最新刊期
- 摘要：To address the problems of color distortion, edge degradation, and inconsistency in polarization parameters during the full-resolution reconstruction of color division-of-focal-plane (DoFP) polarization images, a polarization image super-resolution reconstruction network based on contextual dual autocorrelation attention is proposed. The network consists of a dual-branch feature extraction module, a contextual dual-autocorrelation attention Transformer, and a polarization feature reconstruction module. Specifically, the dual-branch feature extraction module is employed to enhance shallow structural and texture features. The contextual dual-autocorrelation attention Transformer establishes spatial structural relationships and channel correlations among different polarization orientations through contextual gated fusion, position-aware autocorrelation attention, and local depth-wise autocorrelation attention mechanisms. The polarization feature reconstruction module restores full-resolution color polarization images through feature refinement and pixel rearrangement. Experimental results show that, on the SunPol-Real, OL, and Monno datasets, the proposed network achieves average peak signal-to-noise ratio values of 38.16, 39.39, and 40.02 dB, respectively, and average structural similarity index values of 0.967 6, 0.965 8, and 0.981 5, respectively, across the reconstructed images at the 0°, 45°, 90°, and 135° polarization orientations. These results demonstrate that the proposed network can effectively alleviate edge blurring and texture loss while improving the consistency of polarization information. This study provides an effective technical approach for the super-resolution reconstruction of color DoFP polarization images.关键词：Polarization image;super-resolution;Self-correlation attention;Position awareness;Local depth4|2|0更新时间：2026-07-21
- 摘要：To address the low electron utilization efficiency in photofermentative hydrogen production, sulfidated amorphous zero-valent iron (SAZVI) with different interfacial structures was synthesized by regulating sulfur precursors and subsequently assembled with Rhodobacter capsulatus SB1003 to construct a biohybrid system. Na2S, Na2S2O4, and K2S6 were employed to sulfurize amorphous zero-valent iron, enabling the construction of efficient electron donors through modulation of the Fe-S interfacial structure. Material characterization, electrochemical analyses, and response surface methodology were combined to evaluate the electron-donating capability and hydrogen-production performance of the resulting systems. The results demonstrated that sulfur precursors markedly influenced the Fe-S interfacial configuration and electron-transfer capability. Among them, the Na2S-derived SAZVI exhibited a uniform FeS conductive layer while retaining a high proportion of Fe0 active sites. XPS analysis revealed that the enrichment of Fe0, Fe2+, and S2− species optimized the interfacial electronic structure, resulting in a more negative corrosion potential (-1.27 V). Accordingly, the biohybrid system constructed with Na2S-derived SAZVI achieved the best hydrogen-production performance, with a cumulative hydrogen yield of 2865.43 mL·L-1, a maximum hydrogen production rate of 60.24 mL·(L·h)-1, and a shortened lag phase of 2.78 h. Response surface optimization identified the optimal operating conditions as pH 7.53, glucose concentration of 11 g·L-1, and lamp power of 42 W, under which hydrogen production was enhanced by 52.34% compared with the control. This study elucidates the critical role of sulfur-precursor-regulated Fe-S interfacial structures in facilitating electron transfer. It offers a reference for developing highly efficient biohybrid systems for photofermentative hydrogen production.关键词：photofermentative hydrogen production;biohybrid system;sulfur precursor;interface electron transfer6|0|0更新时间：2026-07-20
- 摘要：To address the limitation of traditional methods in capturing complex nonlinear relationships and multi-factor interactions in small-sample experimental data from supercritical water co-gasification of coal and sludge, this study developed a machine learning-based framework for gasification product prediction and model interpretation. Leave-one-out cross-validation was employed to evaluate random forest (RF), extreme gradient boosting (XGBoost), gradient boosting regression (GBR), and support vector regression (SVR) models for predicting H₂, CH₄, CO, and CO₂ yields, while grid search and Optuna were used for hyperparameter optimization. The results showed that XGBoost exhibited the best overall predictive performance, with an average R² of 0.828, MAE of 0.614, and RMSE of 0.839 after Optuna optimization. SHAP analysis identified reaction temperature as the most influential factor, contributing approximately 45% to the model output. Partial dependence plots further revealed the nonlinear effects of key experimental parameters on gasification product distribution and their interaction characteristics. The proposed framework is suitable for small-sample SCWG datasets and provides a data-driven approach for parameter optimization and product distribution regulation in coal and sludge co-gasification.关键词：coal and sludge;supercritical water co-gasification;interpretable machine learning;leave-one-out cross-validation;Extreme Gradient Boosting2|0|0更新时间：2026-07-20
