In order to study the influence and distribution law of different width to thickness ratios of the corrugated web I-beam on the welding residual stress, the specimens with different width to thickness ratios were studied by numerical simulation and experiment. By using ABAQUS software and complete coupling method, a thermoelastic plastic three-dimensional finite element model of the corrugated web I-beam was established, and its welding stress field was also obtained. The distribution law of welding residual stress in different path positions and the magnitude and distribution of residual stress of the specimens with different width to thickness ratios were studied, and the distribution model was drawn. A systematic study was conducted on the influence of plate thickness and width to thickness ratio on residual stress of components, and the mechanism of influence between components was further explored. The research results show that the results obtained by the finite element model of the corrugated web I-beam welding has a good agreement with the residual stress measured by the test. The overall distribution of residual stress is mainly longitudinal residual stress, and the maximum residual tensile stress exists near the weld of the flanges, and the maximum tensile stress is 415 MPa. Under the condition that other factors remain unchanged, the residual compressive stress of flange and web overhang is inversely proportional to the change of width to thickness ratio and the thickness of plate. Unlike this, the magnitude of residual tensile stress does not directly change significantly with the width to thickness ratio or the thickness of the plate. Finally, based on the numerical simulation analysis, the residual stress distribution model and calculation formula of 1/2 section of corrugated web I-beam are proposed, which provides reference for the subsequent experimental research on residual stress of corrugated web.

Dynamic shear rheological test and multiple stress recovery creep test were used to measure the high temperature rheological properties of ultraviolet (UV) aged asphalt. Meanwhile, the component composition test, element composition test, gel permeation chromatography test and infrared spectrum test were used to measure the changes of microscopic composition such as component composition, element composition, molecular weight distribution, characteristic peaks and functional groups of asphalt. The variation rules of macro and micro aging behavior of asphalt under ultraviolet radiation and its evaluation indicators were studied. The results show that after UV aging, the rutting factor and unrecoverable compliance of asphalt are increased, the phase angle and recovery rate are decreased, and the high-temperature stability is increased. At the same time, as a polymer material, asphalt breaks its internal chemical bonds and long chain and branch chain molecules under the action of ultraviolet, and oxidizes under the action of oxygen, which increases the asphaltene and converts the asphalt structure to gel type. After UV aging, the molecular weight distribution of asphalt becomes narrower, and there is a phenomenon of small molecules continuously aggregating into large molecule compounds inside, causing changes in characteristic peaks and the production of carbonyl and sulfoxide functional groups. It is found that the rutting factor and recovery rate can be used as the evaluation indicators of high-temperature rheological properties of asphalt UV aging, while carbonyl and sulfoxide functional groups can be used as the indicators of aging generation and anti-aging ability, respectively.

Traditional slope monitoring focuses on the slope as a whole and pays less attention to the local deformation and damage of the slope, and the traditional monitoring technology has low precision and poor timeliness, which couldnt realize the timely capture of local deformation. Therefore, the soil slope was taken as the research object. Firstly, a slope hazard level model was constructed on the basis of analytic hierarchical analysis method, and the slope was monitored and classified according to the hazard level. Then, initial zoning and zoning optimization plans were developed for slope with different monitoring levels, which realized the zoning delineation of slopes. Finally, the feasibility and reliability of the proposed zoning technology system were verified by combining with the soil slope of a highway in Wanzhou, Chongqing. The research results show that the targeted monitoring based on slope zoning can realize rapid capture of local deformation, and slope zoning monitoring has a higher level of non-redundant equipment deployment compared to conventional monitoring. The monitoring scheme based on slope zoning is more scientific and the monitoring results are more reliable, which indicates that the zoning technology system is feasible. The results of the study can provide new solutions for monitoring other types of slopes.

