Ming Liu

Detecting defects on steel surfaces is crucial for ensuring product quality and production safety in industrial settings. Object detection using deep learning, particularly the YOLOv5 model, has become a widely adopted method for this purpose. However, the complex shapes of current steel surface defects pose challenges for precise detection, especially when using low-cost recognition devices with small resolution images. To address these challenges, we integrated the RepBi-PAN fusion network into YOLOv5, enhancing the detection capability for large targets in complex backgrounds. To mitigate issues related to the premature introduction of shallow features and decrease in Precision, we optimized the model structure by incorporating the DenseNet structure into the backbone for improved feature extraction. Additionally, we introduced the Normalized Attention Module (NAM) to enhance the detection capability for small targets. Experimental results demonstrate the effectiveness of the enhanced model, showing a 4.1% increase in mean average precision (mAP), a 3.2% improvement in precision, and a 2.4% enhancement in recall. The improved algorithm outperforms the original in complex backgrounds and recognizing small targets, addressing limitations of the Rep-Bi network. Compared to other YOLO algorithms, our approach achieves optimal values for recall and mAP while maintaining a smaller model size. When compared to YOLOv9, which is the best-performing algorithm in the YOLO series on the dataset used in this study, our model demonstrates several advantages. Specifically, it maintains superior overall performance with fewer parameters and lower computational requirements compared to deeper YOLOv9 variants. Furthermore, when compared to YOLOv9s, our model exhibits better performance in terms of precision, recall, and mAP, while also having fewer GFLOPs, a smaller parameter count, and a reduced model size.

Resistance to poly (ADP-ribose) polymerase (PARP) inhibitors (PARPis) remain a major challenge in ovarian cancer (OC) treatment. However, the underlying mechanism of PARPi resistance is still poorly characterized. Increasing evidence has proven that lymphocyte cytosolic protein 1 (LCP1) promotes tumor progression. The JAK2/STAT3 signaling pathway plays an important role in increasing tumor metastatic ability and chemoresistance in cancer by promoting epithelial - mesenchymal transition (EMT). We established an olaparib-resistant OC cell line and studied its toxicologic effects through cell survival, Transwell, colony formation, western blotting and flow cytometry assays. RNA sequencing and screening were then performed to identify genes associated with olaparib resistance. Lymphocyte cytosolic protein 1 (LCP1) was found to be overexpressed in olaparib-resistant OC cells. The inhibition of cell survival and promotion of cell apoptosis induced by olaparib in parental cells were significantly attenuated in olaparib-resistant cells. LCP1 was upregulated in olaparib-resistant cells compared with parental OC cells. Moreover, we found that the protein levels of JAK2/STAT3 signaling pathway components and EMT markers were increased in olaparib-resistant cells. Overexpression of LCP1 increased olaparib resistance in OC cells, and knockdown of LCP1 attenuated olaparib resistance. The changes in the protein levels of JAK2/STAT3 signaling pathway members and EMT markers between the cell types were similar to the changes in the levels of LCP1. These findings indicate that LCP1 expression may play an important role in the resistance of OC to olaparib by activating the JAK2/STAT3 signaling pathway and EMT. LCP1 could be a potential therapeutic target for patients with OC who are resistant to olaparib. Our study provides a new mechanism of olaparib resistance.

In response to the challenges of limited dataset size and low image resolution in steel surface defect datasets, leading to suboptimal recognition accuracy and overall performance of deep learning algorithms, this paper proposes a deep learning-based model for steel surface defect detection. Leveraging the AFPAN (Asymptotic Feature Pyramid Network) as the fusion network, the model utilizes the YOLOv8 backbone network for feature extraction. Additionally, it introduces the SPD-Conv (Space-to-Depth) block and the EffectiveSE (Effective Squeeze-and-Excitation) attention mechanism, specifically designed for low-resolution images. While maintaining a low parameter count, the model significantly enhances performance. Compared to algorithms such as SSD (Single Shot MultiBox Detector) and DETR (DEtection TRansformer), our algorithm achieves a more balanced Precision and Recall, without exhibiting a noticeable imbalance between the two metrics. Additionally, it surpasses algorithms like RT-DETR (Real Time-DEtection TRansformer) in terms of accuracy and comprehensiveness. In contrast to the YOLO (You only look once) series, the improved network exhibits optimal performance in smaller models, achieving a 4.8% precision improvement and a 3.2% increase in mAP (mean average precision). Experimental results on the NEU-DET (NEU surface defect database) dataset affirm that, compared to the YOLOv8s model, our proposed model excels in accurately and comprehensively identifying defect positions and edges in defect detection.

