Abstract:To study the crack resistance and width calculation method of trapezoidal wet joints in steel-ultra-high performance concrete （UHPC） bridge decks， one full-scale specimen with wet joints was designed and fabricated for a negative bending test. According to the test results， the cracks appeared at the interface of UHPC trapezoidal wet joints under the configuration of reinforcement， the nominal tensile stresses were 10.3 and 12.9 MPa at the crack widths of 0.02 and 0.05 mm at the joint， respectively， and the peak tensile stress of UHPC under the design load of the real bridge was 3.84 MPa， which showed that the wet joints with negative moments had a better performance of cracking resistance in line with the engineering requirements. Meanwhile， FE simulation and parametric analysis were carried out by introducing the damage principal and UHPC shrinkage effect. The results indicate that the UHPC shrinkage effect would lead to high tensile stress in the joint region， which would weaken its cracking resistance. The initial cracking load was more sensitive to the size of the joints， and increasing the diameter of reinforcements would have a slight improvement on the cracking resistance of the joints. The increase of the diameter from 8 mm to 16 mm only increases the cracking load by 2%. In addition， comparing the results of the tests and the formulae， it can be seen that the existing method of calculating the crack width of concrete overestimates the crack resistance of UHPC joints， and the modified formulae were better at predicting the crack width of joints.

Abstract:To address the challenges of optimization for prestressed concrete （PC） continuous rigid frame bridges， particularly the issues of local optima and the difficulty in simultaneously optimizing structural dimensions and prestressing tendons， this study proposes a method combining the artificial bee colony （ABC） algorithm with the back propagation （BP） neural network. The heuristic algorithm helps to avoid local optima. The objective function considers both structural cost and performance， with constraints to meet specifications and construction requirements， achieving systematic optimization of structural dimensions and tendons for PC continuous rigid frame bridges. Based on a continuous rigid frame bridge with a span arrangement of （95+173+95） m， a comparative analysis was conducted using nine different algorithms combined with neural networks. This comparison highlighted the superiority of the ABC algorithm in terms of optimization effect and efficiency. The optimal scheme not only meets the specifications but also reduces the objective value by 35.8%， tendon usage by 46.3%， and stress safety variance by 60.4% compared to the original bridge， showing significant optimization effects. The predicted value of the objective for the optimal scheme differed from the actual value by only 2.1%， demonstrating the effectiveness of the prediction. Furthermore， importance and sensitivity analyses were conducted to explore the impact of different parameters on the objective value.

Abstract:Various sources of uncertainty can threaten the integrity and functionality of engineering structures during their service life. Reliability analysis aims to quantitatively measure the safety of structures by estimating the probability of failure. With the development of machine learning technology， many researchers have developed surrogate model-based methods to better balance the efficiency and accuracy of reliability analysis. By defining loss functions informed by physics-related equations， such as partial differential equations or ordinary differential equations， Physics-informed Neural Networks have received much attention as universal approximators of functions. Some researchers have proposed incorporating random variables into the input layer of neural networks， thereby facilitating highly efficient reliability analysis without the need for numerical solvers， such as finite element analysis. However， existing methods typically design collocation points with uniform sampling and calculate the failure probability with Monte Carlo Simulation. The existing methods， therefore， may not effectively handle rare events. Thus， this work proposes subset simulation-based adaptive training of Physics-informed Neural Networks， along with an associated reliability analysis framework， to enhance computational efficiency. Several numerical and engineering examples demonstrate the performance of the proposed method.

Abstract:The hybrid beam cable-stayed bridge is widely used for fully utilizing the advantages of steel girders and concrete girders. The range of the interface position between the steel girder and the concrete girder significantly affects the internal force distribution of the cable-stayed bridge. Determining this interface location expeditiously remains a crucial technical challenge to bridge conceptual design. Hypothesizing the dependent cables as a membrane model， the influence factors of the sum of horizontal component forces on the bridge tower are investigated， such as tower height， span length， live load and constant load of concrete and steel girders. Based on the rational dead load state of cable-stayed bridge and the horizontal force balance of bridge tower， the calculation methods for determining the optimal range of the interface position between the steel girder and the concrete girder are derived. The feasibility and applicability of the methods are confirmed by the finite element analysis results of hybrid girder cable-stayed bridges in two references.