- 摘要：As a cutting-edge field representing the deep integration of artificial intelligence (AI) and robotics, embodied AI robots have become a key driver for developing new forms of productive capacity and advancing the ‘AI+’ initiative, and serve as a crucial vehicle for achieving general intelligence. This paper systematically reviews the modern history of robotics and traces the technological evolution of the three major schools of thought in artificial intelligence: symbolism, behaviorism and connectionism. It summarizes the definitions and core concepts of embodied AI proposed by leading scholars both domestically and internationally, and proposes the DREAM technical framework for embodied intelligent robots, which encompasses five core elements: Digital Twin, Reality, Engagement, Awareness, and Motion. It also summarizes its three core characteristics: physical carrier dependency, twin mapping, and self-growth. It provides an in-depth analysis of the current mainstream technical approaches in embodied AI, including the integrated end-to-end model approach, the hierarchical model approach, and the world model approach. It categorizes and describes the carrier forms of embodied AI robots, covering fixed-base, wheeled, legged, flying, and special types such as underwater, space-based, and soft robots. It summarizes the classification criteria for embodied AI robots. By examining their current applications in industrial manufacturing, domestic services and specialized environments, alongside the challenges they face, the paper outlines future development trends for embodied AI robots, thereby providing a reference for future research in this field.关键词：embodied AI;robot;artificial intelligence;large model;DREAM architecture24|4|0更新时间：2026-07-15
- 摘要：To address the challenges of diverse target characteristics and visual occlusion in grasping target shape reconstruction for robots in unstructured home service scenarios, this paper proposes a shape reconstruction method for grasping targets driven by visual-tactile fusion. First, a block feature transformation network is designed to transform the single-view input point cloud and the tactile array into patch feature proxies. Second, the Transformer architecture is utilized to achieve visual-tactile cross-modal feature fusion and completion, and a geometric perception module is designed to predict and model local geometric relationships. Finally, combined with a shape query module to generate dynamic condition-based query embeddings, a decoder folding reconstruction module is designed to complete the point cloud based on patch feature predictions, thereby outputting the complete target shape point cloud. The experimental results demonstrate that the proposed method significantly improves the grasp success rate of robotic grasping tasks. Compared with the current best baseline model, the proposed approach achieves a 12.5% improvement in shape reconstruction performance and attains an average accuracy of up to 80.3% on unseen objects. Furthermore, the integration of the tactile modality reduces the Chamfer distance by 1.532. Through cross-modal data complementarity, the proposed method achieves more complete and robust point cloud shape reconstruction for grasping targets, demonstrating strong robustness and generalization capabilities when handling complex and variable objects, thereby providing a solution for shape reconstruction of grasping targets for robots in unstructured home service scenarios.关键词：shape reconstruction;visual point cloud;tactile array;visual-tactile fusion12|1|0更新时间：2026-07-07
