A detection model named dynamic kernel perception YOLO (DKP-YOLO) was constructed to address the problems of contextual information loss, multi-scale distress coexistence, and insufficient detail restoration in pavement distress detection from a UAV perspective. A lightweight crack context module was proposed to mitigate the variability of crack morphology and the loss of contextual information in complex scenes. This module enhanced feature discriminability through heterogeneous parallel depthwise separable convolutions and a dynamic feature fusion mechanism. In order to tackle the challenges of complex backgrounds and multi-scale distress, a pavement distress perception module was designed, which improved the multi-scale distress perception capability using parallel multi-scale convolutions and a dual-attention mechanism. A large-kernel perception feature fusion module was introduced to overcome the limitations of traditional convolutions in modeling long-range dependencies and responding to small targets. This module captured global contextual relationships by integrating extra-large receptive field convolution with channel-spatial attention. Finally, a lightweight dynamic upsampling operator was incorporated into the neck network to reduce detail loss and semantic ambiguity during upsampling. Experiments on the UAV-PDD2023 dataset showed that the proposed model significantly improved detection accuracy while effectively reducing model complexity. Compared with the baseline model, DKP-YOLO achieved a better balance between being lightweight and maintaining high detection performance, making it more suitable for UAV-based pavement distress detection.

To address the issue of inconsistent thermal stress between the upper and lower devices of the same bridge arm in converters under conditions of uneven aging or layout deviations, a novel thermal balancing method for upper and lower bridge arms based on dead-time shifting adjustment was proposed. By monitoring the on-state voltage of insulated gate bipolar transistor (IGBT) online, the real-time junction temperature of IGBTs was deduced based on the mapping relationship between the on-state voltage and junction temperature of IGBTs. Then, the temperature difference between the upper and lower devices of the same bridge arm was determined. When the thermal stress difference between the upper and lower bridge arms exceeded the threshold, the effective conduction time of the upper and lower bridge arms was adjusted by fine-tuning the position of the dead time, thereby actively regulating the power loss distribution of the upper and lower bridge arms. This achieved an increase in the loss of the bridge arm with lower temperature and a decrease in the loss of the bridge arm with higher temperature without affecting the total system loss, thus improving the thermal stress consistency of power devices in the converter and enhancing system reliability. Theoretical analysis and experimental results showed that the proposed method did not affect system efficiency and had no significant negative impact on output performance. Under typical operating conditions, the average temperature difference between the upper and lower devices was reduced by approximately 14% after adjustment, effectively improving the thermal stress consistency of the upper and lower bridge arms.

A heterogeneous smoke and flame visual detection method integrating visible and infrared images was proposed, in order to address the problems of low detection efficiency, poor accuracy, and consequent delays in fire disposal and misallocation of resources in the early stage of tunnel fires caused by lighting interference and structural environment. For infrared images, an OTSU threshold segmentation optimized by a grey wolf optimizer fused with differential evolution was applied, and the maximum entropy algorithm was combined to improve flame region segmentation accuracy. Considering the tunnel scene characteristics, YOLO11 was adapted, and the backbone network was optimized using a Swin-Transformer modified with a neural attention mechanism to enhance weak feature extraction. During feature fusion, a lightweight BiFPN was employed to adapt to multi-scale smoke and flame features, and an adaptive WIoU loss function was introduced to improve localization accuracy under fuzzy boundaries. At the decision layer, weighted fusion and conflict resolution were used to achieve dual-modal collaborative decision-making. Experimental results demonstrated that the proposed method achieved at least a 23.55% reduction in average localization error compared with single-modal detection. Compared with nine fusion algorithms such as evidence theory, the recall, F1, and mAP values of the proposed method reached 92.52%, 90.83%, and 93.47%, respectively, all exceeding those of the compared algorithms, while maintaining the smallest localization error fluctuation. The inference speed of the model increased by an average of 14.61%. The research results not only contribute to tunnel fire prevention and control but also effectively improve overall tunnel safety management and emergency response capability.

