Abstract:To address the problem of numerous borders and weak border protection in metro cloud platform， the collaborative interaction between the cloud and the industrial control network is analyzed， and a dynamic trust management method for border security protection of metro cloud platform is proposed. The method consists of abnormal behavior recognition， trust evaluation， trust updating， and trust-based dynamic access control. Based on the network topology of metro cloud-based integrated supervisory control system， three kinds of abnormal control commands are simulated， i.e.， unauthorized control commands， non-conforming control commands， and interference with normal control commands. The results show that the proposed method can effectively resist abnormal control commands initiated by malicious nodes. The changes in trust values vary for different nodes and different types of misbehaviors following the rule of “slow rise and fast fall”， thus ensuring fine-grained boundary protection for the metro cloud platform.

Abstract:Railway signal equipment is essential for ensuring traffic safety and improving transportation efficiency. Strengthening the intelligence operation and maintenance of signal equipment is essential to mitigate the risks associated with railway operations. Currently， the intelligence operation and maintenance platform based on building information model （BIM） in China is unable to accurately depict the behavior and mutual feedback mechanism of each device， thus relying on experiential knowledge for inference. To address this issue， initially， the knowledge graph was constructed using the text related to the operation and maintenance of railway equipment； Subsequently， a convolutional neural networks-clique group graph convolutional neural networks （CNN-cgGCN） model was developed to process BIM image modal data and annotate the information of 20 specific railway signal equipment part drawings. The experimental results show that the accuracy of the model reaches 95.38 %， and the harmonic mean F1 of precision and recall reaches 95.58 %； Finally， BIM image information is integrated into the visual knowledge graph of operation and maintenance text. This multi-modal knowledge graph is then visualized using the Neo4j graph database， so as to accurately map the mechanism of mutual feedback between signal equipment， and offer online services and guidance to on-site railway operation and maintenance personnel， facilitating safety management and operational decision-making.

Abstract:Aiming at the demand of autonomous positioning of rail construction vehicles， this paper proposes a method of absolute position extraction of rail vehicles based on the video of 100-metre markers captured by the on-board forward-looking camera. The method first improves the YOLOX-s network and constructs the target detection model of the 100-metre marker to complete the target detection of the 100-metre marker. Then， when the 100-metre marker is detected， it combines image preprocessing with convolutional recurrent neural network （CRNN） network to construct the 100-metre marker digital text recognition model to extract the digital text information of the 100-metre marker， so as to achieve the absolute position location of rail construction vehicles. The method is verified to be able to quickly and accurately extract the absolute position information of rail construction vehicles.

Abstract:With the rapid advancement of urban rail transit systems， the existing pre-evaluation practices for safety-related domains within domestic railway transport remain immature. The inability to share accrued cases of faults and accidents expedites the challenge in executing practical pre-evaluation tasks. First， a 5M（man， machine， media， management， and mission） multi-factor signal system security knowledge based on ontology reasoning rules is constructed. Then the pre-evaluation process of the signal system is elaborated. Evolving from case analysis and practical application， a novel rail transit safety pre-evaluation system is formed relying on a safety knowledge base. Finally， the rail transit safety pre-evaluation method based on safety knowledge base is compared with the general pre-evaluation method from the aspects of risk identification quality and risk control accuracy. The results verify that this approach can identify and appraise risk factors within signal system stages， spanning its complete lifecycle， with superior quality and precision in risk identification and control. The deployment of this methodology bolsters the strength of pre-safety management effectiveness in urban rail transit and provides a reference framework for theoretical investigations on risk pre-evaluation within safety-related domains.

Abstract:As one of the most important diseases of tunnel lining， reinforcement corrosion is a serious threat to the structural durability and safe operation of tunnels in service. The reinforcement corrosion and the resulting cracks directly threaten the durability and safety of the lining. Therefore， it is very important to timely perceive the reinforcement corrosion degree of tunnel lining. The current detection method mainly relies on contact measurement， which is low in efficiency. To explore the feasibility of non-contact rapid detection， a combination of indoor experiments and numerical analysis was used to study， in detail， the heat conduction law of shallow steel corrosion. The influence of the degree of steel corrosion and the thickness of concrete protective layer on the temperature field of the lining surface under active thermal excitation conditions was analyzed. The results show that under the condition of active thermal excitation， the corrosion area of concrete surface is shown as a high temperature area in thermal image， When the corrosion rate of steel bars is 4.36 %–23.16 %， the temperature difference of concrete surface is 2.3 ℃–4.4 ℃. The temperature difference on the surface of the specimen increases with the increase of reinforcement corrosion degree， presenting a cubic function relationship. The thickness of concrete protective layer is negatively correlated with surface temperature difference. The larger the thickness of protective layer is， the smaller the temperature difference is. In summer， the heat conduction effect caused by the temperature difference between the inside and outside of the lining will occur on the lining surface at the position of reinforcement corrosion. By detecting the temperature distribution of the lining surface with infrared thermal image in combination with the apparent characteristics of the lining， the corrosion degree of reinforcement can be determined comprehensively， which provides a new method and technical means for the rapid detection of reinforcement corrosion.