- 摘要：To meet the demand for fast and accurate eddy-current loss assessment in laminated cores for energy-efficient transformer design, this paper proposes a fast and high-accuracy multiscale computational method based on eigenfunction-informed prior knowledge. The method addresses the high computational cost of refined modeling in the conventional finite element method (FEM), as well as the relatively large errors in local eddy-current field and loss calculations caused by the lack of physical relevance in the enrichment functions used in existing multiscale finite element methods (MSFEMs). First, according to the hyperbolic-cosine distribution characteristic satisfied by the magnetic vector potential in the laminated region, an eigenfunction capable of representing the field variation inside the laminations is extracted and introduced into the MSFEM framework as prior information. Then, by combining the level-set function with the partition of unity method, an enrichment function is constructed to characterize the nonlinear field variation near material interfaces and edge skin-effect regions, enabling a low-degree-of-freedom solution of the local eddy-current field in the laminated core. On this basis, the local eddy-current loss density and the eddy-current loss per unit length are calculated from the eddy-current density distribution. Finally, the proposed method is validated using a two-dimensional coil-lamination model that represents the local electromagnetic behavior of a transformer laminated core. The results show that, using the fine-mesh FEM result as the reference, the proposed method reduces the number of mesh nodes and degrees of freedom by 97.99% and 97.41%, respectively, while reducing the computational time from 628 s to 51 s, corresponding to a reduction of 91.88%. Compared with the original MSFEM under the same coarse-mesh condition, the proposed method reduces the average relative error of the magnetic vector potential by 98.80% and the relative error of the eddy-current loss per unit length by 96.77%. These results demonstrate that the proposed method reduces the computational scale and time while improving the accuracy of local field calculation and eddy-current loss assessment under an extremely low mesh density, providing an efficient computational tool for eddy-current loss assessment and refined electromagnetic analysis of transformer laminated cores.关键词：transformer;laminated core;eddy-current loss;multiscale finite element method;enrichment function42|4|0更新时间：2026-06-23
- 摘要：To meet the requirements of humanoid robot joints for high load capacity, high precision, and compact integration, this study investigates the effects of machining errors on the meshing characteristics of a novel inverted recirculating planetary roller screw mechanism (IRPRSM). Unified three-body helical-surface equations and a spatial meshing model are established to determine the actual meshing position, meshing radius, deflection angle, and axial clearance. Results show that eccentricity errors cause periodic fluctuations in the meshing radius and axial clearance, with the frequency increasing with the number of screw starts. Within ±0.03 mm, the difference between the axial clearances induced by screw and roller pitch-diameter errors is less than 2 μm. When a flank-angle error occurs separately in the screw or roller, the axial clearance peaks at 24.58 μm at 0° and decreases to approximately 10–12 μm at ±3°. When the screw and roller flank-angle errors vary equally in the same direction, the axial clearance increases only from 22.87 μm to 26.11 μm. Thread-start indexing errors do not change the axial contact position but redistribute the clearances between the upper and lower helical flanks. The findings provide a theoretical basis for the manufacturing and precision assembly of IRPRSMs in humanoid robot joint actuators and support the diversified development of related joint technologies in China.关键词：humanoid robot;joint;planetary roller screw mechanism;meshing characteristics;machining error8|1|0更新时间：2026-06-23
- 摘要：As China’s new-type power system accelerates its development, distributed renewable energy sources and new types of loads are being integrated into the power grid on a large scale. As a result, power equipment is subjected to complex stresses such as broadband harmonic disturbances and high load fluctuations, which impose higher requirements on operational reliability and maintenance technologies. To address the challenges in intelligent operation and maintenance of power equipment—such as the difficulty in fusing multi-source heterogeneous data, insufficient transfer of domain knowledge, and the lack of trustworthiness in complex fault reasoning—this paper systematically investigates the key technologies and development pathways of large model-empowered intelligent operation and maintenance for power equipment. It outlines the key technologies and development paths of large models in this domain and constructs a large model framework for intelligent operation and maintenance of power equipment, comprising modules such as data, perception, cognition, and decision-making. By drawing on typical scenarios such as globally data-driven operation, operation optimization coordination, and intelligent operation and inspection support, the paper elucidates the mechanisms through