A dual-stage image deraining network based on rain streak mask suppression and non-local reconstruction collaboration was proposed to address severe rain streak noise interference and insufficient spatial global modeling capability of existing attention mechanisms in single-image deraining networks. In the first stage of the network, a rain streak mask attention mechanism was designed, in which rain streak masks were generated through morphological operations, to enhance the model’s ability to suppress rain streak interference by selectively masking rain-affected regions during feature extraction. In the second stage, a non-local attention mechanism was devised by employing a feature clustering-based non-local similarity measurement method to guide pixel rearrangement, which broke spatial constraints, thereby augmenting the long-range modeling capability of the sliding window attention mechanism and improving the deraining performance. Through progressive optimization based on the dual-stage “rain streak suppression-detail reconstruction” process, high-quality reconstruction of rain-free images was achieved. Experimental results on multiple public datasets demonstrate that the proposed network achieves significant improvements in both PSNR and SSIM metrics compared to other networks, effectively removing rain streaks while better preserving image details and producing high-quality restored results with natural-looking appearance and fine-grained texture representations.

Aiming at the degradation of underwater images due to light propagation attenuation, scattering, and dissolved suspended matter, a novel underwater image enhancement network integrating a spatial-wise refined feature Transformer (SRFT) module and a channel-wise feature attention enhanced reconstructed Transformer (CFART) module was proposed. Serialization processing and positional encoding on underwater image sequences were performed by the SRFT module and four-layered spatial self-attention mechanisms were applied to capture global degradation differences while establishing long-range feature dependencies. In the CFART module, features were projected via heterogeneous convolutional kernels and then fed into a multi-head self-attention module through channel fusion. And feature information was reconstructed using multilayer perceptron layers with residual connections. Experimental results showed that the proposed algorithm effectively improved issues such as color shift, blurriness, and low contrast issues in underwater images. In terms of objective evaluation metrics, the proposed algorithm outperformed similar methods with MSE of 253.558, PSNR of 25.421, Entropy of 7.488, and UIQM of 4.461. It also demonstrated significant advantages in SSIM and UCIQE tests with SSIM of 0.893 and UCIQE of 0.592. Subjective visual assessments further confirmed that the proposed algorithm provided excellent color correction and detail restoration capabilities for degraded underwater images. By optimizing the quality of underwater images, the proposed algorithm helps to enhance the precision and efficiency of seabed investigations and assessments.

A spatiotemporal joint lane-change intention prediction model based on Transformer and graph attention neural network (GATv2) was proposed in order to accurately predict lane-changing intention of surrounding vehicles and improve the safety and reliability of autonomous vehicles. Symmetric exponential average filtering was applied to smooth the raw NGSIM data. Then velocity and acceleration were recalculated by using the difference method, and outliers were removed. The data of the target vehicle and its surrounding vehicles were unified by establishing an eight-neighborhood for each vehicle, providing a foundation for model training and validation. Transformer multi-head attention was used to mine temporal dependency, while GATv2 was employed to quantify the interaction relationship between vehicles at each time step. Then a Transformer-GATv2 lane-changing prediction model based on spatiotemporal feature and multi-vehicle interaction was constructed. The experimental results showed that the proposed model significantly outperformed other baseline models in accuracy, recall, F₁-score, and other metrics. A prediction accuracy of 98.04% was achieved, which demonstrated superior lane-changing intention prediction performance and strong early prediction capability. The model was deployed on a six-degree-of-freedom driving simulator integrated with the CARLA co-simulation platform for model-in-the-loop testing, verifying the usability of the model in real-world scenario.

A small target detection algorithm (DLSRF-Net) for multi-scale complex scenarios from UAV viewpoint was proposed by reconstructing the YOLOv11, to address the insufficient feature extraction and poor detection performance of existing algorithms in small target detection under UAV viewpoint due to small target sizes, complex backgrounds, and multi-scale information in the scenarios. The adaptive depthwise separative receptive field attention convolution module (DWRFAConv) was proposed to improve the model’s ability to extract the receptive field features of small targets and reduce the model load. The multi-branch lightweight multi-scale linear attention mechanism was designed to enhance the model’s attention to small targets. The RSCDI module was designed as the upsampling layer and fully connected layer of the model to solve the problem of feature information loss, suppress the useless information, and improve the model’s detection accuracy. The model sizes were classified into two categories based on parameter count and computational complexity, and experimental validation was carried out on the VisDrone2021 dataset. The results showed that the proposed algorithm achieved the optimal performance under both model size categories, and the generalization ability of the proposed algorithm was verified on the DOTA and the SSDD datasets.