Abstract:In order to study the mechanical deformation performance of prefabricated double-sided superimposed utility tunnel side wall connected by spiral stirrup sleeve bottom joint under out of plane load， two full-scale side wall specimens connected by this joint were tested under low cyclic reciprocating loading at different axial compression ratios， and compared with two cast-in-situ specimens， The effects of the bottom joint of the spiral stirrup sleeve and the vertical axial compression ratio on the seismic deformation performance of the side wall of the utility tunnel were preliminarily revealed. The test results show that the hysteretic curves of composite precast specimens and cast-in-situ specimens are relatively full， and have a good seismic energy dissipation capacity； which meet the seismic deformation design requirements of the fabricated structure； The shear deformation of the upper area of the composite assembly specimens are higher than that of the cast-in-situ specimens， and the cracks are all over the whole height range of the wall， which has good out of plane deformation ability

Abstract:Health monitoring in the field of civil engineering is of great significance to ensure the long-term and stable service of infrastructure. Compared with traditional monitoring methods， the computer vision technology based on deep learning has the advantages of high efficiency and accuracy. This paper provides a systematic review on the application of the deep learning-based computer vision technology in the field of civil engineering life cycle health monitoring. First， a scientific econometric analysis of the literature in this field is conducted with the help of literature visualization software. Then， the development process of computer vision technology is briefly described， and the methods of data acquisition， data processing， and data annotation in the process of constructing deep learning data sets are summarized. Afterwards， the development and practical engineering application value of the computer vision technology based on deep learning in safety management of construction site， local damage detection of in-service structures and overall damage assessment of structures after disaster are reviewed. Finally， the future application directions are prospected.

Abstract:To quickly and accurately simulate the evolution process of urban rainstorm waterlogging， a dynamic interaction method between underground sewer network and above surface was proposed， and a bidirectional coupling model of SWMM /LISFLOOD-FP was constructed to solve the two-way flow exchange and time synchronization problems. Taking the Waigaoqiao area of Shanghai as an example， the coupling model was calibrated and verified using two short-duration rainfall processes， and the simulation results of unidirectional and bidirectional coupling models were compared and analyzed. The results show that the bidirectional coupling model has a higher simulation accuracy and a better applicability in the study area. For the mild （<0.2 m） waterlogged area that accounts for more than 80 % of the submerged area， the ratio of simulated waterlogged area between unidirectional and bidirectional coupling models is 1.21. However， for moderate （0.2-0.3 m） and heavy （>0.3 m） waterlogged areas， the unidirectional coupling simulations tend to overestimate， and the ratios increase to 1.88 and 2.1， respectively. The bidirectional coupling model can effectively reveal the whole process of urban stormwater accumulation， diffusion and regression， which can be used for urban rainstorm waterlogging simulation and inference， and provide scientific basis for waterlogging control and disaster prevention.

Abstract:The delimitation of the boundary of the living circle is both the focus and the difficulty in the study of community living circles. There are three existing delimitation methods， administrative boundaries or regulatory planning units， service radius and accessibility of facilities， and daily activities of residents， of which the daily activities of residents are the closest to the original definition of the community living circle. However， the current application of this method mainly relies on the data of the global positioning system （GPS）， which is costly and limited in sample size， making it difficult to describe the activity patterns of most residents in a community and lacking a comprehensive understanding of a wide range of different types of community living spheres. In this paper， based on the large sample of the data of location-based services （LBS）， a complex network analysis technique is used to measure the community living circle in the central city of Chengdu （549 km²）， and by identifying its scale and boundary features， the spatial factors such as location， road network density and point of interest （POI） density on the community living circle are further explored. The results show that better network clustering results （close to the actual size of a 15-minute living circle） can be obtained by choosing carefully the unit resolution and distance constraint d-values. The size of community living circles classified based on LBS data varies widely， but most of the sizes are between 1 and 5 km². Among the three types of representative spatial elements， road network density and community living circle size， there is a moderate correlation， and location and commercial POI density are not significantly correlated with the size of the community living area. However， there exists a strong correlation between these three spatial elements and population density of community living circle. This paper reveals that community living circle measurement method based on LBS data， will deepen our understanding of the community living circle phenomenon and provide theoretical and technical support for community living circle planning.

