Recently, deep-learning-based methods have achieved excellent performance in semiconductor wafer defect inspection. However, these trained networks are only effective for a certain background pattern and fail in defect transfer detection between multiple background patterns. Once the background pattern has changed, images are needed to be recollected and relabeled, which is very time-consuming. To achieve defect transfer detection in semiconductor wafers, a novel prototype learning (PL)-based network is proposed in this article. First, a multiscale residual fusion module (MSRFM) is elaborately designed to generate feature maps that minimize the interference of background patterns. Second, some labeled defect images are chosen as a defect library, which is further extracted into a defect prototype. Then, a distance map will be calculated between feature maps and the defect prototype, resulting in a precise prediction. Finally, an extra self-constraint loss ( $L_{\text {sc}}$ ) is designed to regulate the defect prototype calculation process. Different from previous methods, only defect-free images are required for the proposed method to achieve defect transfer detection. Experiments on real-world semiconductor wafer production lines show that the proposed method achieves mean intersection over union (mIoU) of 83.49% and 80.12% in defect transfer detection between two background pattern wafers. Furthermore, the excellent performance on other classical industrial datasets demonstrates that the proposed network has great robustness to various defects and industrial scenarios.

Municipal underground pipelines are closely intertwined with the daily lives of citizens. However, as a result of manufacturing imperfections and prolonged usage, pipelines frequently develop defects such as bulges, cracks, and damage. These issues can lead to leaks or complete failures, resulting in significant inconvenience for the public and substantial economic losses. Therefore, regular inspection and quality assessment of the inner walls of pipelines are essential. This paper presents a monocular structured light-based system for the 3D reconstruction of pipeline inner walls. The proposed system utilizes a laser emission unit and an industrial camera acquisition unit to generate a 3D point cloud of the pipeline’s inner surface, enabling the quantitative detection of defects. The system only requires a single calibration and can achieve a 360° unobstructed reconstruction of the entire inner wall. The camera is triggered by an encoder to produce real-time point cloud data. The system underwent testing on pipelines without defects, pipelines with quantified defects, and pipelines with irregular defects. Experimental results demonstrate that the generated point clouds can effectively quantify defects and assess the quality of the pipeline’s inner walls with millimeter-level accuracy.

Semiconductor wafers are the cornerstone of the intelligent electronic devices, and the precise fabrication of them is regarded as the crown of the manufacturing industry. With the total size of the semiconductor wafers getting smaller and their packaging process getting more complex, it becomes more and more difficult to inspect the surface defects of semiconductor wafers. Herein, a high resolution guided up-sampling(HRGU) based semantic segmentation network is demonstrated for high-accuracy wafer surface defect inspection. First, a novel HRGU block is developed to recuperate the information loss caused by down-sampling of the encoder, by guiding the up-sampling process in the decoder. Second, a boundary refine module(BRM) taking the image gradient map into account is designed to further improve the segmentation performance at edges. Finally, a boundary focal loss(BFLoss) is utilized to constraint the training process. Experiments are carried out on several typical open industrial defect datasets and our own wafer surface defect datasets. The proposed network exhibits a more superior detection performance compared to the other classical traditional counterparts, achieving a high segmentation metric mIoU of 80.71%, 87.05%, 91.23%, and 94.18% on KolektorSDD dataset, Magnetic Tile dataset and two our own wafer surface defect datasets gathered from industrial production lines.

This study adopts numerical simulation methods to explore the acoustic transmission characteristics of pipe strings in the upper and lower completions of a monitoring system for test production of natural gas hydrate. A finite-element simulation model for acoustic transmission in the pipe string system is established through COMSOL. The sound pressure level attenuation and the sound pressure amplitude ratio are chosen as evaluation indexes. Parametric numerical simulations are carried out to study the effects of the number of tubing cascades and the size of connection joints in the pipe string system on the acoustic transmission characteristics of the pipe string. The Light Gradient Boosting Machine (LightGBM) algorithm is adopted to predict the acoustic transmission characteristic curves of the pipe string. Based on this prediction model, with the maximum transmission distance, maximum sound pressure amplitude ratio, and minimum transmission attenuation as objective functions, the NSGA-II (Non-dominated Sorting Genetic Algorithm-II) optimization algorithm is adopted to obtain the optimal combinations of the pipe string system structure and the transmission frequency. The findings show that within the range of 20–2000 Hz, when the acoustic wave propagates in the column system, the amplitude attenuation caused by structural damping is positively correlated with the transmission distance, and the high-frequency acoustic wave attenuates faster. When the frequency exceeds 500 Hz, the sound pressure amplitude ratio is lower than 0.4, and the attenuation is stabilized at 90% above 1500 Hz. The thickness of the joints has a weak impact on the transmission, while an increase in length raises the characteristic frequency but exacerbates sound pressure attenuation. The LightGBM algorithm has a high prediction accuracy, reaching up to 88.54% and 84.82%, respectively. The optimal parameter combinations (n, hkg, lkg, freq) optimized by NSGA-II provide an optimization scheme for the structure and frequency of acoustic transmission in down-hole pipe strings.

Regular assessment of underground sewer pipe is vital for the maintenance of the sewer system. Previous studies mainly focus on automated defect detection in closed-circuit television images based on computer vision techniques. Nevertheless, challenging environmental conditions inside pipelines, such as liquid surface reflection and uneven illumination conditions, compromise the accuracy of automated sewer assessment systems. In order to address these issues, a multi-task visual perception framework (MVPF) is proposed for robust sewer defect segmentation. Firstly, we employ Pipe-SegNet to perform pipe defect segmentation task. This architecture, comprising dual-encoder branches, demonstrates strong robustness against diverse defect types inside pipelines. Subsequently, in order to cope with challenging environmental conditions, water segmentation task and dominant vanishing point detection task are assigned to Pipe-WSNet and Pipe-DVPNet, respectively. Liquid-air interface M_w and weak illumination region M_d can be derived from these two tasks. Eventually, M_w and M_d are further integrated to guide the training of Pipe-SegNet and refine its inference results, adhering to the distribution patterns of real-world defects. Extensive experiments based on practical engineering demonstrate that Pipe-SegNet achieves a more superior segmentation performance across multiple defect categories compared to other counterparts, and the mIoU of segmentation results is significantly improved by 4.07% through the integration of Pipe-WSNet and Pipe-DVPNet.