Abstract:The interaction mechanism between the double-layer lining of shield tunnel is crucial for the overall safety performance of the tunnel. To investigate the shear strength performance of the contact surface between double-layer linings， numerical simulation tests of the shear behavior between the segment and the second lining concrete were conducted based on the PFC^3D discrete element program. The tests were performed under different conditions of the strength of the second lining concrete， interface roughness， and normal stress. Shear stress-strain curves and the development of microscopic cracks in the segment-second lining concrete were obtained under different working conditions. According to the test results， two modes of failure were proposed which are shear-crack failure and shear-sliding failure. Local mechanical analyses of individual convex bodies on the contact surface were conducted for the two failure modes. The research results indicate that with the increase of roughness and normal stress， the crack propagation gradually shifts from the contact surface to the inside of the concrete， which means the shear failure mode transitions from shear-sliding failure to shear-crack failure. Based on the shear failure mechanism of the contact surface and considering the coupling effects of multiple factors， a calculation method for the shear strength of the double-layer lining contact surface was proposed based on the Mohr-Coulomb criterion. The research results provides a theoretical basis for determining the shear strength parameters of the interface between the primary and secondary linings in shield tunnels.

Abstract:This study is based on a deep foundation pit project in a soft soil area， conducted in-situ micro-disturbance double-liquid grouting experiments. By integrating real-time monitoring， static cone penetration tests， and on-site excavation methods， the study explores the grout diffusion range in soft soil strata. It also analyzes the fracture-compaction morphology of grout veins within the diffusion range and reveals the spatial and temporal variations in surrounding soil displacement， stress， and pore water pressure during continuous grouting along the depth of a single borehole. The findings demonstrate that the lateral displacement of the soil， pore water pressure， and soil pressure induced by grouting follow a power-law attenuation with distance， with rapid decay within 2 m. Excavation results showed that grout veins extend as far as 2.5 m， primarily exhibiting compaction filling within 0.5 m of the grouting borehole and fracturing beyond 1 m. Moreover， the measured grout volume beyond 1 m aligns with the theoretical volume expansion ratio calculated through inclinometer readings of soil deformation. These results provide a solid experimental basis and extensive data reference for grouting corrections in soft soil shield tunnels.

Abstract:This paper proposes an Embedded Discrete Fracture Model （EDFM）-based method for computing water inflow into fractured rock tunnels. The method explicitly represents rock fractures using embedded discrete fractures. By introducing flow exchange terms to couple matrix and fracture flow， it circumvents the conformity constraint between matrix and fracture grids， thereby significantly simplifying the meshing process. The validity of the method is verified by comparing its results with those obtained from the El-Tani analytical solution and COMSOL numerical simulations. Further analysis is conducted on the influence of fracture parameters—including spacing， aperture， dip angle， length， and gap—on water inflow. Finally， the method is applied to compute water inflow into the Nanwan Tunnel. The calculated maximum fracture-induced inflow is on the same order of magnitude as the reference value， demonstrating that the method is applicable for preliminary water inflow assessment in practical engineering.

Abstract:To investigate the impact of spatial variability of different mechanical parameters on the initial fracturing pressure of gas storage cavern. A thermomechanical coupling random phase field model， applicable to lined caverns for compressed air energy storage， was established based on the principles of random field theory， conservation laws of mass and energy， Fourier's law， and the variational principle. The correctness of the model was validated through experimental results and numerical simulation. Combining Monte Carlo method， a compressed air energy storage （CAES） case from literature was selected for analysis and study. The research indicates that the initiation pressure of a gas storage cavern is significantly influenced by the spatial variability of rock mass， decreasing with increasing coefficients of variation of the Young’s modulus and the critical energy release rate， and exhibiting an approximately inverse relationship under their combined variability. However， the autocorrelation distances of the Young’s modulus and critical energy release rate have a relatively minor impact. The spatial variability of Young’s modulus has a greater influence on the initiation pressure compared with the critical energy release rate.

