future research directions that will shape the next generation of SHM systems for civil engineering.
文摘This systematic review aims to comprehensively examine and compare deep learning methods for brain tumor segmentation and classification using MRI and other imaging modalities,focusing on recent trends from 2022 to 2025.The primary objective is to evaluate methodological advancements,model performance,dataset usage,and existing challenges in developing clinically robust AI systems.We included peer-reviewed journal articles and highimpact conference papers published between 2022 and 2025,written in English,that proposed or evaluated deep learning methods for brain tumor segmentation and/or classification.Excluded were non-open-access publications,books,and non-English articles.A structured search was conducted across Scopus,Google Scholar,Wiley,and Taylor&Francis,with the last search performed in August 2025.Risk of bias was not formally quantified but considered during full-text screening based on dataset diversity,validation methods,and availability of performance metrics.We used narrative synthesis and tabular benchmarking to compare performance metrics(e.g.,accuracy,Dice score)across model types(CNN,Transformer,Hybrid),imaging modalities,and datasets.A total of 49 studies were included(43 journal articles and 6 conference papers).These studies spanned over 9 public datasets(e.g.,BraTS,Figshare,REMBRANDT,MOLAB)and utilized a range of imaging modalities,predominantly MRI.Hybrid models,especially ResViT and UNetFormer,consistently achieved high performance,with classification accuracy exceeding 98%and segmentation Dice scores above 0.90 across multiple studies.Transformers and hybrid architectures showed increasing adoption post2023.Many studies lacked external validation and were evaluated only on a few benchmark datasets,raising concerns about generalizability and dataset bias.Few studies addressed clinical interpretability or uncertainty quantification.Despite promising results,particularly for hybrid deep learning models,widespread clinical adoption remains limited due to lack of validation,interpretability concerns,and real-world deployment barriers.
基金financially supported by the National Science and Technology Major Project——Deep Earth Probe and Mineral Resources Exploration(No.2024ZD1003701)the National Key R&D Program of China(No.2022YFC2905004)。
文摘An image processing and deep learning method for identifying different types of rock images was proposed.Preprocessing,such as rock image acquisition,gray scaling,Gaussian blurring,and feature dimensionality reduction,was conducted to extract useful feature information and recognize and classify rock images using Tensor Flow-based convolutional neural network(CNN)and Py Qt5.A rock image dataset was established and separated into workouts,confirmation sets,and test sets.The framework was subsequently compiled and trained.The categorization approach was evaluated using image data from the validation and test datasets,and key metrics,such as accuracy,precision,and recall,were analyzed.Finally,the classification model conducted a probabilistic analysis of the measured data to determine the equivalent lithological type for each image.The experimental results indicated that the method combining deep learning,Tensor Flow-based CNN,and Py Qt5 to recognize and classify rock images has an accuracy rate of up to 98.8%,and can be successfully utilized for rock image recognition.The system can be extended to geological exploration,mine engineering,and other rock and mineral resource development to more efficiently and accurately recognize rock samples.Moreover,it can match them with the intelligent support design system to effectively improve the reliability and economy of the support scheme.The system can serve as a reference for supporting the design of other mining and underground space projects.
基金supported by the National Key Research and Development Program of China(No.2022YFB4300902).
文摘As joint operations have become a key trend in modern military development,unmanned aerial vehicles(UAVs)play an increasingly important role in enhancing the intelligence and responsiveness of combat systems.However,the heterogeneity of aircraft,partial observability,and dynamic uncertainty in operational airspace pose significant challenges to autonomous collision avoidance using traditional methods.To address these issues,this paper proposes an adaptive collision avoidance approach for UAVs based on deep reinforcement learning.First,a unified uncertainty model incorporating dynamic wind fields is constructed to capture the complexity of joint operational environments.Then,to effectively handle the heterogeneity between manned and unmanned aircraft and the limitations of dynamic observations,a sector-based partial observation mechanism is designed.A Dynamic Threat Prioritization Assessment algorithm is also proposed to evaluate potential collision threats from multiple dimensions,including time to closest approach,minimum separation distance,and aircraft type.Furthermore,a Hierarchical Prioritized Experience Replay(HPER)mechanism is introduced,which classifies experience samples into high,medium,and low priority levels to preferentially sample critical experiences,thereby improving learning efficiency and accelerating policy convergence.Simulation results show that the proposed HPER-D3QN algorithm outperforms existing methods in terms of learning speed,environmental adaptability,and robustness,significantly enhancing collision avoidance performance and convergence rate.Finally,transfer experiments on a high-fidelity battlefield airspace simulation platform validate the proposed method's deployment potential and practical applicability in complex,real-world joint operational scenarios.