Abstract:Considering the two measures of information release and congestion charging under uncertain environment， and based on traffic network equilibrium， an optimal toll model of the simple two-route road network is derived and is extended to the general road network. Based on the two-route network， a day-to-day route choice behavior experiment is conducted. The impact of the two measures on travelers’ route choice behavior is analyzed. The results show that the implementation of perfect information and congestion charging can both reduce the flow fluctuation， and the effect of congestion charging alone is the best. However， only when the two measures are combined， the trend of road network flow towards user equilibrium is the most stable. Perfect information will increase route switching behavior， while congestion charging can effectively inhibit large route switching. When there is no congestion toll， route switching often increases the travel cost. On the contrary， when congestion toll is implemented， route switching can often reduce the travel cost. The travel time when the two measures of congestion tolling and traffic information release are implemented together is better than that when the two measures are implemented separately.

Abstract:As the number of newly-emerged technologies is rising， the influence of security check procedures on the pedestrian transfer efficiency of railway stations and the corresponding design strategies are increasingly concerned. After the pedestrian simulations are conducted by using the MassMotion Software， the influence of arriving passengers in different parts of railway stations， and the interaction effects between external factors are compared in two scenarios： the mutual recognition of security checks and the application of face recognition systems. Based on the results above， a reference table for impacts from arriving passengers is proposed as a supporting tool to help architects respond to complex scenarios in the early design stages.

Abstract:The coordinated development of the port-city economy can enhance the competitiveness and influence of the port. This paper conducts a comparative analysis between the international shipping centers of Shanghai and Singapore. Employing the entropy weight-gray correlation model and the vector autoregressive model， it examines the interactive relationship between the port and the city， focusing on foreign trade， gross product value， and industrial structure. The findings reveal a strong correlation within the port-city economy. In Singapore， foreign trade and port exhibit a stable two-way promotion effect， whereas the export-oriented economy in Shanghai has a modest driving effect on the shipping center. The GDP of Shanghai emerges as a pivotal support for the port， although the radiating effect of the shipping center on GDP appears less pronounced. A long-term driving force is identified between tertiary industry and the port in Shanghai. The secondary industry in Shanghai exerts a predominant influence on the shipping center. Therefore， policies promoting the expansion of the tertiary industry demonstrate a more potent influence on the shipping center than those favoring the secondary industry. Based on the conclusions， it proposes some suggestions， such as bolstering port resilience and integrating port-industry-city development.

Abstract:Aimed at the problems of low structural efficiency and incongruity with the stress field of bearing parts in the filling of uniform isomorphic mesoscopic structures such as grid and triangle used in traditional 3D printing， and inspired by the fact that naturally growing woodgrain can improve the overall performance of wood， a woodgrain inspired 3D printing toolpath planning method based on image recognition is proposed aiming to learn from the fiber arrangement of wood to improve the mechanical properties of parts. Through wood milling layering， the woodgrain image of each layer is extracted in turn and the corresponding printing path is generated. Then， the biomimetic woodgrain specimen is printed layer by layer. The tensile test results show that the maximum tensile load of the printed specimen based on woodgrain is increased by 115.03 %， 72.89 % and 64.39 % respectively compared with the three traditional uniform isomorphic filling specimens，i.e.， grid filling， triangular filling，and rectilinear filling. It is proved that the non-uniform heterogeneous filling structure based on biomimetic woodgrain can significantly improve the tensile strength of the printed specimen.

Abstract:Aluminum thin-walled tube can be cracked during axial compression connection， which can result in connection failure. To improve the compressional plastic formability of aluminum alloy， a heating plastic connection tool for unequal diameter thin-walled tubes was designed in the paper. The thermal stress and heat transfer of the tool in the process of heating plastic connection were numerically simulated by finite element software. The cooling system structural parameters were analyzed by using the orthogonal test method. The tool and the trial unequal diameter aluminum joint were manufactured. The results show that the maximum stress of the tool at 20 ℃ and 600 ℃ are lower than the yielding strength of the material at the corresponding temperature， which meets the requirements of heating plastic connection process for steel and aluminum. The cooling system designed can effectively slow down the heat transfer from the tool to the tensile machine. The joint of φ44 and φ30 is successfully connected by the tool at 300 ℃， which verifies the rationality of the tool design and the feasibility of the connection process， provideing guidance for subsequent engineering application.

