Abstract:A method of generating adversarial examples against object detectors was proposed for object detection system in vehicle environment perception scenarios. The method achieves white-box adversarial attacks on object detectors， i.e.， object invisible attacks and object targeted mis-detectable attacks. On the Rail dataset and Cityscapes dataset， experimental results indicate that the method has good performance on the object invisible attacks and the object targeted mis-detectable attacks in the process of YOLO object detection.

Abstract:By using the features of each level in convolutional neural network， a clothing image parsing method based on multi-scale fusion enhancement was proposed. Through the fusion enhancement module， the semantic information and the features in different scales were effectively fused with the consideration of global information. The results show that the average F1 score on the Fashion Clothing test set reaches 60.57%， and the mean intersection over union（MIoU） on the Look Into Person（LIP） validation set reaches 54.93%. The method can effectively improve the accuracy of clothing image parsing.

Abstract:In order to realize the intercommunication between different data management systems， we proposed a framework of multi-source heterogeneous data governance based on semi-supervised learning. Then， we designed， implemented and tested an automatic alignment method of unstructured data and structured data. The named entity recognition （NER） technology was firstly employed in the framework to convert the unstructured data into the structured one， and the string-similarity-based method and supervised-learning-based method were respectively used for the schema matching of structured data. With the semi-supervised learning method， the structured data and its corresponding entity in database were matched and integrated. Finally， natural language processing （NLP） technology and deep learning methods were used to impute missing values in the integrated dataset. It is shown that the F1-scores for the alignment on the paper dataset and video metadata set are 89.70% and 96.50%， respectively； and that the accuracy of missing value imputation on different attributes is all above 78%， which is a great improvement compared with the baseline methods.

Abstract:The implementation of DeepTRE was improved to adapt to large-scale dataset scenarios， which greatly reduces the space complexity of DeepTRE when retaining the excellent verification ability of DeepTRE. The improved DeepTRE was evaluated in the high-speed rail operating environment recognition scenarios and was compared with other mainstream verification tools， i.e.，DLV and SafeCV. The experimental results show that the memory usage of the improved DeepTRE tool is significantly lower than that of the original DeepTRE tool. Compared with other neural network verification tools， the improved DeepTRE tool has better verification effect on the premise of faster verification speed.

Abstract:Based on the assumption of no bending moment， taking the node coordinates as the parameters to be determined， a numerical algorithm suitable for the form-finding of cable-net，i.e.，Euler coordinates form-finding method， was proposed to establish the Euler space coordinate system. Taking the axial force and rod length as unknown parameters， the equilibrium iteration matrix of the structural node coordinates and axial force was directly established according to the mechanical properties and balance conditions of the rod element. A case study shows that the Euler coordinates form-finding method can adapt to various types of cable net structure optimization， with good robustness and stable convergence.

Abstract:Three specimens of steel-concrete connection with different structures were designed to analyze the effects of connectors and back bearing plates on force transmission in the connection. It is shown that under 1.80 times of the design load， the connection in practical engineering is in the elastic state； when the axial force is 4.65 times of the design load and the shear force is 1.80 times of the design load， the connection reaches the elastic limit state； when the axial force is 6.65 times of the design load， the concrete inside the steel cells in the connection collapses and reaches the failure state. The back bearing plate is the most important component for axial force transmission in the structure. The back bearing plate bears 50%~60% of the axial force in the elastic stage， while the back bearing plate bears 35%~42% of the axial force in the plastic phase. In the mixed joint mode of perfobond plate connectors and stud connectors， the stud connectors played a minor role. The connection and force transmission between the steel box plates and the filled concrete could be realized by only setting the perfobond plate connectors.