Hydraulic tunnels carry important functions such as water diversion, water transmission and drainage, and are an important part of the national water network construction. With the emergence of large diameter and deep buried hydraulic tunnel structures in western region, many scholars have studied the formation mechanism of lining concrete cracks. Firstly, based on the literature review and organization, starting from the types of defects in hydraulic lining concrete, it was found that lining cracks were one of the important causes of diseases, which became a key factor restricting the efficiency of water transportation and diversion. Secondly, according to the causes of lining cracks and their serious engineering consequences, anti-cracking technologies such as surrounding rock prediction, temperature control measures, concrete material performance improvement and intelligent temperature-controlled maintenance technology were discussed. Then, the causes of deterioration of reinforced concrete lining were summarized from the multi-scale coupled analysis model. Finally, based on the real loads and environmental conditions that lining concrete was subjected to in actual projects, it was expected to use a multi-scale coupled analysis model to provide a new method to reveal the cracking mechanism of lining concrete.

In order to clarify the deformation law of diaphragm walls caused by deep foundation pit excavation of subway stations in soft soil areas along the eastern coast, the statistical analysis was carried out on the deformation characteristics of diaphragm walls in deep excavation of subway stations in eastern coastal areas represented by Hangzhou, Shanghai and Ningbo. The results show that the lateral deformation of the diaphragm walls in the above three areas is concentrated between 0 and 30 mm, which is greater than the statistical value of lateral deformation of the enclosure structures in other regions. The maximum lateral deformation of the diaphragm wall increases with the increase of the excavation depth of the foundation pit. The maximum lateral deformation of the diaphragm wall in Hangzhou region varies greatly with the excavation depth of the foundation pit, but the impact of the excavation depth of the foundation pit on the deformation of the diaphragm wall is smaller than that in Shanghai and Ningbo regions. The non-dimensionalization treatment of the lateral deformation curve of the diaphragm wall shows that the deformation of the diaphragm wall in the foundation pit of the three subway stations exhibits a "convex" pattern. The lateral displacement of the diaphragm wall increases rapidly within the range from 0 to 0.5 x (x represents the ratio of the corresponding position of the maximum lateral deformation of the diaphragm wall within the excavation depth to the excavation depth of the foundation pit), and the maximum lateral displacement occurs between 0.5 x to 1.1 x, which gradually decreases thereafter.

The behavior pattern of companion passengers is a comprehensive embodiment of the transportation needs of each member of the group, especially the regular companion behavior based on close social relations, which is very important for the transportation organization and planning department to understand the needs of companion passengers and provide targeted services and management. Based on the subway card data of smart card data (SCD) of public transportation in Beijing, the identification rules of regular companion passengers were proposed by analyzing the interpersonal distance and the frequency of companion behavior during travel. In order to explore the travel patterns of different types of companion passengers, starting from the dimensions of time and space, the mean value of first departure time, the mean value of last departure time, the coefficient of variation of the first departure time, the coefficient of variation of the last departure time, and the OD repetition rate of the companion passengers were selected as the clustering indicators. Based on these indicators, companion passenger travel patterns were classified into three main types and named according to their travel characteristics: entertainment early travel companion mode, entertainment late return companion mode and commuting companion mode. For different types of companion passengers, the personalized ticketing policies were proposed. These personalized ticketing policies aim to meet the needs of different types of companion passengers and improve the attractiveness and competitiveness of public transport.

In the Yangtze River waterway, navigation marks are the "eyes" of ships, guiding them to safely navigate along the waterway or the scheduled route, so it is extremely important to scientifically distribute navigation marks. At present, navigation marks on key segments such as ferries in the lower Yangtze River still rely on manual experience. However, with the rapid increase in the size and number of ships, the contradiction between the layout of navigation marks in key water areas and the safety of ship navigation is becoming more and more prominent, so it is urgent to rationally allocate and scientifically optimize the navigation marks in key water areas. Therefore, a simulation platform of traffic flow for inland vessel was established based on Multi-agent traffic flow simulation model and conflict detection method of ship domain. Taking the Fubei section of the lower reaches of the Yangtze River in Jiangsu province as the research object, the high-risk water areas for navigation in this channel were identified and the navigation routes were optimized by the traffic flow simulation platform and historical navigation data. And based on this, a layout plan combining virtual and physical navigation marks was developed, effectively enhancing the support capability of ship's safe navigation.