In order to measure absolute distances between multiple targets over a wide range, a measurement estimation method is designed which is based on a fusion strategy of temporal features and nonlinear fitting of satellite receiver positions. The method takes the undifferential-corrected latitude and longitude data collected by a low-cost satellite receiver through wavelet denoising, and then realizes the accurate distance estimation of the measurement target through the fitted regression model designed to fuse the Temporal Convolutional Network (TCN) and the Fully Connected network (FC).To enable model training, a dataset featuring deep fusion characteristics is constructed, while employing a Genetic Algorithm (GA) to optimize the weights and thresholds of the model’s fully connected layer, thereby enhancing its nonlinear regression capability. The method also makes up for the defect of the Haversine formula for latitude and longitude distance calculation, which has a high demand for localization accuracy. The experimental results show that the neural network model designed and optimized in this paper is better than the other six models in terms of performance, which fully verifies that the method has good data fitting ability and prediction accuracy, and is practical in the static distance measurement problem.

Detecting defects on steel surfaces is crucial for ensuring product quality and production safety in industrial settings. Object detection using deep learning, particularly the YOLOv5 model, has become a widely adopted method for this purpose. However, the complex shapes of current steel surface defects pose challenges for precise detection, especially when using low-cost recognition devices with small resolution images. To address these challenges, we integrated the RepBi-PAN fusion network into YOLOv5, enhancing the detection capability for large targets in complex backgrounds. To mitigate issues related to the premature introduction of shallow features and decrease in Precision, we optimized the model structure by incorporating the DenseNet structure into the backbone for improved feature extraction. Additionally, we introduced the Normalized Attention Module (NAM) to enhance the detection capability for small targets. Experimental results demonstrate the effectiveness of the enhanced model, showing a 4.1% increase in mean average precision (mAP), a 3.2% improvement in precision, and a 2.4% enhancement in recall. The improved algorithm outperforms the original in complex backgrounds and recognizing small targets, addressing limitations of the Rep-Bi network. Compared to other YOLO algorithms, our approach achieves optimal values for recall and mAP while maintaining a smaller model size. In comparison with YOLOv8, the improved model surpasses all V8 models, being only 0.5% below the precision of the largest YOLOv8x model. Simultaneously, the improved model is smaller and has fewer parameters compared to all models in the YOLOv8 series, with slightly higher GFLOPs than the smaller v8 models.

The human pose has a natural graphic structure. Therefore, in recent years, many studies on human pose estimation is based on Graph Convolutional Networks (GCN). However, the classical GCN method has a small receptive field and the transformation matrix is shared by all nodes, which limits their feature extraction ability. To address these issues, a novel multi-branch attention graph convolution (MultiBA_GConv) operation is proposed and a regression network model for human 3D position estimation based on the MultiBA_GConv operation is developed in this paper. Several different transformation matrices are used to extract the feature information contributing to the node itself, its neighbors, and other global nodes, and the corresponding attention mechanism is used to focus on these features in the MultiBA_GConv operation. The ablation experiments show that the multi-branch processing and multi-attention mechanism proposed in the MultiBA_GConv operation can greatly improve the feature extraction ability of graph convolution. The comparison experiments with other GCN methods show that MultiBA_GConv has the best feature extraction performance in all graph convolution operations. Based on the MultiBA_GConv, a 3D human pose regression network architecture is designed and the proposed method is evaluated on a challenging benchmark dataset for 3D human pose estimation. The experimental results of the comparison with the state of the art show that our method is highly competitive.