Abstract:In this paper， through the full-scale experiment of high strength steel flange-welded web-bolted connections， the mechanical properties of Q690 and Q960 connections at ambient temperature and elevated temperatures are studied. The theoretical values of bearing capacity recommended by the existing American， European and Chinese codes are compared with the experimental values to verify the effectiveness of the codes for the high strength steel flange-welded web-bolted connections. The recommendations for the design of the high strength steel flange-welded web-bolted connections are given. Finally， the difference of fire resistance between the high strength steel endplate connections and the high strength steel flange-welded web-bolted connections is analyzed. It is shown that the high strength steel flange-welded web-bolted connections are better than the high strength steel endplate connections in terms of fire resistance.

Abstract:A series of model tests was carried out on silt in the East China Sea to study the load-lateral rotation response charateristics and analyze the relationship between lateral stiffness and lateral rotation of a suction caisson subjected to cyclic loading. The accumulated rotation variation of the suction caisson with the number of loading cycles during the service life of offshore wind turbines was discussed. The test results show that the deformation of the silt around the suction caisson has an increasing trend under cyclic loading and its stiffness decreases with the increase of the number of cycles. The rotation point has an obvious upward trend whether the suction caisson is subjected to loading or unloading， which will accelerate the overturning of the suction caisson. The relationship between the unloading stiffness and the number of cycles follows a logarithmic increase.

Abstract:The concept of “Map of Urban Vibration Pollution” is proposed in this paper， which is used to demonstrate the scope and scale of ambient vibration pollution in a quantitative and visual approach. Apps of online data collecting are programed， and such data are acquired through intelligent portable devices， thus forming a cloud database. After data cleaning， examples are given at the national， urban， and regional levels. It is shown that the “Map of Urban Vibration Pollution” is feasible and useful to support the operation and maintenance of an intelligent city.

Abstract:Based on the monthly average discharge calculated from measured runoff from 1950 to 2018 at the Datong Station in the Yangtze River Estuary， the two-dimensional hydrodynamic and salinity transport model was established by MIKE21 to investigate the effect of runoff change on the position and intensity of horizontal salinity fronts in the four branches of the Yangtze River Estuary after the Three Gorges Project was completed. The numerical results show that： the annual average salinity increases in the mouth of the Yangtze River Estuary and the increase is the most obvious in the South Passage. Except the North Channel， all salinity fronts have a double frontal structure. There is a linear relationship between the position of the salinity front and the runoff， and the change of position at the first front is more obvious than that at the second front. The relationship between the intensity of each front and the runoff is a quadratic function. The coefficient of the quadratic term is positive at the first front and negative at the second front.

Abstract:The traditional indexes test， multiple stress creep and recovery （MSCR） test， amplitude sweep and frequency sweep test were conducted to study the rejuvenation effect of fog seal material on aging asphalt. Then， the freeze-thaw splitting test and Cantanbro stripping test of Marshall specimens soaked in fog seal materials were also conducted to explore the effect of fog seal materials on the durability of asphalt mixture， respectively. It is shown that the waterborne acrylate and waterborne polyurethane play a synergistic enhanced effect and improve the elastic properties of asphalt. The fog seal materials with carbamate and carboxyl polar groups enhance water stability and anti-loosening performance of asphalt mixture significantly. According to the significance testing and grey relational analysis， the regeneration indexes of the fog seal materials present a strong consistency.

Abstract:Taking the superconducting electrodynamic maglev train as the research object， a field-circuit-motion coupling numerical model of a bogie and levitation/guidance coil was established. Then， the model was solved and its correctness was verified. Finally，the levitation force and guidance force of a bogie under different vertical displacements and lateral displacements were analyzed and the relationship between non-cross connection and cross-connection figure-eight-shaped coils was revealed based on the model. The results show that the force generated by no-cross connection figure-eight-shaped coils is mainly the levitation force， and the force generated by cross connection figure-eight-shaped coils is mainly the guidance force. With the increase of lateral displacement， the guidance-drag ratio increases first and then decreases. With the increase of vertical displacement， the levitation-drag ratio also increases first and then decreases.