In order to scientifically evaluate the level of civil aviation safety supervision, according to the categories of administrative counterpart safety problems in civil aviation safety supervision, a safety supervision capability evaluation index system covering facilities and equipment, employees, rules and regulations or procedures and management-related issues was established by use of SHEL model, which was based on the systematic characteristics of civil aviation supervision and operation. The subjective and objective weights of indicators were determined by improved G1 method and CRITIC method, and the comprehensive weights of each index were combined and calculated by game theory. Finally, the rank-sum ratio method was used to comprehensively evaluate the safety supervision capacity of civil aviation regulatory agencies. The results show that the established model can objectively evaluate the safety supervision capability of civil aviation administrative departments, which is consistent with the actual supervision performance level of the units. It is proved that the proposed model is scientific, effective and accurate.

Aiming at the problem of low prediction accuracy of the model caused by the imbalance of traffic accident data categories, the oversampling technology and Bayesian network were integrated to improve Catboost algorithm, so as to predict the severity of traffic accidents. PowerSHAP algorithm was used to screen out 21 key features. Borderline-SMOTE oversampling was used to perform balanced sampling on minority class samples located at the boundary. Bayesian optimization method was used to adjust the hyperparameters of Catboost algorithm, and the traffic accident severity prediction model was established based on the optimized Catboost algorithm. Empirical verification shows that the improved model outperforms other classification models in predicting the severity of traffic accidents, with accuracy, precision, recall, and F1 score of 83.03%, 87.01%, 80.79%, and 83.50%, respectively. Compared with other parameter optimization methods, the parameter performance after Bayesian optimization is improved by more than 3 percentage points. Borderline-SMOTE oversampling technology has achieved performance evaluation indicators of over 88% in serious injury accidents with a small data proportion, which is at least 3.3% higher than other models. Moreover, in terms of predicting the severity of traffic accidents with imbalanced categories, Borderline-SMOTE oversampling technology has higher accuracy than other five data sampling techniques.

Urban intersections usually involve traffic flow, pedestrians and non-motorized vehicles from multiple directions. The interaction and conflicts between various traffic participants make the traffic scene very complex, which causes the increase of difficulty of data collection. Vehicle-mounted LiDAR (VM-LiDAR) could scan the surrounding environment in real time and obtain traffic data around vehicles. Based on these data and combined with vehicle networking technology, the intelligent traffic control of complex road sections can be achieved. VM-LiDAR technology was used to collect point cloud data at urban road intersections, and support vector machine was used to identify urban road intersection targets after the conditional Euclidean clustering algorithm was introduced to segment the point cloud data. According to the results of target recognition, vehicle delay, parking frequency and traffic capacity, a multi-objective optimization model for urban road intersection traffic control was established in the vehicle networking system. Moreover, the improved genetic algorithm was used to solve the proposed model, and the optimal control scheme was obtained, which realized the intelligent traffic control at urban road intersections. The research results show that the proposed control method can effectively identify targets at urban road intersections, reduce the average number of stops per vehicle, improve the average traffic rates and reduce traffic delay rates, whose traffic control effect at urban road intersections is significant.

Aiming at the U-shaped steering scenario of unmanned vehicles on two-way single lane, a U-shaped steering trajectory planning method was proposed based on the spatial distribution relationship and forward and reverse trapezoidal acceleration. In this method, a U-shaped steering trajectory was planned firstly according to the spatial distribution characteristics of vehicles, and the linear interpolation method was then used to smooth the trajectory. Then, the lateral acceleration programming method of forward and reverse trapezoid was used to design the multi-objective function to obtain the optimal solution of lateral acceleration and lateral acceleration change rate. Finally, an improved Stanley control method based on the spatial distribution relationship of vehicles was proposed, and the trajectory error converged to 0 by adding a steady-state error penalty term into the input function. The simulation results show that compared with the Stanley method, both the maximum lateral trajectory error and the maximum heading angle error of the proposed control method are improved at different initial speeds. The results of driver-in-the-loop experiments show that the proposed method can effectively complete the U-shaped steering.