Abstract:The prediction of parents' temporary parking needs is extremely important for the planning and control of parking spaces in front of school gates. Based on the analysis of the characteristics of temporary parking demand， the queuing theory model is used to predict the temporary parking demand of primary schools’ parents. And considering that the time-segmented school commuting is an effective measure to alleviate parking conflicts， a time-interval parking demand prediction model is established， and considering several key parameters， such as the number of students， the proportion of parents' vehicle using， the interval time， the proportion of early arrival， to provide reference for the planning and management of school parking. Finally， combined with a practical case， the application of the method is demonstrated， and the suggestions to reduce the contradiction between supply and demand of parking are put forward from the school side， which is of great significance for the alleviation of traffic congestion during school commuting hours.

Abstract:Connected and automated vehicle technologies offer new opportunities to improve the capacity of signalized intersections. This paper proposes a trajectory guidance strategy for CAVs at signalized intersections to enhance traffic efficiency and reduce travel costs. Firstly， a 0-1 integer programming model is established based on vehicle paths and potential car-following and overtaking costs， allocating the traffic flow into multiple platoons. Secondly， longitudinal and lateral trajectory control models are developed： the longitudinal model incorporates both delay and fuel consumption， while the lateral model employs a Stackelberg game theory approach to achieve cooperative vehicle platooning. Finally， the effectiveness of the proposed strategy is validated through simulations in SUMO. The results show that under a traffic volume of 350 per hour per lane， the proposed strategy reduces travel time and fuel consumption by 34.3% and 30.1%， respectively， and decreases stop time and number of stops by 38.7% and 81.9%， compared to a speed-only guidance strategy.

Abstract:The pre-crash scenarios for trucks were constructed based on crash data and real world in-vehicle data. Forty types of functional scenarios for trucks were derived， and the logical scenarios for trucks were parametrically described by combining vehicle motion parameters and driving behavior parameters. The crash causation chain for trucks were mined and the crash causations were analyzed by the Driving Reliability and Error Analysis Method （DREAM）. It was found that： Rear-end and side collision were the main trucks crash scenarios， accounting for 24.7% and 37.2%， respectively. For longitudinal collision scenarios， driver reaction time， braking effort， and following distance are important factors affecting the crashes of trucks. Differences in lateral departure distance and steering angle for longitudinal scenarios. Differences in drivers' lane departure behavior are important factors affecting the risk of crashes for trucks. Inadequate braking， late action， and too close a following distance are the key causes of longitudinal collisions. Wrong direction of travel， late braking and too high speed are the causes of lateral collisions.

Abstract:In order to better predict the short-term inbound passenger flow of urban rail transit in different scenarios， a multi-scenario inbound passenger flow prediction method based on the combined TCN-Transformer deep learning model is proposed. The method takes into account the internal characteristics of passenger flow such as temporal features and periodic factors such as date attributes. A multifactor passenger flow feature input matrix is constructed through the feature embedding layer. The Temporal Convolutional Network （TCN） and causal attention Transformer modules are used in parallel to extract and learn local and global information from passenger flow data. Then the prediction layer composed of fully connected layers outputs the prediction results. The effectiveness of the model is validated by using the passenger flow data and related information of Shanghai Stadium Station， and compared with the prediction results of several comparative models. The experimental results show that the TCN-Transformer model can better capture the characteristics of inbound passenger flow in different scenarios， and has better prediction accuracy and generalization ability. Compared with several other models， the root mean square error （R_MSE）， mean absolute error （M_AE）， mean absolute percentage error （M_APE） of our model are reduced by at least 8.42%， 7.32%， and 6.18% respectively.