Abstract:In this paper， the Miller cycle process of hybrid engine is mainly studied. A one-dimensional hydrodynamic model of intake camshaft with 130° wrap angle is established. The effect of Miller cycle realized by independent and joint adjustment of inlet valve opening （IVO） and exhaust valve closing （EVC） on engine performance is simulated and analyzed. It is shown that with the deepening of Miller cycle， under low load conditions，the intake exhaust gas reabsorption mass increases， the volumetric coefficient and pump mean effective pressure （PMEP） decreases， and the brake specific fuel consumption （BSFC） decreases； under medium load condition， the volumetric coefficient increases， the pump mean effective pressure decreases significantly，the crankshaft angle corresponding to 50% heat release rate （CA50） decreases， and the brake specific fuel consumption decreases； under high load conditions，the brake specific fuel consumption increases sharply. The combined regulation of IVO and EVC can further improve the engine performance and increase the thermal efficiency at the best fuel consumption point by 0.43%.

Abstract:Taking the aerodynamic sound source generated by the airflow flowing through the solid surface as the research object， based on the acoustic wave equation’s solution of dipole sound source corresponding to the force point source， according to both the relationship between the pressure of the sound source and the pulsation force and the relationship between the pulsation force and the pressure gradient in the flow field， the relationship between the pressure of the dipole sound source and the pressure gradient in the flow field was established. Based on the concept that two dipole sound sources can form a quadrupole sound source， the relationship between the pressure of quadrupole sound source and the pulsation velocity of flow field was also established. Finally， taking the flow around cylinder as the research object， the strength and distribution of the dipole and quadrupole sound sources of the flow around cylinder were obtained with a combination of the above two equations and the numerical simulation method. It is shown that the pressure of the dipole sound source to the far field is determined by ??p?t， and the pressure of the quadrupole sound source to the far field is determined by ?2ur2?t2.

Abstract:In this paper， the China’s first global ocean climate data records （CDRs） are used to analyze and predict the sea level changes in the Yellow Sea with obvious seasonal changes. Based on the singular spectrum analysis （SSA）， the time and spatio-temporal series of sea level anomalies （SLAs） in the Yellow Sea are decomposed and de-noised. Then the SSA-long short-term memory （LSTM） network （SSA-LSTM combined model） is established to predict the sea level trends of the Yellow Sea. Compared with the traditional methods， the prediction accuracy of the SSA-LSTM combined model is significantly improved with 35.04 mm of the minimum root-mean-square error for the SLAs time series prediction length of 5 years. For the first-year prediction of spatial-temporal series of SLAs， the minimum root-mean-square error is only 19.68 mm. The law of spatial-temporal differentiation of the sea level change in the Yellow Sea is also analyzed by the spatial modes. It is found that the sea level trend of the Yellow Sea is highly consistent and significantly related to the season and latitude. According to the SSA-LSTM combined model， the sea level rise rate of the Yellow Sea will remain at 3.65±0.79 mm per year from 2016 to 2025.

Abstract:In this paper，the Italian domination number of direct product of paths P_n×P_m is studied. With the combination of computer construction proof and mathematical derivation proof， the exact values of the Italian domination number of P_n×P₁， P_n×P₂ and P_n×P₃ are determined. For P_n×P_m （m≥4）， a bound on the Italian domination number is presented.

Abstract:The Moran’s I index was used to conduct the spatial autocorrelation analysis of innovation output in 30 provincial adminstrative region. Then， the impact of different factors on regional innovation output was explored using the geographically weight regression （GWR） model. It is shown that regional innovation output has the characteristics of positive corelation. Jiangsu， Zhejiang， and Shanghai have become the typical “hot spots” of regional innovation，and promote the development of surrounding areas；the effect of the factors，i.e.，R&D full-time equivalent， R&D input intensity， the number of college students per 100 000 persons and the number of mobile Internet users， on innovation output is significantly different in different areas.