which large models function in equipment state perception, task decomposition, tool invocation, and human–machine collaborative decision-making. The study further analyzes current application challenges from four perspectives: data quality and security, model architecture and physical consistency, computational resource deployment, and multi-scenario adaptation. Finally, it proposes development pathways including private deployment, multimodal semantic alignment, neural-symbolic integration, cloud–edge–device collaboration, and the combination of large model hubs with small model plug-ins, thereby providing a reference for the trustworthy construction and engineering application of large models for intelligent operation and maintenance of power equipment关键词：large language model;power equipment;intelligent operation and maintenance;multi-source heterogeneous data;human-machine collaborative decision-making;multimodal semantic alignment4|0|0更新时间：2026-06-23
- 摘要：Cooperative perception allows vehicles to share information with other agents, effectively reducing perception blind spots for single-vehicle intelligence in complex, long-tail scenarios. However, real-world deployment faces major bottlenecks, primarily due to limited wireless bandwidth and multi-source heterogeneous data. This article provides a comprehensive review of communication-efficient transmission methods for cooperative perception in autonomous driving, alongside open-source datasets and future trends. To reduce communication overhead, the review begins by examining information filtering strategies within the perception process. It then explores cooperative relationship modeling and feature alignment mechanisms among heterogeneous agents. Furthermore, this article summarizes feature compression techniques designed to tackle strict data transmission constraints. Additionally, the review categorizes large-scale open-source datasets in this field and analyzes current bottlenecks, particularly in scene cognition and dynamic network security. Ultimately, this article aims to provide a theoretical reference for building more robust and secure vehicle-infrastructure cooperative networks.关键词：Autonomous Driving;Cooperative Perception;Intelligent Connected Vehicles;Multimodal Information Fusion;Communication Efficiency9|1|0更新时间：2026-06-22
- 摘要：To address the problems of sparse LiDAR point clouds, ambiguous features, and mismatches of keypoints caused by low overlap in intelligent driving scenarios, this paper proposes an adaptive multi-scale 3D point cloud registration network for intelligent driving. First, the network employs an adaptive multi-scale feature extraction module to obtain multi-level features of sampled points, combined with an adaptive neighborhood stabilizer for precise keypoint localization, thereby resolving matching difficulties caused by feature ambiguity. Second, by integrating multi-scale feature matching consistency with geometric consistency verification, the inlier confidence of point pairs to be registered is improved. Finally, based on this confidence, a coarse-to-fine iterative optimization strategy is adopted to achieve accurate 3D point cloud registration. Test results on the KITTI dataset demonstrate that the network achieves a relative translation error as low as 0.0618 m and a relative rotation error as low as 0.21°, with a registration recall of up to 99.8%. Results on the nuScenes dataset indicate that with inter-frame distances of 10 m, 20 m, and 30 m, the registration recall reaches 90.1%, 83.3%, and 78.8%, respectively. These findings validate that the proposed network effectively suppresses mismatches in complex road scenarios, achieving high-precision and highly robust large-scale 3D point cloud registration. This research meets the practical application requirements of dynamic and complex traffic conditions, providing reliable technical support for precise localization and environmental perception in autonomous driving scenarios.关键词：intelligent driving;three-dimensional point cloud registration;adaptive;multi-scale features6|2|0更新时间：2026-06-19