文摘Bearing is an indispensable key component in mechanical equipment,and its working state is directly related to the stability and safety of the whole equipment.In recent years,with the rapid development of artificial intelligence technology,especially the breakthrough of deep learning technology,it provides a new idea for bearing fault diagnosis.Deep learning can automatically learn features from a large amount of data,has a strong nonlinear modeling ability,and can effectively solve the problems existing in traditional methods.Aiming at the key problems in bearing fault diagnosis,this paper studies the fault diagnosis method based on deep learning,which not only provides a new solution for bearing fault diagnosis but also provides a reference for the application of deep learning in other mechanical fault diagnosis fields.
基金financially supported by Shenzhen Science and Technology Program(Nos.JSGG20220831105002005 and KJZD20231025152759002)the National Natural Science Foundation of China(Nos.52274133 and 523B2101).
文摘Natural gas hydrate(NGH)has a bright future as a clean energy source with huge reserves.Coring is one of the most direct methods for NGH exploration and research.Preserving the in-situ properties of the core as much as possible during the coring process is crucial for the assessment of NGH resources.However,most existing NGH coring techniques cannot preserve the in-situ temperature of NGH,leading to distortion of the physical properties of the obtained core,which makes it difficult to effectively guide NGH exploration and development.To overcome this limitation,this study introduces an innovative active temperature-preserved coring method for NGH utilizing phase change materials(PCM).An active temperature-preserved corer(ATPC)is designed and developed,and an indoor experimental system is established to investigate the heat transfer during the coring process.Based on the experimental results under different environment temperatures,a heat transfer model for the entire ATPC coring process has been established.The indoor experimental results are consistent with the theoretical predictions of the heat transfer model,confirming its validity.This model has reconstructed the temperature changes of the NGH core during the coring process,demonstrating that compared to the traditional coring method with only passive temperature-preserved measures,ATPC can effectively reduce the core temperature by more than 5.25℃.With ATPC,at environment temperatures of 15,20,25,and 30℃,the duration of low-temperature state for the NGH core is 53.85,32.87,20.32,and 11.83 min,respectively.These findings provide new perspectives on temperature-preserving core sampling in NGH and provide technical support for exploration and development in NGH.
基金The National Key Research and Development Program of China grant No.2022YFB3706704 received by Yuan Renthe National Natural and Science Foundation of China grant No.52308150 received by Xiang Xu.
文摘To improve the accuracy of thermal response estimation and overcome the limitations of the linear regression model and Artificial Neural Network(ANN)model,this study introduces a deep learning estimation method specifically based on the Long Short-Term Memory(LSTM)network,to predict temperature-induced girder end displacements of the Dasha Waterway Bridge,a suspension bridge in China.First,to enhance data quality and select target sensors,preprocessing based on the sigma rule and nearest neighbor interpolation is applied to the raw data.Furthermore,to eliminate the high-frequency components from the displacement signal,the wavelet transform is conducted.Subsequently,the linear regression model and ANN model are established,whose results do not meet the requirements and fail to address the time lag effect between temperature and displacements.The study proceeds to develop the LSTM network model and determine the optimal parameters through hyperparameter sensitivity analysis.Finally,the results of the LSTM network model are discussed by a comparative analysis against the linear regression model and ANN model,which indicates a higher accuracy in predicting temperatureinduced girder end displacements and the ability to mitigate the time-lag effect.To be more specific,in comparison between the linear regression model and LSTM network,the mean square error decreases from 6.5937 to 1.6808 and R^(2) increases from 0.683 to 0.930,which corresponds to a 74.51%decrease in MSE and a 36.14%improvement in R^(2).Compared to ANN,with an MSE of 4.6371 and an R^(2) of 0.807,LSTM shows a decrease in MSE of 63.75%and an increase in R^(2) of 13.23%,demonstrating a significant enhancement in predictive performance.