For large and medium-sized cities, the planning and development of urban road networks may not keep pace with the growth of urban vehicles, resulting in traffic congestion on urban roads during peak hours. Take Jinan, a mid-sized city in China’s Shandong Province, for example. In view of the daily traffic jam of the city’s road traffic, through investigation and analysis, the existing problems of the road traffic are found out. Based on real-time, daily road traffic data, combined with the existing road network and the planned road network, the application of a road intelligent transportation system is proposed. Combined with the application of a road intelligent transportation system, this paper discusses the future development of urban road traffic and puts forward improvement suggestions for road traffic planning. This paper has reference value for city development, road network construction, the application of intelligent transportation systems, and road traffic planning.

Due to the high risk factors in the electric power industry, the safety of power system can be improved by using the surveillance system to predict and warn the operators’ nonstandard and unsafe actions in real time. In this paper, aiming at the real-time and accuracy requirements in video intelligent surveillance, a method based on edge computing architecture is proposed to judge unsafe actions of electric power operations in time. In this method, the service of unsafe actions judgment is deployed to the edge cloud, which improves the real-time performance. In order to identify the action being executed, the end-to-end action recognition model proposed in this paper uses the Temporal Convolutional Neural Network (TCN) to extract local temporal features and a Gate Recurrent Unit (GRU) layer to extract global temporal features, which increases the accuracy of action fragment recognition. The result of action recognition is combined with the result of equipment target recognition based on the yolov3 model, and the classification rule is used to determine whether the current action is safe. Experiments show that the proposed method has better real-time performance, and the proposed action cognition is verified on the MSRAction Dataset, which improves the recognition accuracy of action segments. At the same time, the judgment results of unsafe actions also prove the effectiveness of the proposed method.

Due to the high risk factors in the electric power industry, the safety of power system can be improved by using the surveillance system to predict and warn the operators' nonstandard and unsafe actions by real time. In this paper, aiming at the real-time and accuracy requirements in video intelligent surveillance, a method based on edge computing architecture is proposed to judge unsafe actions of electric power operations in time. In this method, the service of unsafe actions judgment is deployed to the edge cloud, which improves the real-time performance. In order to identify the action being executing, the end-to-end action recognition model proposed in this paper uses the Temporal Convolutional Neural Network (TCN) to extract local temporal features and a Gate Recurrent Unit (GRU) layer to extract global temporal features, which increases the accuracy of action fragment recognition. The result of action recognition is combined with the result of equipment target recognition based on yolov3 model, and the classification rule is used to determine whether the current action is safe.Experiments show that the proposed method has better real-time performance,and the proposed action cgnition is verified on the MSRAction Dataset,which improves the recognition accuracy of action segments.At the same time,the judgment results of unsafe actions also prove the effectiveness of the proposed method.

The performance of vehicle body vibration and ride comfort of active or semi-active suspension with proper control is better than that with passive suspension. The key to achieve good control effect is that the suspension control system should have strong real-time learning ability according to changes in the road surface and suspension parameters. In the control strategies adopted by previous researchers, the classical neural network controller has some learning ability, but it is mainly based on offline learning with a large number of samples. In this paper, the deep reinforcement learning strategy is used to solve the above problems. Aiming at the continuity of state space and execution action in vehicle active suspension system, the control of the semi-active suspension is realized by using improved DDPG (Deep Deterministic Policy Gradient) algorithm. To overcome the shortcoming of low efficiency of this algorithm in the initial stage of learning, the DDPG algorithm is improved and using empirical samples in the learning method is proposed. Based on Mujoco, the physical model of semi-active suspension is established, and its dynamic characteristics are analyzed under the condition of various road level and vehicle speed. The simulation results show that compared with the passive suspension, the semi-active suspension based on improved DDPG algorithm with learning method using experienced samples can better adapt to various road level, more effectively reduce the vertical acceleration of the vehicle body and the dynamic deflection of the suspension, and further improve the ride comfort.