- 摘要：To investigate the dynamic wetting behavior of droplets during anti-gravity migration on an inclined aluminum plate surface under 35 kHz ultrasonic excitation, the effects of droplet volume, ultrasonic vibration velocity, and surface inclination angle on droplet diameter spreading ratio and contact angle hysteresis were quantitatively analyzed. The evolution of droplet wetting behavior was revealed, and the synergistic relationship between microscopic wetting properties and macroscopic dynamic behavior was established. Furthermore, based on the force competition mechanism, a comprehensive dimensionless number including the acoustic Weber number and the Bond number was constructed, and an empirical relationship for predicting the droplet dynamic response was established. The results show that ultrasonic excitation enhances droplet wetting ability and drives droplets to migrate against gravity on inclined surfaces at angles of 30-90°, with a maximum displacement of approximately 5.04 mm. The droplet diameter spreading ratio exhibits a nonlinear evolution of "increase-peak-retraction-steady" under ultrasonic, with the maximum values ranging from 1.16 to 1.51, primarily positively modulated by vibration velocity. Ultrasonic vibration effectively weakens gravity-induced wetting asymmetry, driving the contact angle hysteresis from a gravity-dominated negative state to an ultrasonic-dominated positive state, and the dimensionless number corresponding to the critical value of this transformation is 18.79. This study provides theoretical and experimental basis for the control of droplet transport and surface wetting based on the ultrasonic field.关键词：ultrasonic wave;inclined surface;droplets;anti-gravity migration;wetting properties5|1|0更新时间：2026-06-16
- 摘要：To address the issues of harmonic disturbance in flux observers and poor robustness against electrical parameter mismatches and load variations in position sensorless drive of permanent magnet synchronous motor (PMSM), a position sensorless drive strategy based on a novel
function-based equivalent sliding mode control ( -ESMC) is proposed. Firstly, to achieve finite time convergence of system state without requiring knowledge of the disturbance or its derivatives, a novel -ESMC method is designed by combining the equivalent control theory with a barrier function. The -ESMC can real time observe disturbance and its convergence is proved based on Lyapunov stability theory. Then, the proposed -ESMC is adopted to design the adaptive hybrid active flux observer (HAFO) and adaptive speed observer respectively to improve the observation performance of rotor flux and speed, as well as the robustness to disturbances such as electrical parameter mismatches and load. The experimental results show that compared to the classical HAFO method, the proposed strategy has improved the dynamic performance of the motor and its robustness against electrical parameter mismatches and load variations. In wide-speed range variable-speed test, the maximum speed observation error is reduced by about 34 rpm, the maximum position observation error is reduced by about 0.4 rad. In the 1.5 N·m step load test, the motor can recover to 500 rpm operation within approximately 1.93 s. In addition, it has demonstrated excellent robustness in the stator resistance mismatches test of 50% and 200%. 关键词：permanent magnet synchronous motor;position sensorless drive;hybrid active flux observer; function-based equivalent sliding mode control8|2|0更新时间：2026-06-16 - 摘要：To clarify the influence mechanism of windage loss and flow characteristics in shaft-type rotor and stator gap under the condition of wider gap-radius ratio, systematically analyzing the influence rules of key parameters, such as gap-radius ratio, rotor radius, and rotor rotational speed on the windage loss coefficient based on numerical methods. That reveals the flow mechanism with the range of gap-radius ratio less than 1.3, proposes a prediction method for the optimal gap-radius ratio, and develops a windage loss coefficient model for wider range of gap-radius ratio. The results show that, the influence mechanism of the gap-radius ratio on the windage loss coefficient mainly includes wall shear effects and dissipative effects inside the fluid. At smaller gap-radius ratio, the shear effect dominates, while at larger gap-radius ratio, the dissipative effects inside the fluid become the primary factor. The windage loss coefficient first decreases and then increases as the gap-radius ratio increases, with an optimal gap-radius ratio which minimizes the windage loss coefficient. As the rotor radius increases, the optimal gap-radius ratio gradually decreases. When the rotor radius increases from 20 mm to 60 mm, the optimal gap radius ratio decreases from 0.425 to 0.167. Based on the present numerical results, a method for predicting the optimal gap-radius ratio for shaft-type rotor and stator gap and a model for the windage loss coefficient in wide gap-radius ratio were proposed, with maximum prediction deviations of 5.58% and 11.81%, respectively. The research results provide technical support for the study of windage loss and gap structure optimization in shaft-type rotor and stator gap with wider gap-radius ratio.关键词：shaft-type rotor and stator gap;windage loss;Taylor vortex;gap-radius ratio24|5|0更新时间：2026-06-16
