The boundaries of underground geological bodies are effectively highlighted through gravity inversion based on a typical focusing stabilizer, but this process is significantly affected by the focusing factor. The focusing characteristics are possessed by the exponential focusing stabilizer. To a certain extent, the issue of selecting the focusing factor can be circumvented by it, and it has the potential to have improved the effectiveness of three-dimensional gravity inversion.Furthermore, the multiplicity of solutions in inversion can be improved by the constraint imposed by the penalty function, which confine the inverted physical property values within a certain range. Based on this, in order to explore the improvement effects on three-dimensional gravity inversion by the exponential focusing stabilizer and penalty function constraint, the exponential focusing stabilizer and penalty function were incorporated into the three-dimensional gravity regularization inversion objective function. Comparisons were made of the effects resulting from three-dimensional gravity inversion, both with and without the involvement of the exponential focusing stabilizer, as well as with and without the application of the penalty function constraint. Model experiments have demonstrated that: the physical properties and spatial distribution of anomalous bodies can be accurately restored by the exponential focusing stabilizer, but there exist instances where false anomalies arise and physical property values surpass the true values. Situations like these can be improved by the exponential focusing inversion that is based on penalty function constraints. Furthermore, the accuracy of the solution can be enhanced by the zonal processing of penalty function constraints. The above situation indicates that the inversion method based on the exponential-type focusing stabilizer and constrained by penalty functions has certain potential to be generalized.

Traditional retinal vessel segmentation methods often face challenges such as missegmentation caused by optic disc confusion, lack of continuity in segmentation results, and imprecise segmentation in detailed regions. To address these issues, a retinal vessel segmentation algorithm was proposed based on UNet. The algorithm replaced traditional square convolutions with a fusion of horizontal and vertical one-dimensional convolutions and two-dimensional square convolutions, enhancing the representation capability of the eye region. A multi-scale branch approach was adopted to increase feature space diversity, thereby improving the network’s feature learning and expression capabilities. Additionally, to further enhance segmentation performance, multi-layer dilated convolutions was introduced into the deep structure of the autoencoder, replacing traditional simple pooling operations. This approach enlarged the convolution kernel size and expanded the receptive field, achieving a fusion of multi-scale shallow and deep feature information. The proposed algorithm was evaluated on the public DRIVE and CHASE_DB1 datasets. Experimental results demonstrates that the algorithm achieves precision (0.956 8 and 0.959 8) and F₁ scores (0.832 6 and 0.830 4), respectively. Compared with traditional UNet and recent UNet-based retinal vessel segmentation methods, the proposed algorithm shows advantages in accuracy, sensitivity, specificity, and F₁ metrics, these validation results fully demonstrate the proposed model’s strong capability in precise segmentation tasks.

During the production of monocrystalline silicon, defects generated during the crystal pulling process are recognized to severely impact product quality. Traditional visual-based defect detection methods, when applied to the detection of small protrusions in crystal pulling images, are confronted with challenges such as slow detection speeds, large parameter volumes, and difficulties in deployment on embedded terminals. In response to these challenges, an improved YOLOv8 object detection model was proposed incorporating a ContextGuided module to enhance the inference efficiency of the model. An efficient DySample was introduced into the feature fusion network to optimize the efficiency and depth of feature fusion. A lightweight network structure was adopted to reduce the complexity and computational demands of the model, making it suitable for devices with limited computing resources. The model has been trained and tested on an industrial dataset, demonstrating a more accurate detection of small protrusions with a mean average precision (mAP) of 97.7%. Compared to YOLOv8n, it exhibits an increase of 11.6% in precision and a reduction in parameter volume by 31.9%, facilitating its deployment on embedded terminals.