Abstract:High energy consumption and poor durability faced by traditional cement-based curing agent when repairing heavy metal contaminated soil， this study prepared red mud （RM） -blast furnace slag （BFS） based geopolymers ??solidifying/stabilizing composite heavy metal contaminated soil by alkali activation. However， the stability of soil solidification will decrease under the long-term environmental influence of ground water level and acids and alkalis in rainfall. Wet-dry cycling and pH tests were conducted to evaluate the evolution of strength and leaching concentration. The mechanism of solidification and stability degradation were also elucidated. With increasing dry-wet cycles， the UCS decreased， while the leaching concentration of heavy metal rose significantly. In contrast， a higher content of RM+BFS mitigated the strength loss and suppressed the increase in heavy metal leaching. For low/medium level contaminated soils， the leaching concentrations initially decreased and then increased with the increase of pH， except for the continuous decrease of Cu（II）. However， regardless of pH， the leaching concentration is basically lower than the limit value， and the curing effect can remain stable for a long time. The hydrogels generated through the polymerization can immobilize heavy metal pollutants through physical absorption， chemical precipitation， and other mechanisms. Environmental influences lead to changes in internal pore connectivity， heavy metal adsorption sites， and occurrence forms， resulting in the degradation of stability. This study serves as a significant reference for utilization of industrial solid waste in solidifying composite heavy metal contaminated soil.

Abstract:This paper takes the Yangtze River Estuary as the research object， studies the changes in heavy metal concentrations and the sources of heavy metals in sediments during the tidal cycle， and combines the health risk assessment model and the Monte Carlo model to evaluate potential human health risks. The results show that during the tidal cycle， heavy metals in sediments have higher levels at the beginning of high tide， before and after high tide， and at the end of low tide. The heavy metals in the study area mainly come from transportation， industry， atmospheric deposition and agriculture， with contribution rates of 25.9%， 33.9%， 18.2% and 21.9%， respectively. Heavy metals in the soil in the study area have an acceptable carcinogenic risk to children and adults， and the risk level for children is greater than that for adults， and the non-carcinogenic risk can be ignored. Comprehensively considering the relationship between the distribution and source of heavy metals and health risks， it is found that agricultural pollution is the main source of heavy metals in sediments during the study period， and arsenic has the greatest carcinogenic risk.

Abstract:To address the issue of disruptions in original schedules caused by the mid-arrival and insertion of new orders in practical workshop scheduling， the uncertainty of order arrival time and quantity was considered based on the hybrid flow shop scheduling problem. A mathematical model was established with the optimization objectives of quality robustness and solution robustness. A proactive scheduling algorithm with a two-layer decision structure was designed for this model. The outer layer of the algorithm used an improved genetic algorithm framework to search for key decision variable encodings， while the inner layer formulated heuristic rules with buffer time settings to decode the outer layer chromosomes by analyzing different scenarios formed by the uncertainty of order arrival. Simulation evaluation was conducted using Monte Carlo sampling and reaction strategies， and effectiveness was verified by comparison with Gurobi and various existing template schedule generation mechanisms.

Abstract:In order to investigate the influence of consumers' emotional tendencies on automotive interior design， a perceptual evaluation method of automotive interior design based on sentiment analysis was established， and a regression model of interior design perceptual imagery was constructed. Firstly， the qualitative perceptual design elements are obtained by differentiating the interior design elements， then the quantitative perceptual scores are obtained by quantifying the consumer's review data using the Bert-BiLSTM sentiment analysis model.Finally， the mapping relationship between the qualitative perceptual design elements and the quantitative perceptual scores is constructed through the quantitative theory I model. Through validation， the prediction error of the method is 7.8%， which proves the feasibility of the method. This method can effectively explore the emotional tendency of consumers toward automobile interior design， and can provide the basis of emotional design for automobile interior designers， which has certain practical application value.