The task expression and planning methods of aircraft assembly robots were analyzed aiming at the problems of isolated operation, insufficient universality, and low degree of autonomy and intelligence of industrial robots in aircraft assembly. A task driven aircraft assembly robot processing system framework was proposed, and an integrated system for task management and planning of aircraft assembly robots was established. A hierarchical decomposition of common processing tasks in aircraft assembly sites was conducted, and a task expression method for aircraft assembly robots based on a hierarchical task network was established. A task planning process oriented towards process constraints was proposed based on the characteristics of processing tasks. A hierarchical replanning scheme for the processing system was designed based on the range of disturbance effects considering the possible external disturbances on site. The test results showed that the proposed method was used to effectively achieve task planning and replanning solution for assembly robots, improving the generality of aircraft assembly robots and the level of autonomy and intelligence in processing systems.

An optimized model KUNet based on Kolmogorov-Arnold network (KAN) and convolutional KAN (CKAN) was proposed to enhance the performance of the UNet model in order to address the limitation of the UNet model in complex feature extraction and generalization capability for medical image segmentation task. Traditional convolutional layer was replaced with CKAN, KAN feature enhancement module was introduced, and skip connection was optimized. Then the diversity and accuracy of feature extraction were improved while preserving structural information by incorporating an adaptive basis function learning mechanism. Comparative experiments were conducted against the UNet baseline model, nnUNet model and Swin-UNet model on four different multimodal datasets: LiTS, CORN, DRIVE and Lungs. Results showed that the average maximum absolute performance gap (MAPG) between the UNet baseline model and the KUNet model across the four datasets were 0.679 9 and 0.620 3 for Dice coefficient and IoU coefficient, respectively, and the KUNet model achieved average improvement metrics of 0.3213 and 0.2625 compared with the optimal or suboptimal model across the four datasets. The KUNet model was utilized to effectively extract more feature within short training cycle and improve the accuracy of image segmentation.

A compact and lightweight zero-knowledge proof algorithm named CZKP-1t was constructed in order to address the issue of insufficient positioning reliability in traditional global navigation satellite systems (GNSS) under atmospheric variation and intentional signal interference. A maritime position data-sharing scheme called CZKP-1t-MPS was proposed by integrating with blockchain technology. This scheme enables position data sharing between conventionally positioned vessels and dynamically positioned vessels equipped with high-precision sensors, effectively enhancing the overall positioning accuracy of the maritime vessel network. The designed position-sharing method severs the connection between the two sharing parties and disassociates the position data from its requester, thereby ensuring strong privacy protection during the sharing process. CZKP-1t-MPS reduces the computational overhead by approximately 78% in the data-sharing process compared with existing maritime position-sharing methods, which guarantees real-time performance in heterogeneous maritime environment.

A control strategy integrating braking force redistribution and path tracking was proposed to address the problem that instability and yawing were prone to occur when vehicles equipped with electromechanical brake (EMB) system experience braking failures during cornering. Gaussian perturbation and staged optimization were introduced to improve the algorithm in order to mitigate the deficiencies of the slime mould algorithm (SMA). The enhanced SMA was employed to optimize the weight matrix of the linear quadratic regulator (LQR). The improved LQR algorithm was utilized to compute the vehicle’s yaw moment upon detection of a single-wheel failure in the EMB system, followed by braking force redistribution to maintain vehicle stability. The pure pursuit algorithm was modified by shifting the tracking control point to enhance the response speed of the algorithm. An adaptive fuzzy control algorithm was incorporated to accommodate dynamic factors such as road conditions and vehicle speed, thus improving its adaptability. Path tracking was implemented to guide the vehicle along a predefined trajectory until a safe stop when a double-wheel failure was detected in the EMB system. The experimental results demonstrated that the maximum lateral deviation was reduced by 59.15% for single-wheel failure and by 41.95% for double-wheel failure compared with conventional methods. The proposed control strategy can more effectively ensure driving safety during cornering braking failure.