基金Guangdong Major Project of Basic and Applied Basic Research(2020B0301030004)Postdoctoral Fellowship Program of CPSF(GZC20232598)+1 种基金China Postdoctoral Science Foundation(2024M753168)National Key Scientific and Technological Infrastructure Project“Earth System Numerical Simulation Facility”(EarthLab)。
文摘Seasonal prediction of summer rainfall in China plays a crucial role in decision-making,environmental protection,and socio-economic development,while it currently has a low prediction skill.We developed a deep learning-based seasonal prediction bias correction method for summer rainfall in China.Based on prediction fields from the flexible Global Ocean-Atmosphere-Land System Model finite volume version 2(FGOALS-f2),we optimized the loss function of U-Net,trained with different hyperparameters,and selected the optimum model.U-Net model can extract multi-scale feature information and preserve spatial information,making it suitable for processing meteorological data.With this endto-end model,the precipitation distribution can be obtained directly without using the traditional method of data dimensionality reduction(e.g.,Empirical Orthogonal Function),which could maximize the retention of spatio-temporal information of the input data.Optimization of the loss function enhances the prediction results and mitigates model overfitting.The independent prediction shows a significant skill improvement measured by the anomalous correlation coefficient score.The skill has an average value of 0.679 in China(0°–63°N,73°–133°E)and 0.691 in the region of the Chinese mainland,which significantly improves the dynamical prediction skill by 1357%and 4836%.This study suggests that the deep learning(U-Net)-based seasonal prediction bias correction method is a promising approach for improving rainfall prediction of the dynamical model.
文摘Deep learning-based object detection has revolutionized various fields,including agriculture.This paper presents a systematic review based on the PRISMA 2020 approach for object detection techniques in agriculture by exploring the evolution of different methods and applications over the past three years,highlighting the shift from conventional computer vision to deep learning-based methodologies owing to their enhanced efficacy in real time.The review emphasizes the integration of advanced models,such as You Only Look Once(YOLO)v9,v10,EfficientDet,Transformer-based models,and hybrid frameworks that improve the precision,accuracy,and scalability for crop monitoring and disease detection.The review also highlights benchmark datasets and evaluation metrics.It addresses limitations,like domain adaptation challenges,dataset heterogeneity,and occlusion,while offering insights into prospective research avenues,such as multimodal learning,explainable AI,and federated learning.Furthermore,the main aim of this paper is to serve as a thorough resource guide for scientists,researchers,and stakeholders for implementing deep learning-based object detection methods for the development of intelligent,robust,and sustainable agricultural systems.
基金supported by the National Natural Science Foundation of China(42272200)The Science and Technology Major Project of China National Petroleum Corporation(2023ZZ18-03)+1 种基金The Science and Technology Major Project of Changqing Oilfeld(2023DZZ01)The Technology project of Huaneng Group Headquarters(Medium-deep Low-Rank Coalbed Methane Resource Potential Evaluation and Key Development Technologies of Zhalainuoer Coalfeld,HNKJ23-H51).
文摘Currently,regression prediction methods based on logging data is one of the main methods for analyzing gas content of coal seams.However,the complexity of logging parameters for deep coal seams and the scarcity of measured gas content data signifcantly afects the accuracy and generalizability of data regression models.Accurately predicting the gas content of coal seams under small-sample condition become a difcult point in deep coalbed methane(CBM)exploration.The ModelAgnostic Meta-Learning(MAML)and Support Vector Regression(SVR)algorithms are among the few suitable for smallsample learning,exhibiting strong adaptability under limited sample conditions.In this study,logging parameters are used as input variables to construct MAML and SVR models,and their performance in predicting gas content of deep coal seams across diferent regions and layers is compared.The results demonstrate that the MAML algorithm efectively addresses the complex relationships between gas content of deep coal seam and logging parameters.The prediction errors for test dataset and new samples are merely 3.61%and 4.52%respectively,indicating exceptional adaptability,robust generalization capability,and stable model performance.In contrast,the dependency of SVR model on input parameters restricts its accuracy and generalizability in predicting gas content in deep coal seams with varying geological conditions.Although achieving a test dataset error of 4.71%,the SVR model demonstrates substantially degraded performance when applied to novel samples,with prediction errors escalating to 12.46%.Therefore,the MAML model is selected to predict gas content in the unknown areas of the Baijiahai region.The prediction results reveal that the gas content of coal seams in the Xishanyao formation(J2x)ranges from 1.32 m^(3)/t to 16.11 m^(3)/t,while that in the Badaowan Formation(J1b)varies between 1.73 m^(3)/t and 11.27 m^(3)/t.Notably,the gas enrichment areas are predominantly distributed in well blocks adjacent to fault systems,such as wells C31 and BJ8,etc.,which align with the favorable geological conditions for deep CBM accumulation in the Baijiahai region.These spatial distribution patterns not only corroborate existing geological insights but also further validate the reliability of the MAML model in predicting gas content within deep coal seams.