- 摘要：To solve the problems of tube bundle freezing and uneven heat transfer in a direct air-cooling island under winter low-temperature and low-load conditions and to improve its operational reliability in severe cold environments, an adjustable louver structure installed on the side heat exchange plane of the air-cooling island is proposed. By adjusting the blade angle, the airflow entering the tube bundles can be actively controlled to meet the heat dissipation requirements under different seasons and operating conditions. Based on the overall geometric model of the air-cooling island and a local single air-cooled unit model, numerical simulations were carried out using Fluent to analyze the flow and heat transfer performance of the louvers at different opening angles, and the effects of louver regulation on winter anti-freezing performance and heat transfer capacity under conventional conditions were systematically investigated.The results show that under an extreme winter condition with an ambient temperature of −20 ℃ and a wind speed of 20 m/s, closing the louvers reduces the total heat transfer capacity of the direct air-cooling island by 83.6 %, effectively decreasing the convective heat transfer intensity on the outer surface of the tube bundles and significantly suppressing freezing risk, thereby maintaining safer operation in severe cold environments. Under a conventional condition with an ambient temperature of 29 ℃ and a wind speed of 4 m/s, adjusting the blade angle to 60 ° limits the heat transfer loss of a single air-cooled unit to 1.8 %, with only a minor influence on the normal heat dissipation performance of the system. The proposed louver structure is easy to regulate in operation and highly adaptable, providing a flexible and effective technical solution for winter anti-freezing operation of direct air-cooling islands in cold regions.关键词：direct air-cooling island;louver;anti-freezing;numerical simulation12|4|0更新时间：2026-06-16
- 摘要：To meet the pressing need for high-performance, compact neon gas high-speed centrifugal compressors in cryogenic reverse-Brayton refrigeration system, this study presents the design and performance optimization of neon gas compressor. Three-dimensional viscous-flow CFD coupled with entropy-generation analysis is employed to simulate the internal flow and quantify losses, and the numerical predictions are validated against experimental data. Based on this, the meridional profile and blade geometry were optimized using a genetic algorithm, and the effects of blade leading-edge inclination angle and tip clearance were analyzed. The results show that the compressor model predictions agree well with the experimental data, with maximum deviations of 1.3% and 2.4% for the pressure ratio and isentropic efficiency, respectively. The compressor's low efficiency is primarily attributed to the tip-clearance leakage vortex and the secondary vortices arising from its interaction with the mainstream. Entropy-generation analysis reveals that the shroud-side region accounts for 55.1 % of the total entropy generation within the impeller. The optimized blade geometry improves efficiency by shortening the flow path and reducing the pressure differential across the blade tip. Ultimately, the isentropic efficiency is raised to 75.4 %, an improvement of 4.9 percentage points over the initial design.关键词：neon gas compressor;high-speed centrifugal compressor;reverse-Brayton refrigeration system;performance optimization;entropy generation2|0|0更新时间：2026-06-16
- 摘要：To address the stringent demands of high-rate fast charging and dynamic driving conditions on heat transfer and dynamic temperature control in electric vehicle thermal management systems, this study comparatively investigates the operational energy efficiency and thermodynamic characteristics of direct cooling and liquid cooling architectures for power batteries. First, a test bench for an integrated thermal management system with multi-mode switching capabilities was developed. Second, a system operation and exergy analysis model was established based on the first and second laws of thermodynamics. Finally, under the dynamic driving conditions of the China light-duty vehicle test cycle-passenger and multi-rate (0.5C–3C) fast charging scenarios, the temperature control response, energy efficiency, and component irreversible exergy losses of the two systems were quantitatively evaluated. The results indicate that: Under dynamic driving conditions, the liquid cooling system exhibits superior cabin supply air comfort and system stability due to its large heat capacity. Conversely, the direct cooling system achieves higher steady-state energy efficiency, with its coefficient of performance stabilizing above 2.5 in the middle and late stages of operation. During high-rate fast charging at 1C and above, the direct cooling system eliminates