Addressing the limitations of current railway track foreign object detection techniques, which suffer from inadequate real-time performance and diminished accuracy in detecting small objects, this paper introduces an innovative vision-based perception methodology harnessing the power of deep learning. Central to this approach is the construction of a railway boundary model utilizing a sophisticated track detection method, along with an enhanced UNet semantic segmentation network to achieve autonomous segmentation of diverse track categories. By employing equal interval division and row-by-row traversal, critical track feature points are precisely extracted, and the track linear equation is derived through the least squares method, thus establishing an accurate railway boundary model. We optimized the YOLOv5s detection model in four aspects: incorporating the SE attention mechanism into the Neck network layer to enhance the model’s feature extraction capabilities, adding a prediction layer to improve the detection performance for small objects, proposing a linear size scaling method to obtain suitable anchor boxes, and utilizing Inner-IoU to refine the boundary regression loss function, thereby increasing the positioning accuracy of the bounding boxes. We conducted a detection accuracy validation for railway track foreign object intrusion using a self-constructed image dataset. The results indicate that the proposed semantic segmentation model achieved an MIoU of 91.8%, representing a 3.9% improvement over the previous model, effectively segmenting railway tracks. Additionally, the optimized detection model could effectively detect foreign object intrusions on the tracks, reducing missed and false alarms and achieving a 7.4% increase in the mean average precision (IoU = 0.5) compared to the original YOLOv5s model. The model exhibits strong generalization capabilities in scenarios involving small objects. This proposed approach represents an effective exploration of deep learning techniques for railway track foreign object intrusion detection, suitable for use in complex environments to ensure the operational safety of rail lines.

The identification of safflower filament targets and the precise localization of picking points are fundamental prerequisites for achieving automated filament retrieval. In light of challenges such as severe occlusion of targets, low recognition accuracy, and the considerable size of models in unstructured environments, this paper introduces a novel lightweight YOLO-SaFi model. The architectural design of this model features a Backbone layer incorporating the StarNet network; a Neck layer introducing a novel ELC convolution module to refine the C2f module; and a Head layer implementing a new lightweight shared convolution detection head, Detect_EL. Furthermore, the loss function is enhanced by upgrading CIoU to PIoUv2. These enhancements significantly augment the model’s capability to perceive spatial information and facilitate multi-feature fusion, consequently enhancing detection performance and rendering the model more lightweight. Performance evaluations conducted via comparative experiments with the baseline model reveal that YOLO-SaFi achieved a reduction of parameters, computational load, and weight files by 50.0%, 40.7%, and 48.2%, respectively, compared to the YOLOv8 baseline model. Moreover, YOLO-SaFi demonstrated improvements in recall, mean average precision, and detection speed by 1.9%, 0.3%, and 88.4 frames per second, respectively. Finally, the deployment of the YOLO-SaFi model on the Jetson Orin Nano device corroborates the superior performance of the enhanced model, thereby establishing a robust visual detection framework for the advancement of intelligent safflower filament retrieval robots in unstructured environments.

Traditional processing technology is not suitable for the requirements of advanced manufacturing due to the disadvantages of large repeated experiments, high cost, and low economic effect. As the latest additive technology, 3D printing technology has to deal with many issues such as process parameters and nonlinear mathematical models. A three-layer backpropagation (BP) artificial neural network with a Lavenberg–Marquardt algorithm was established to train the network and predict orthogonal experimental data. Additionally, the best combination of parameters of material deformations were predicted and verified by experiments. The results show that the predicted value obtained by the BP model is in good agreement with the experimental value curve, with a small relative error and a correlation coefficient of 0.99985. Moreover, the deformation errors of the printed model are not more than 3%. The incorporation of the BP neural network algorithm into the 3D printing process can, therefore, help cope with related problems, which is a future trend.

Considering that the vibration signals of gears and bearings in the automatic transmission device are complex and the fault features are difficult to extract. This paper proposes a method for extracting fault features of transmission device using adaptive variational modal decomposition (AVMD), and uses deep belief network (DBN) for pattern recognition. The vibration signal is decomposed by AVMD using energy ratio method. The intrinsic mode function (IMF) with abundant fault information is obtained. By calculating the energy entropy of each IMF component and form a high-dimensional feature vector as the input of DBN to establish an early fault identification model. The early fault data of the PHM2009 transmission device experimental platform was selected for identification and analysis. The identification results show that AVMD can extract the weak features of transmission device fault signals more accurately than empirical mode decomposition (EMD). Moreover, DBN has a higher fault identification accuracy rate than support vector machine (SVM), probabilistic neural network (PNN), back propagation neural network (BP) and Kohonen self-organizing competition neural network.

Local characteristic-scale decomposition (LCD) is an adaptive decomposition method for non-stationary and nonlinear time-varying signals. In this paper, kernel mapping is used to replace random mapping in LCD, and a fault feature extraction method based on kernel local characteristic-scale decomposition-Hilbert envelope spectrum (KLCD-Hilbert) is proposed. This method first performs wavelet noise reduction on the bearing fault signal, and then uses the KLCD method to adaptively decompose the motor bearing vibration signal into several intrinsic scale components (ISC), the kernel function determines the number of ISCs. Finally, create the Hilbert envelope spectrum of each state vibration signal in turn, and input the extracted characteristic data into the structure of the extreme learning machine (ELM) to realize the fault identification of the motor bearing. The experimental results show that the KLCD-Hilbert envelope spectrum can better reflect the fault characteristics of the motor bearing than the time domain or frequency domain amplitude in the process of identifying the state of the motor bearing. Moreover, the KLCD method has a higher recognition rate than the local mean decomposition (LMD) and empirical mode decomposition (EMD) methods.

The accurate and rapid identification of safflower filaments is a prerequisite for automating harvesting. This paper proposes a lightweight, high-precision detection model for safflower filaments based on YOLOv5s, named YOLOv5s-MCD, to address the issues of large existing network model sizes and low detection accuracy in complex natural environments. The Backbone of the YOLOv5s-MCD model was optimized into a lightweight improved network MobileNetv2 with DSC and CA modules, and the neck part incorporated the CA attention mechanism. The loss function is improved from DIoU’s non-maximum suppression method to CIoU to reduce the model size and improve the detection accuracy and speed. The experimental results show that the size of the YOLOv5s-MCD model is its size by 7.69 MB compared to the original YOLOv5s model, with a mean Average Precision (mAP) of 95.6% and an average detection time of only 3.2ms per image. When tested under unobstructed, obstructed, backlighting, shaking, and wide-angle natural environments, the improved YOLOv5s-MCD model increased the mAP value by 4.4, 0.7, 3.3, 3.4, and 1.0 percentage, respectively, compared with the YOLOv5s model, with improved F1 scores and confidence levels. This indicates that the improved lightweight model can achieve fast, real-time, and accurate detection of the safflower filaments in complex environments. The research results can provide a technical reference for the development of field safflower filament-harvesting robots.