基金the support of Research Program of Fine Exploration and Surrounding Rock Classification Technology for Deep Buried Long Tunnels Driven by Horizontal Directional Drilling and Magnetotelluric Methods Based on Deep Learning under Grant E202408010the Sichuan Science and Technology Program under Grant 2024NSFSC1984 and Grant 2024NSFSC1990。
文摘Porosity is an important attribute for evaluating the petrophysical properties of reservoirs, and has guiding significance for the exploration and development of oil and gas. The seismic inversion is a key method for comprehensively obtaining the porosity. Deep learning methods provide an intelligent approach to suppress the ambiguity of the conventional inversion method. However, under the trace-bytrace inversion strategy, there is a lack of constraints from geological structural information, resulting in poor lateral continuity of prediction results. In addition, the heterogeneity and the sedimentary variability of subsurface media also lead to uncertainty in intelligent prediction. To achieve fine prediction of porosity, we consider the lateral continuity and variability and propose an improved structural modeling deep learning porosity prediction method. First, we combine well data, waveform attributes, and structural information as constraints to model geophysical parameters, constructing a high-quality training dataset with sedimentary facies-controlled significance. Subsequently, we introduce a gated axial attention mechanism to enhance the features of dataset and design a bidirectional closed-loop network system constrained by inversion and forward processes. The constraint coefficient is adaptively adjusted by the petrophysical information contained between the porosity and impedance in the study area. We demonstrate the effectiveness of the adaptive coefficient through numerical experiments.Finally, we compare the performance differences between the proposed method and conventional deep learning methods using data from two study areas. The proposed method achieves better consistency with the logging porosity, demonstrating the superiority of the proposed method.
基金the Natural Science Foundation of Shandong Province of China(Nos.ZR2022QD080 , ZR2025MS575)the National Natural Science Foundation of China(Nos.W25322063,42250410333,52250410357)+1 种基金the Fundamental Research Funds for the Central Universities,CHD(No.300102263103)the Young Talent Fund of Association for Science and Technology in Shaanxi,China(No.20230703)。
文摘The accurate identification of microporosity is crucial for the characterization of hydrocarbon reservoir permeability and production.Scanning electron microscopy(SEM)is among the limited number of methods available to directly observe the microscopic structure of the hydrocarbon reservoir rocks.Nevertheless,precise segmentation of microscopic pores at different depths in SEM images remains an unsolved challenge,known as the‘depth-related resolution loss'problem.Therefore,in this study,a 3D reconstruction technique for regions of interest(ROI)was developed for in-depth pixel analysis and differentiation among various depths of SEM images.The processed SEM images,together with the processing outcomes of this technique,were used as the input database to train a stochastic depth with multi-channel residual pathways(SdstMcrp)deep learning model programmed in Python to develop a tool for segmenting the microscopic pore spaces in SEM images obtained from the Beibuwan Basin.The more accurate segmentation helped to detect an average of 1.2 times more microporosity in SEM images,accounting for about 1.6 times more pixels and 1.2 times more pore surface area.Finally,the impact of the accurate segmentation on the calculation of permeability,a significant reservoir production property,was investigated using fractal geometry models and sensitivity analysis.The results showed that the obtained permeability values would vary by a factor of 6,which represents a considerable difference.These findings demonstrate that the proposed models can effectively identify features across a wide range of grayscale values in SEM images.
