Surface defects are usually the early phenomenon of rail failure, which threatens the safety of railroad transportation critically, and the timely detection of surface defects helps to eliminate the potential risk of rail and reduce the chance of railroad safety accidents. The existing methods of detecting surface defects on rails suffer from a large performance degradation in the application of rails containing pollutions such as rust and oil. Therefore, this article proposes a multilevel, end-to-end fast rail surface defect detection method. First, rail extraction was performed based on the stability of the standard deviation of the edge pixels. Then, differential box-counting (DBC) and GrabCut algorithm are then combined for defect segmentation to boost the speed and accuracy of extracting complex defects. Finally, YOLO v2 is used to precisely locate and detect defects. The experimental results show that the proposed method performs well, with an average accuracy of 97.11%, an average recall of 96.10%, and an average frame rate of 0.0064 s. In addition, the proposed method offers a high robustness in the tests of different use cases.

Local failure of rail track commonly grow from surface defects. Hence, timely detection of surface defects helps to identify potential hazards on the track and reduce the occurrence of railroad accidents. Since track surface defects are scattered and diverse, and different service life leads to different types of defects, it is crucial to detect surface defects in real time, with efficiency, reliability and robustness. To this end, a pixel-level defects segmentation method is proposed. In this paper, features are tessellated together at the Channel level to form denser features, allowing additional information on surface defects textures to be propagated among high-resolution layers. Dropout is performed on the weak correlations learned during the convolution, so that the convolution blocks can share a uniform weight matrix, reducing computation redundancy and model complexity. Firstly, the track datasets of different ages are categorized into four sets, then, the samples are normalized to grey scale by pre-processing, and fed into the proposed network for training. An evaluation of the proposed model on defective samples was performed to demonstrate the performance of the method, with an Accuracy of 97.47%, a Loss of 0.0061 and an Average Frame Rate of 0.033s. The results of different networks tested on the same dataset show that the proposed model exhibits strong stability, adaptability and robustness. In addition, the proposed model is assessed on two different datasets with distinct challenges, with Mean Intersection over Union yielding 2.13% and 3.77% boosts respectively.

Addressing the challenge of surface defect detection in load-bearing rails within auto-motive assembly workshops, which operate in complex environments and under long-term service, this paper proposes an innovative detection framework based on an improved YOLOv5 network. This framework, designed specifically for the unique challenges presented by load-bearing rails, integrates advanced machine vision and deep learning technologies. Initially, a Multi-Scale Pyramid Pooling (MSPP) module, incorporating the concept of residual stacking, is introduced to effectively enhance the extraction of complex features; Subsequently, the coordinate attention mechanism is optimized, leading to the development of a novel Spatial Coordinate Attention Mechanism (DAM), focused on detecting small-sized defects; Thereafter, a Dual Sampling Transition Module (DSTM) is applied to enhance information retention during the down-sampling process; Finally, the DBDAMN clustering algorithm is utilized to optimize anchor sizes, allowing for more precise adaptation to the diversity of defect sizes. These innovations significantly improve the accuracy of surface defect detection in load-bearing rails, particularly in identifying small defects, offering an effective means of preventing workshop safety incidents. The experimental results demonstrate that this method achieves 97.3% on AP50, marking a 4.2% improvement over the standard YOLOv5 model, thus indicating a significant performance enhancement. To validate the superiority of our model, a comparison with popular current models was conducted, achieving optimal values in recall rate, accuracy, and mAP, which were 91.4%, 92.6%, and 88.9%, respectively. Therefore, the proposed method meets the requirements for precision in rail defect detection. Our shareable code and data sources are available at https://github.com/yhf177900/ EHA-YOLOv5.git.

Unmanned aerial vehicles(UAVs) as flight service platform are promising to deliver continuous required coverage to target users through efficient charging strategies. The agility-and-resilience of UAVs allow them to be rapidly assigned in accordance with mobile user status and charging strategies that require energy-efficient scheduling without compromising the current quality of service. However, as the limited coverage radius and endurance, efficiently scheduling multi-UAVs to adaptively charge, considering service fairness, and minimizing the replacement consumption of UAVs is difficult. Therefore, a multi-objective charging coverage problem is designed to address these factors, which involves the constrained Pareto dominance principle. Subsequently, an assisted deep reinforcement learning model based on the survival incentive (ADRL-SI) is proposed to adaptively charge UAVs while taking into account user fairness and energy-efficient deployment. ADRL-SI adopts a multi-objective particle swarm optimisation approach to assist different agents in learning. The survival incentive mechanism is responsible for securing the UAVs with minimum residual energy. Finally, simulation results display that the proposed ADRL-SI is more advantage than current DRL-based methods in the matter of user service level, UAV energy efficiency, adaptive charging security, and system convergence.