The rod of a fire drive production well in Xinjiang oilfield is seriously corroded, and there are many corrosion pits on the surface. In order to find out the cause of rod corrosion failure, the metallographic structure and physical and chemical properties of the failed rod were tested and analyzed, and the corrosion characteristics were analyzed by scanning electron microscope, energy spectrometer and X-ray diffraction analysis, and the failure causes were found out combined with the service conditions of the rod. The results show that the chemical composition, metalloid structure and inclusions of the D-class rod meet the standard requirements of the rod. The rod has only been in service for 1 year and 8 months and its diameter has decreased from 19 mm to 16 mm, the average corrosion rate is 1.796 mm /a, and the maximum surface corrosion pit depth is 2.1 mm. The surface of the rod is mainly attached with FeCO₃, FeS and CaCO₃. It is judged that the rod is corroded by CO₂/H₂S and underscale corrosion caused by high salinity produced water. At the same time, a large amount of Cl^- in the produced liquid promotes the development of pitting pits. It is recommended to inject corrosion and scale inhibitor into production wells with temperature lower than 50 ℃, pH value between 6 and 8, and formation water belonging to high calcium and high chlorine to protect the D-class rod and extend the service life of the rod.

In order to reasonably and efficiently carry out the risk assessment of road tunnel construction collapse, the risk assessment model of road tunnel construction collapse was studied by rough set (RS), grid search method(GS) and support vector classification (SVC). Firstly, the index system of highway tunnel construction collapse risk evaluation was constructed by integrating advanced geological prediction. At the same time, the information of 100 tunnel collapse cases was collected and the index data was discretized. Secondly, attribute reduction was conducted based on the condition information entropy of rough set to obtain the reduced core index set. Then, grid search method was used to find the optimal parameters of the support vector classification training set, the risk assessment model of highway tunnel construction collapse based on rough set-grid search-support vector classification (RS-GS-SVC) was established. Finally, the model was used to predict the test samples. The results show that under the condition of the same learning sample, compared with rough set-genetic algorithm-support vector classification (RS-GA-SVC) model and Rough set-particle swarm optimization-support vector classification (RS-PSO-SVC) model, RS-GS-SVC model has higher classification accuracy; Under the same proportion of training set and test set, the prediction accuracy of RS-GS-SVC model is higher than that of GS-SVC model, with the accuracy rates of 93.33% and 90% respectively, and the operation time of RS-GS-SVC model is shorter. It can clearly be seen that the model complexity is effectively reduced and the classification accuracy is improved through the reduction of rough set conditional information entropy attributes.

In order to improve the problem of easy cracking in the negative moment zone of the composite beam, the UHPC(ultra-high performance concrete)-narrow steel box composite beam structure was proposed, and the shear load capacity calculation method of UHPC-narrow steel box composite beam was obtained based on the ultimate equilibrium method and the sub-stacking method. In order to verify the accuracy of the calculation method, one normal concrete narrow steel box composite beam and three UHPC-narrow steel box composite beam specimens were designed with the thickness of UHPC layer and steel fiber as variables, and the test beams were subjected to mid-span reverse loading test to obtain the shear resistance of the composite beam under negative bending moment. The test results show that compared with the normal concrete narrow steel box composite beam, when the UHPC material is used in the wing slab, the crack distribution in the wing slab was more regular, and with the increase of the thickness of the UHPC slab, the crack distribution gradually shows the characteristic of vertical equidistant distribution, and when the thickness of the wing slab was kept constant and the thickness of the UHPC layer was increased by 50 mm, the yield load increases by 12.5% and the ultimate load increases by 8.3%, and the deflection value corresponding to the ultimate load is reduced by 22%. By analyzing the yield deflection as well as the ultimate deflection of each specimen, it can be seen that the yield deflection tends to increase gradually with the increase of UHPC airfoil thickness, and the proportion of the elastic phase of the combined beam keeps increasing. The comparison of the test data with the theoretical calculation results shows that the calculation results obtained by the sub-stacking method can more effectively reflect the contribution of UHPC layer, ordinary concrete layer, steel box, reinforcement and filled part of concrete to the shear load capacity, and the results are more accurate.

