In response to the current situation of relying on manual alignment of the optical path in existing velocity interferometer system for any reflector(VISAR) devices, and to meet the future demand for remote automated control, a new method for automatic alignment of the optical path was proposed. The complementary metal oxide semiconductor(CMOS) of this method was measured indirectly, and the pixel deviation of the light spot was used as a system input. Coefficient matrix transformation and discrete fuzzy feedback control methods were used to quickly eliminate the errors. Based on the modules such as vision and motion in the Windows control and automation technology(TwinCAT), each of which was run in a different real-time kernel, the communication link between the vision and motion control modules was eliminated, and fast real-time closed-loop control was realized. After the experimental verification of shock wave velocity measurement, the remote “one-button” automatic alignment was realized. The system can shorten the alignment time to 2 s and improve the alignment accuracy to 4.5 μm. The problem of inefficient manual adjustment of the existing device was solved, and the accuracy and stability of the system were improved.

Dust deposition can affect the normal operation of equipment. To accurately and efficiently detect dust on equipment and formulate a scientific cleaning strategy, a lightweight dust deposition detection method based on Fast-UNet was proposed. By effectively pruning UNet and adopting max pooling and bilinear interpolation for down-sampling and up-sampling operations, the parameter redundancy was reduced, and a compact basic network was obtained. The lightweight Ghost Module was used to replace the ordinary convolution in the basic network, further reducing the complexity of the network. An convolutional block attention module(CBAM) that integrated channel and spatial attention was embedded in the encoding process, which made the network pay more attention to the target area while introducing minimal parameters. Experiments on a simulated dust deposition dataset show that, compared with the original model, Fast-UNet reduces the number of parameters by 99.6%, decreases computational complexity by 98.7%, achieves an inference speed of 94.18 frames per second, and maintains a recognition accuracy of 91.17%. Compared with five other mainstream segmentation models, Fast-UNet also demonstrates advantages in both accuracy and speed. This method meets the needs of dust detection for both accuracy and efficiency, providing a technical reference for dust quantitative analysis.

Tunnel lining detection is an important element of quality management in tunnel construction and maintenance. Due to the variety of internal lining defects and unclear boundaries makes it challenging to identify these problems and train models effectively. Relying on manual detection or existing models, it is not possible to achieve fast and accurate defect detection. To address the above problems, A dataset consisted of 1 922 liner radar samples collected from Yunnan Tunnel B-scan was developed for training the model. A tunnel lining defect detection model YOLO-Tunnel based on YOLOv5 was proposed, which improved the model feature extraction ability, increased the receptive field, and improved the model localization ability by upgraded the Backbone and Neck. And further improved the model detection ability by selected the appropriate model size and balanced weight based on the dataset's scale and target size proportions. The results show that YOLO-Tunnel has better defect detection accuracy compared to YOLOv5s and also meets the real-time detection requirements, in which the precision, recall, and mAP are increased by 2.5, 9.0, and 8.1 percentage points, respectively, with the inference time increases by 2.7 ms to 21.8 ms. The research results provide a reference for further improving the performance of the detection of tunnel lining detection and the direction of optimization of the model reference.

To explore the factors affecting customers' evaluation of fresh logistics service quality, a logistics service quality evaluation model was proposed and established based on sentiment analysis of online reviews and latent Dirichlet allocation (LDA). A convolutional neural network (CNN) model integrating a multi-head self-attention mechanism and bidirectional long short-term memory network (BiLSTM) was constructed for sentiment analysis of online comments. Additionally, LDA topic model was carried out for positive and negative comments after classification. The key factors affecting the evaluation of fresh product logistics service quality were obtained by exploring the focus of customers' demand for fresh product logistics service. The sentiment analysis based on CNN-BiLSTM-Attention was implemented through Python programming, and the results of sentiment analysis on online comments were compared with those of support vector machine (SVM), CNN, BiLSTM, and CNN-BiLSTM. The comparison results show that, compared with the classification results of other models, the CNN-BiLSTM-Attention model is superior in accuracy, precision, recall rate, F₁, and other indexes, effectively improving the accuracy of text emotion classification. The research results demonstrate that researching the factors affecting the logistics service quality of fresh e-commerce based on online review data can help e-commerce enterprises better improve logistics efficiency and service quality from the perspective of consumer demand.

To address the high cost and low accuracy of manual inspection for steel surface defects, as well as the excessive computational resource requirements caused by complex traditional target detection models,a lightweight defect detection algorithm named YOLOv8n-MDC was proposed by integrating MobileNetv3 with YOLOv8.Firstly, based on YOLOv8n, the original intersection over union(IoU)-based bounding box loss function was replaced with weighted IoU(WIoU), enhancing model robustness through a non-monotonic focusing mechanism. Secondly, the backbone feature extraction network of YOLOv8n was substituted with MobileNetv3, utilizing its lightweight architecture to reduce network complexity and redundant computational overhead. Finally, during the feature fusion stage, depthwise separable convolution (DWConv) and C3Ghost modules replaced the original components, further minimizing model parameters and accelerating detection speed. Evaluated on the NEU-DET steel surface defect dataset, the YOLOv8n-MDC achieves an mAP of 81.3%, representing a 5% improvement over the baseline YOLOv8n, while its parameter count and computational complexity are reduced to 1.02 M and 2.1 GFLOPs (33.9% and 25.9% of the original model, respectively), meeting industrial requirements. This lightweight algorithm significantly reduces computational complexity and resource consumption while enhancing detection accuracy, offering an optimized solution for industrial steel surface defect inspection.