A new traffic flow prediction model named CFHD-Former was proposed in order to address the limitations of existing traffic flow prediction models that struggle to capture dynamic variations across different time slices and cannot adequately consider the heterogeneous characteristics of traffic volume distributions across regions. An adaptive high-frequency heterogeneity module and a progressive optimization mechanism were introduced to enhance its adaptability to traffic states under different time slices. A core flow node identification module was employed building on the captured temporal heterogeneity feature in order to partition the road network into core and non-core flow networks based on nodal traffic flow characteristics. Heterogeneous modeling of the two types of road network was implemented via a spatial encoder. A frequency-domain autocorrelation MAE loss function was incorporated during backpropagation in order to consider the dependencies among different time steps within the prediction sequence, thereby reducing multi-step prediction errors. The experimental results demonstrated that the MAE of the proposed CFHD-Former model was reduced by 1.70%, 4.58% and 4.44% on the PEMS04, PEMS08 and METR-LA datasets, respectively compared with the best-performing baseline model. Results verified the effectiveness of CFHD-Former in modeling the spatio-temporal heterogeneity of complex road networks and provided a new solution for urban traffic flow prediction.

A multi-constrained human-robot collaborative disassembly line balancing problem was proposed for U-shaped disassembly lines in order to address the issues that existing studies on human-robot collaborative disassembly lines neither simultaneously consider differences in human and robotic task time and task attribute constraint, nor incorporate robot procurement costs into the long-term costs of collaboration. An integer programming model for the U-shaped disassembly line was constructed, with the objectives of minimizing the number of workstations, the idle time balancing index, and the long-term cost. Constraints considering various problem characteristics, including human-robot task attributes, human-robot task time, and AND/OR precedence relations were incorporated. An improved hybrid clonal simulated annealing algorithm was proposed. Double-layer encoding and decoding were designed, along with mutation and crossover operations specifically considering the problem characteristics. Cloning operations were introduced to enhance the local search capability of the algorithm, and a two-stage annealing process was implemented to accelerate convergence speed. Gurobi software was applied to solve small and medium-scale problems, and the results were compared with those obtained by the algorithm to verify the correctness and effectiveness of the model and algorithm. The cost variations of different disassembly line modes with the estimated operational time of the disassembly line were calculated and compared. Results demonstrate that the proposed model possesses the advantage of agile disassembly line planning.

Direct numerical simulations were performed to investigate the interaction between a horseshoe vortex and a free surface at various Weber numbers. The incompressible Navier-Stokes equations were solved using the volume-of-fluid method coupled with the piecewise linear interface construction scheme. The continuum surface force model was incorporated to account for surface tension effects during interface deformation. The simulations covered a range of Weber numbers from 0.24 to 0.84. By quantitatively analyzing the evolution of total kinetic energy, dissipation rate, total helicity, and surface energy, the influence of Weber number on the vortex-interface interaction was elucidated. At low Weber numbers, surface tension dominated, maintaining the interface as a coherent ring-like structure and suppressing deformation and topological changes, while the vorticity field remained stable with negligible variations in helicity. Conversely, at high Weber numbers, inertial forces dominated over capillary effects, leading to significant interface deformation, breakup, and reconnection. This process resulted in vortex breakdown, accelerated kinetic energy decay, distinct peaks in energy dissipation, and a continuous decrease in helicity. These findings underscored the critical role of the Weber number in energy transfer and dissipation mechanisms, providing theoretical support for multiphase flow dynamics and guidelines for engineering applications such as the design of surface and trans-media vehicles.

The fluid dynamic behavior of halide perovskite precursor solutions during thin-film formation was systematically elucidated, and a unified theoretical framework for regulating solute transport, evaporation dynamics, and crystallization evolution was established. Rheological measurements, kinetic analysis, and numerical simulations of representative coating processes were combined to characterize velocity distributions and concentration migration during the stages of viscous flow, interfacial spreading, and evaporation-induced solidification. In addition, particle image velocimetry and interfacial tracking techniques enabled coordinated observation of flow-field structures, evaporation front positions, and crystal orientation. The results indicated that the solvent system, shear conditions, and interfacial properties were key variables governing solute distribution and crystallization pathways, and the appropriate control of evaporation gradients and interfacial energy could effectively suppress the coffee-ring effect. Overall, a systematic mechanistic framework spanning solution engineering, process control, interfacial regulation, and structural characterization was established, providing a predictable and designable pathway for achieving highly uniform, low-defect perovskite thin films and a theoretical basis for the controlled fabrication of solution-processed perovskite optoelectronic materials.