In order to solve the problem that it is difficult to make certain and reasonable decisions under the current ethical dilemma of autonomous vehicle, which hinders the application and development of self-driving technology, therefore, an ethical decision-making model based on multi-scale mixed attention mechanism convolutional neural network (CNN) was designed for autonomous vehicles. Firstly, according to the structure and function of CNN as well as the characteristics of ethical dilemmas, the training model and parameters of CNN were designed, then ethical dilemmas in training set were used to train the proposed training model, and the model based on multi-scale mixed attention mechanism CNN was constructed. Secondly, the ethical dilemmas in testing set were used to test and verify the proposed CNN model. The results show that multi-scale CNN model has better accuracy than traditional CNN does. Furthermore, the stability of CNN model with attention mechanism is enhanced compared to the model without attention mechanism. Therefore, the multi-scale mixed attention mechanism CNN model has higher accuracy and stability, with the highest accuracy rate reaching 89%.

As a new computing model, mobile edge computing has shown great potential in the internet of vehicles scenario. Task offloading is a key problem in mobile edge computing, which is crucial to improve system performance and user experience. In this regard, the research on vehicle task offloading under mobile edge computing was reviewed. Firstly, the basic concept of mobile edge computing was introduced, and then the research on the aspects such as evaluation index, task offloading problem formula and offloading decision was summarized, and the task offloading methods and some algorithms were also summarized. Finally, the future research direction and challenges of mobile edge computing task offloading were discussed. Through the overview of the current research status and the discussion of future challenges, it is helpful to better understand the frontier dynamics of mobile edge computing task offloading, which provides guidance and inspiration for the related research and applications.

In view of the problem of bolt bending and bushing fracture at the joint caused by excessive force on the hard point of the triangle link of electric motocross under the flying slope condition, an optimization scheme of the hard point coordinates of the multi-link rear suspension system was formulated. Firstly, the whole vehicle model and the rigid-flexible coupling model of the triangular connecting rod were established. The dynamic strength response of the motorcycle on the ramp with 1m slope was simulated and analyzed, the force and stress at the hard point of the triangular connecting rod were obtained, and the accuracy of the simulation model was verified by road tests. Then, the multi-objective optimization of the multi-link rear suspension system was completed by using the NSGA-II genetic algorithm, which took the force at the hard point position of the triangular connecting rod as the optimization target and the hard point coordinates as the design variable. The results show that after optimization, the force at the junction between the triangle connecting rod and the flat fork is greatly reduced, the force at the connection between the shock absorber and the triangle connecting rod is reduced, the force at the connection between the triangle connecting rod and the tie rod is greatly reduced, and the maximum stress of the triangle connecting rod is reduced. The reliability of the optimization scheme was verified by the experiment of the sample production.

A vehicle chassis adaptive dynamics model with different sub-models was proposed to address the common problem of low computational efficiency in the simulation process of vehicle chassis dynamics. The proposed adaptive model could automatically match the chassis dynamics model with the highest computational efficiency and satisfactory solution accuracy according to different driving conditions. Using the MATLAB/Simulink simulation platform, three different models, such as the adaptives model, dynamics model with high complexity and CarSim vehicle dynamics model, were simulated, compared and analyzed under different working conditions. The simulation results under double line change, circular working condition, and comprehensive working condition show that compared with the other two models, the proposed adaptive model has higher computational efficiency while ensuring the accuracy of the results.

To investigate the influence of the opening area of rotating wheel on the flow characteristics of the wheel area and the whole vehicle model, an entire vehicle model with wheels based on the Ahmed model was constructed and the surface pressure characteristics of the vehicle body was obtained by wind tunnel tests. A simulation model of the aerodynamic characteristics of the entire vehicle considering the rotation state of the wheels was established by use of Xflow software. The impact of grid partitioning strategies on simulation results was explored and the reliability of the numerical method was verified by comparing with the pressure coefficient results of wind tunnel tests. Furthermore, the influence of different wheel-spoke opening areas on the aerodynamic drag of the entire vehicle model was explored. The influence mechanisms of variation of wheel-spoke opening area on the flow field characteristics and overall vehicle aerodynamic performance were systematically analyzed, considering vortex structures, streamline patterns and other flow field characteristics. The study shows that as the wheel-spoke opening area increases, the aerodynamic drag of the whole vehicle firstly increases and then slightly decreases, while the increase of wheel-spoke opening area leads to a simultaneous increase in aerodynamic drag of the front and rear wheels, but a decrease in the vehicle body drag.