In the field of fingerprint recognition technology, ridge density, as one of the morphological features of fingerprints, has demonstrated increasing research value. Aiming at the problems of time-consuming and labor-intensive existing measurement methods, an algorithm based automated measurement method was proposed. The algorithm first preprocessed fingerprint images, including grayscale conversion, edge detection, noise reduction, and ridge enhancement, to improve image quality and clarity. Subsequently, it strengthened fingerprint features, performed array transformation, determined directional vectors, detects peaks, and finally plotted a grayscale fluctuation diagram to visually present the measurement results. Experimental results show that the automated measurement algorithm performs well in terms of efficiency and accuracy, exhibiting high consistency and significant statistical correlation with manual measurements. This further validates the scientific robustness and effectiveness of the automated measurement method, providing new perspectives and approaches for the automation and intelligence of fingerprint recognition.

In order to better address incidents of online violence resulting from uncontrolled public opinion in the era of self-media, network platforms are involved in decision-making during the early stages of online violence opinion formation. This can effectively prevent the formation of online violence opinions. Firstly, based on the inducement behaviors of online violence and considering the internal self-purification effects among platforms, self-media, and netizens, the costs and benefits of their autonomous behaviors during the initial stages of online violence opinion formation were defined. Next, a “platform-self-media-netizen” three-party evolutionary game model was constructed, and the behaviors of each subject and their evolutionary stable strategies were analyzed. Finally, numerical simulation experiments were conducted using MATLAB to verify the accuracy of the model and evolutionary results. To further reveal the factors influencing the cooperation among these parties, the impact of their initial cooperation willingness and related parameters on the system was explored. Simulation results show that a strong regulatory strategy adopted by network platforms, with clear guidance, can effectively enhance the cooperation willingness of self-media and netizens, effectively curbing the formation of online violence opinions.

As a key link between carbon source and carbon sink in carbon capture, utilization and storage(CCUS) technology, CO₂ pipeline transportation will play an important role in the process of carbon neutralization in the future. For the pipeline water hammer condition, the pressure oscillation may exceed the pressure in the pipe and be lower than the inlet pressure of the pump. At present, the water hammer and control theory of supercritical CO₂ pipeline is not mature. A mathematical model based on the law of conservation of mass, momentum and energy was established to describe the one-dimensional gas flow in the pipeline. The characteristic line method was used to solve the model, and the MATLAB programming was used to calculate. The simulation results were compared with the simulation results of the gas transmission system model proposed by Kiuchi and the simulation results of the commercial software OLGA. The results show that the simulation results are generally consistent with the simulation results of the gas transmission system model. Compared with OLGA software, the maximum relative errors of pressure and flow are 0.02% and 2.32%, respectively, which meet the requirements of engineering calculation accuracy. For the fast transient process of pipeline parameter change caused by pipeline compressor start and stop, valve emergency switch and rapid change of flow in a short time, the rapid change value is set to simulate. The established model can calculate the parameter change of each node with high accuracy, which can provide theoretical support and technical support for the localization of supercritical CO₂ pipeline transportation process simulation software.

In order to study a non-destructive testing method for concrete beam stress, a ultrasonic tail stress identification algorithm coda wave-deep residual shrinkage network (C-DRSN) based on deep residual shrinkage network(DRSN) was proposed. According to the high-dimensional characteristics of the tail wave signal vector, the interference of signal noise to the measurement stress accuracy was reduced by introducing residual contraction block, using soft threshold function and attention mechanism, and the adaptive recognition and extraction of stress features in the signal were realized, and the recognition accuracy was improved. The characteristics were visually analyzed, and the mapping relationship between the tail wave sign and the stress was established. In order to verify the model's ability of stress recognition, ultrasonic tail wave signals of concrete I-beams under three-point bending and eccentric compression loads were collected respectively. The results show that the recognition rate can reach 99% under both loading modes, indicating that the proposed method is feasible in the stress recognition of concrete beams, and the accuracy of the proposed method is higher than that of the tail wave interference method.

To realize the efficient analysis of composite joints with novel side-plate reinforced connections, a macro model of beam-column-slab composite joint was proposed. The optimal realization method of the connection between floor and steel beam was determined. The accuracy and reliability of the macro model was verified. Furthermore, a beam-column joint frame model with traditional side-plate reinforced connections (TSP), a beam-column joint frame model with novel side-plate reinforced connections (FBSP), and a composite joint frame model with novel side-plate reinforced connections (CJ-FBSP) were established. Elastic-plastic time-history analysis was conducted on three frame models considering the joint performance. The top point lateral displacement, inter-story drift angle, plastic energy dissipation, and plastic hinge distribution of different frame models under seismic waves were obtained. The results show that the performance of TSP and FBSP joints can be well simulated according to the spring stiffness calculation method. At the same time, the connection method considering the shear slip and pull-out performance of studs can precisely simulate the mechanical performance of composite joints. The inter-story drift angle of the three joint frame models do not exceed the specification limit (1/50). The maximum inter-story drift angle of FBSP frame model is smaller than that of TSP frame model but larger than that of CJ-FBSP frame model, and the plastic energy dissipation capacity is the best. Due to the strengthening effect of the floor, the plastic hinge rate of CJ-FBSP frame model is the smallest but the plastic energy dissipation capacity is weak.