An integrated system for path planning and tracking control was developed to address the structural mismatch between the discrete path generated by the A^* algorithm and the continuous inputs required by multi-reference point model predictive control (M-MPC), and to improve the tracking accuracy and smoothness of path tracking control. The A^* reference path was smoothed and discretized at equal arc-length intervals, and the generated reference point sequence was introduced into the prediction horizon of the M-MPC controller. A heading alignment mechanism was designed to mitigate the problem of large initial heading deviations between the reference path and the differential-drive robot, thereby ensuring a seamless transition from path planning to tracking control. Experimental results demonstrated that the proposed system achieved high accuracy and smoothness. Compared with the direct combination system of the A^* algorithm and M-MPC, the peak and average values of the displacement error were reduced by 51.85% and 20.40%, respectively, while the cumulative control increment was decreased by 28.86%. Compared with the systems combining the smoothed A^* algorithm and a pure pursuit controller or a single reference point MPC, the peak and average values of the displacement error was decreased by at least 52.80% and 41.58%, respectively. The proposed system improved the tracking accuracy and smoothness, and enhanced the motion control performance of differential-drive robots in complex environments.

A novel parameter optimization method for model predictive control was proposed to address the problems of insufficient tracking accuracy and poor real-time performance of model predictive controllers in trajectory tracking of autonomous vehicles. An improved non-dominated sorting whale optimization algorithm (NSWOA) based on the improved Sinusoidal mapping and a Lévy flight strategy was proposed to solve the problem of relatively concentrated and out-of-bound optimal solutions obtained by the NSWOA. The parameter optimization problem of model predictive controllers was formulated as a multi-objective optimization problem. The predictive horizon, control horizon, and sampling time were used as optimization variables. The sum of squared lateral trajectory errors and total computation time were used as optimization objectives. The improved NSWOA was employed to solve the multi-objective optimization problem and obtain the Pareto optimal solution set. The optimal controller parameter combination was determined by using the expert scoring method, the continuous ordered weighted averaging operator method, the game theory-based combined weighting method, and the technique for order preference by similarity to an ideal solution. The tracking accuracy of the proposed method was improved by an average of 56.27%, and the computation time was reduced by an average of 21.54%. This method provides a new idea that balances high tracking accuracy and high real-time performance for the tuning strategy of model predictive control parameters.

Existing anomaly detection methods of time series data focus on extracting the temporal variation features, while the spatial dependency features between multiple variables are ignored. To address this problem, a detection method based on a spatiotemporal graph attention network was proposed. The original multivariate time series data were transformed into a time-series graph with spatiotemporal dependencies, and a spatiotemporal graph attention network was designed to separately extract the temporal variation features and spatial dependency features. The periodic patterns of fused spatiotemporal features were learned by a multilayer perceptron, and an anomaly detection was performed based on the anomaly scores between prediction values and observation values. Experimental results on public datasets showed that the proposed method significantly outperformed state-of-the-art baseline methods in terms of anomaly detection accuracy and robustness.

A dual-channel graph neural network method was proposed for user-level fraud detection tasks on E-commerce platforms to address the limitations of existing approaches that overemphasized global modeling of user behavior while insufficiently exploiting comment information. Multi-dimensional user behavior was modeled through the construction of two complementary graphs: an entity interaction graph and a comment semantic graph. The entity interaction graph was designed to capture global interaction patterns based on purchase and rating behaviors, while the comment semantic graph was built to model time-sensitive semantic relations between comments for characterizing fine-grained behavioral features. Parallel modeling of the dual graphs was performed using graph neural networks. Dynamic interaction optimization between dual-channel features was achieved through an attention mechanism, and higher-order node features containing multi-hop neighborhood information were generated. A comprehensive user-level behavior representation was produced by adaptively fusing different neighborhood ranges and feature spaces with a multi-head additive attention mechanism. Experimental evaluations were conducted on public datasets to validate the proposed method, and significant improvements were observed in multiple evaluation metrics compared to traditional approaches. Results show that the proposed method effectively enhances fraud detection performance at the user level.

A knowledge graph-guided 3D visual question answering method was proposed to capture the implicit common-sense semantic relationships between objects in the scene. By introducing external structured knowledge, the model was effectively enhanced in both semantic understanding and reasoning ability. Specifically, key semantic entities were extracted from the question text, and a knowledge graph-guided feature enhancement module was designed to obtain knowledge features using these key semantic entities. The knowledge features were fused with visual features extracted both from the question representation and from the 3D object detection network for answer prediction. Experimental results on the ScanQA dataset showed that the proposed method outperforms existing baseline models on metrics such as EM@1 and BLEU-4.