Abstract:Long-term fatigue driving is an important cause of accidents for operational drivers. To ensure driving safety， companies install cameras on operational vehicles to collect drivers’ facial videos， automatically identify the drivers’ fatigue state based on a fatigue detection model， and use voice reminders or even enable remote escort to prevent fatigue. Most of the existing fatigue detection research is based on the extraction of the key points of drivers’ faces， which has high requirements for video quality. However， in the real commercial vehicle environment， the detection of key points easily fails due to the poor light at night， the imperfect position of the camera and the obscured face of the drivers， thus affecting the accuracy of the model. Therefore， this paper proposes an end-to-end fatigue detection model for operational drivers with a high robustness， based on convolutional neural network （CNN） and long short-term memory neural network （LSTM）. The model takes the drivers' facial videos collected by cameras as input， and uses the CNN network to extract the single-frame features of the videos. On this basis， the temporal single-frame features are used as the input of the LSTM network to finally identify the drivers’ fatigue state. The experimental results show that the area under curve （AUC） of the model is 0.9， which is much superior to existing models based on facial key points. In addition， in order to improve the robustness of the model in the actual driving environment， data augmentation is applied to the training data， simulating both light and camera changes. The accuracy and robustness of the model are further improved through model retraining. Before the improvement， the AUC of the model in the actual driving environment for commercial vehicles is reduced by 37.3 %， compared with the laboratory model， However， after the improvement， the AUC is only reduced by 9.7 % which indicates that the robustness of the model is improved and the model can better adapt to the complex naturalistic driving environment for commercial vehicles.

Abstract:Aiming at the problem of blurred edges in salient object detection， this paper proposes a new method that can fully utilize edge information to enhance the confidence of edge pixels. First， the triple attention module is introduced， which uses the characteristics of the predicted saliency map to directly generate foreground， background and edge attention， and the process of generating attention weights does not add any parameters. Next， the edge prediction module is introduced， which performs supervised edge prediction in the shallowest layer of the network with the biggest feature map， and fuses the predicted edge with the saliency map to refine the edges. Finally， the model is qualitatively and is quantitatively evaluated on six commonly used public datasets， and fully compared with other models， which proves that the proposed model can achieve the best results. The method proposed in this paper has 30.28 M parameters， and can predict saliency maps at 31 frames per second on GTX 1080 Ti graphics card.

Abstract:Coastal wetlands are an important part of the “blue carbon ecosystem”. The spatiotemporal distribution and content of soil organic carbon of the coastal wetlands in the Yangtze River Estuary， the carbon sequestration rate， the lateral input and output fluxes of organic carbon and the quantification methods， and organic carbon circulation quantitative analysis model are summarized. The dynamic response of organic carbon storage and composition to different influencing factors is analyzed. It is found that the horizontal distribution of soil organic carbon ranks as Chongxi Wetland > Chongming Dongtan > Jiuduansha > Nanhui tidal flat； Organic carbon flux and concentration changes are mainly affected by water and soil physicochemical properties, terrestrial inputs and tidal dynamic, porewater exchange, human activities and global climate change. In the future， the uniform observation of the soil carbon pools and organic carbon transportation in wetlands of the Yangtze River Estuary should be strengthened， to accurately quantify the contribution of the main factors to organic carbon， which is of great significance to the study of the carbon circulation mechanism and carbon sink assessment of saltmarsh wetlands.

Abstract:To address the problem of reliability and risk assessment of intelligent manufacturing systems， an improved failure mode and effects analysis （FMEA） model for intelligent manufacturing systems is proposed for the reliability and risk evaluation of intelligent manufacturing systems. The FMEA model is improved from the perspective of innovative use of combined weight， the idea of technique for order preference by similarity to ideal solution and fuzzy vlsekriterijumska optimizacija i kompromisno resenje. In the calculation of the weights of decision makers， the weights of decision makers are obtained based on the idea of technique for order preference by similarity to ideal solution， in which the fuzzy analytic hierarchy process is used to calculate the subjective weight of risk factors and the entropy weight method is used to calculate the objective weight of the risk factors， which reduces the subjectivity. The results obtained from the improved model are analyzed by applying partition around medoids （PAM） clustering algorithm and applied to the risk evaluation of intelligent manufacturing systems， The importance of each failure mode in the intelligent manufacturing systems is determined， and the effectiveness of the improved model is verified by comparative analysis.