Land use and traffic safety is a hot topic of mutual concern in the fields of urban geography and transportation. However, existing research on the impact of land use on pedestrian traffic accidents often incorporates a unified framework of built environment and mainly adopts measures such as land use mix or the proportion of land use types, lacking detailed analysis on land use types, thus making it difficult to translate their findings into actionable design strategies. Taking Yuzhong District, Chongqing City as an example, the land use types were finely characterized based on point of interest (POI) data, and the extreme gradient boosting (XGBoost) model was applied to explore the nonlinear relationships between land use types, pedestrian, road conditions, road environment, and the severity of pedestrian traffic accidents. The study finds these as follows. ①Land use types play an important role in the severity of pedestrian traffic accidents, with hospitals, residential areas, and educational land being the most influential. The presence of hospitals, residential areas, and educational land within a 300-meter buffer zone around accident sites reduces the severity of pedestrian traffic accidents. ②Road sections with curved and sloped road alignments are high-risk areas for severe pedestrian traffic accidents; entrances and exits of road sections, narrow road sections, and intersections have a mitigating effect on the severity of pedestrian traffic accidents. The findings provide policy insights for refined land use planning and governance to reduce the severity of pedestrian traffic accidents.

In response to the tasks of detection and maintenance in small-diameter pipeline transportation, a small-scale soft pneumatic pipeline crawling robot was designed and developed. The robot employed the technique of analytical implicit triply periodic minimal surface offset to form a solid structure with thickness, and achieved the anisotropy of structural stiffness by adjusting the parameters of implicit equations. Based on this, the radial and linear actuators were designed to meet the flexible motion requirements of the robot inside the pipeline. To optimize the performance of the robot, a joint parameter optimization framework based on MATLAB, Rhino, and ABAQUS was proposed, and an automated parameter optimization process was implemented using Python scripts. Through this framework, the motion performance and adaptability of the robot could be effectively improved. Based on two types of deformation actuators, the overall structure of the pipeline crawling robot was designed, and the manufacturing method and motion control gait were elaborated in detail. Experimental results show that the designed pipeline robot can stably move along the pipeline in two different postures and under certain load conditions, verifying the feasibility of the joint parameter optimization framework. The research results provide valuable references for the parameterized design and optimization research of subsequent pipeline crawling robots, which contributes to the application development of soft robots in the field of small-diameter pipeline detection and maintenance.

A stepped-frequency radar transmission and reception system based on RF direct sampling technology was designed to address the issues of slow switching speed of sub pulses and incoherent phase of transmission and reception signals in stepped-frequency radar. This system performs RF direct acquisition of C-band radar signals, and all signal processing processes were carried out in the digital domain. It also had high integration and low power consumption. It was generated stepped-frequency signals by fast frequency hopping, frequency sweep of 1.8 GHz bandwidth within 7 ms was gotten. The system operates in the frequency range of 200~2 000 MHz with a frequency step of 2 MHz. At the same time, the parameters of the stepped-frequency signal in this system can be changed according to detection requirements. The test results show that the stepped-frequency radar system can achieve one-dimensional ranging function.

To study the wind pressure distribution characteristics of long-span roofs of airport terminals at different mountain heights in mountainous areas, a rigid model wind tunnel pressure measurement test of airport terminals roofs at mountain heights of 0 m, 30 m, 60 m, and 90 m was conducted to compare and analyze the effects of the heights on the surface mean and pulsating wind pressure, non-Gaussian characteristics of pulsating wind pressure, peak factor, and extreme wind pressure of the roof surface. The results show that the increase in mountain height significantly increases the mean and fluctuating wind pressure coefficient at the windward leading edge of the roof, and also intensifies the degree of flow separation at the leading edge of the roof. This causes the skewness, kurtosis, and probability density function of the fluctuating wind pressure at the windward leading edge of the roof to deviate significantly from the standard Gaussian distribution, exhibiting strong non-Gaussian characteristics. At the same time, the Hermite moment model was used to calculate the peak factor, and it was found that the peak factor of most measuring points on the roof surface was mainly distributed in the range of 3.5~4, which was much higher than the recommended value of 2.5 in GB 50009—2012. The extreme wind pressure value at the front edge of the roof also increased with the increase of the mountain height, and there was a similar variation pattern at the edge of the roof under all wind directions. Among them, the most unfavorable extreme negative pressure on the roof surface at a mountain height of 90m decreased by 44.9% compared to the 0m mountain height. Research can provide useful suggestions and references for the design and construction of terminals in similar airports.