secondary heat transfer resistance, resulting in a significantly faster cooling response than the liquid cooling system. However, affected by the refrigerant phase-change flow pattern, its maximum steady-state temperature difference within the battery module reaches 7.8 ℃, indicating poorer local temperature uniformity compared to the 4.7 ℃ of the liquid cooling system. The direct cooling architecture effectively reduces irreversible heat transfer losses. Its battery cold plate exergy efficiency reaches 47%, which is significantly higher than the 29% of the battery chiller in the liquid cooling loop. Furthermore, during long-term dynamic operation, the maximum transient total exergy efficiency of the direct cooling system can reach 20.0%. This study reveals the thermodynamic evolution characteristics of different cooling architectures under heavy load and dynamic conditions, providing a theoretical basis for the engineering application and optimization of next-generation EV composite thermal management systems.关键词：electric vehicle;integrated thermal management;direct cooling and liquid cooling;thermodynamic property3|0|0更新时间：2026-06-16
- 摘要：To enhance the safety performance of mining new energy vehicles and further improve the estimation accuracy of lithium battery state of charge (SOC), this study proposes a cloud–edge collaborative SOC estimation framework and algorithm integrating deep learning for mining scenarios. At the edge end, parameters are collected through the battery management system, and local SOC estimation is performed using the ampere-hour integration method. In the cloud, a hybrid neural network model combining a bidirectional long short-term memory network and a multi-head attention mechanism is constructed, and the dung beetle optimization algorithm is introduced to optimize the hyperparameter combination of the model. Finally, adaptive unscented Kalman filtering is used to fuse cloud and edge data for SOC estimation correction. The effectiveness and performance of the model are evaluated under US06, FUDS, and DST operating conditions. The results show that the cloud-based model performs well and can effectively estimate SOC. Under FUDS conditions, the mean absolute error is 0.36% and the root mean square error is 0.50%; under DST conditions, the mean absolute error is 0.40% and the root mean square error is 0.57%. The cloud–edge fusion model further achieves SOC correction. Compared with the cloud-based algorithm, the mean absolute error and root mean square error are reduced by 28.5% and 37.9%, respectively, under FUDS conditions, and by 26.9% and 39.4%, respectively, under DST conditions. The model demonstrates a certain degree of generalization ability and robustness.The proposed cloud-edge framework can provide theoretical reference for the engineering application of lithium battery SOC estimation in mining scenarios, while the cloud–edge fusion model offers a practical approach for improving SOC prediction accuracy of mining lithium batteries.关键词：lithium-ion batteries;state of charge;Ampere-hour integration method;bidirectional long short-term memory;adaptive unscented Kalman filter5|0|0更新时间：2026-06-11
- 摘要：To address the problems that unsupervised methods for 3D medical image registration suffer from limited accuracy, weakly supervised methods rely on high-quality segmentation labels, and labels generated by foundation segmentation models cannot directly and stably guide registration training, a weakly supervised mutual-learning method for 3D medical image registration guided by a foundation segmentation model is proposed. First, masks of major anatomical regions are extracted from different views using segment anything model, and intersection and union masks are constructed through mask fusion to represent structural prior information at different granularities. Second, a mutual-learning mechanism is introduced into the weakly supervised registration framework, where a confidence and position information module is incorporated to promote knowledge distillation and collaborative optimization between registration models under different mask supervision, thereby enabling effective utilization of complementary information and improving training stability. Finally, experimental results on three brain image datasets demonstrate that the proposed method can effectively exploit label information while maintaining stable training, and outperforms existing 3D image registration methods on multiple evaluation metrics, showing better registration accuracy and reliability.关键词：medical image registration;segment anything model;weakly-supervised method;muti-view;mutual learning33|4|0更新时间：2026-06-09
- 摘要：To alleviate the inhibitory effect of ammonium ions (NH4+) on photo-fermentative hydrogen production, this study employed the purple non-sulfur bacterium Rhodobacter capsulatus (R. capsulatus) strain SB1003 as the model organism. First, overlap extension polymerase chain reaction (SOE PCR) was used for targeted modification of the nitrogen fixation regulatory gene nifA. N-terminal deletion fragments of nifA1 and nifA2 were constructed separately, and through homologous recombination screening, the nifA1 deletion mutant ZX01, the nifA2 deletion mutant ZX02, and the double deletion mutant ZX03 were obtained. Subsequently, the above strains were inoculated into hydrogen-producing medium containing glucose as the carbon source and different concentrations of ammonium sulfate and sodium glutamate as mixed nitrogen sources, and were cultured under illuminated, anaerobic conditions for photo-fermentative hydrogen production. The hydrogen production performance and expression levels of relevant genes of the different mutants were systematically compared under ammonium-containing and ammonium-free conditions. The results showed that under the condition of 20% ammonium nitrogen mass fraction, the double mutant ZX03 achieved the highest cumulative hydrogen production of 801.80 ± 60.40 mL·L-1, which was 22.7% higher than that of the wild-type R. capsulatus SB1003. Kinetic fitting based on the Gompertz equation revealed that ZX01 had the shortest hydrogen production lag phase (7.99 ± 3.41 h), indicating that deletion of the N-terminus of nifA1 significantly accelerates the initiation of hydrogen production. Quantitative real-time polymerase chain reaction (qPCR) analysis further demonstrated that under high ammonium conditions (8 mmol·L-1 NH4+), the transcript levels of the nitrogen signaling protein-encoding genes glnB1 and glnB2 in ZX03 were up-regulated by approximately 220-fold relative to the wild-type, while the expression of the electron transport-related gene rnfB and the flavodoxin-encoding gene fldA was also markedly enhanced. These findings indicate that deletion of the N-terminus of nifA1 mainly improves the response rate of the strain to ammonium stress by shortening the lag phase, whereas deletion of the N-terminus of nifA2 primarily enhances ammonium tolerance by elevating the basal hydrogen production capacity. The double mutant ZX03 synergistically integrates the advantages of both modifications and maintains nitrogenase system activity at the transcriptional level by up-regulating the expression of nitrogen signaling and electron transport-related genes, thereby achieving optimal hydrogen production performance under high ammonium conditions. This study provides a clear molecular engineering strategy and a theoretical foundation for constructing ammonium-tolerant, high-efficiency photo-fermentative hydrogen-producing strains.关键词：photosynthetic bacteria;mutant;photo-fermentation;gene editing8|3|0更新时间：2026-06-09
- 摘要：Steer-by-wire (SBW) is a key technology for advanced intelligent driving, yet its tracking accuracy and control performance are hard to guarantee under multiple uncertainties and input constraints. To address this issue, a novel cooperative fault-tolerant control method integrates robust guaranteed performance control and feedforward regulation is proposed. First, an SBW simulation model considering system parameter perturbations, external disturbances, actuator faults, and input constraints is established. Then, multiple uncertainties and faults within a certain range are incorporated into a unified control design framework. By defining a quadratic performance index and solving the guaranteed performance control law via linear matrix inequalities (LMI), the tracking performance and robustness of the system are constrained, and closed-loop system stability is proven via Lyapunov theory. Finally, a segmented variable-curvature feedforward compensator is introduced to improve the initial response speed, forming a complete steering angle tracking controller. Simulations and real-vehicle tests show that the proposed method achieves accurate and stable steering angle tracking with fast dynamic response. The comprehensive simulation tracking error under typical steering conditions is reduced by about 50% compared with guaranteed performance control, 65% compared with PID control, and 63% compared with adaptive sliding mode control. The comprehensive experimental tracking error is reduced by about 17% compared with guaranteed performance control, 4 % compared with PID control. It significantly maintains efficient and stable operation of the SBW vehicle under varying scenarios and fault conditions.关键词：steer-by-wire;fault-tolerant control;guaranteed performance control;multiple uncertainties;input constraints;intelligent vehicle10|2|0更新时间：2026-06-03
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0