Application of spiral stirrups can obviously improve the axial bearing capacity and ductility of reinforced concrete (RC) columns. This paper aims to study the mechanical properties of rectangular short RC columns confined with five-spiral stirrups under axial compression and to present a calculation method for the axial compression bearing capacity by finite element analysis of a number of concrete short column specimens under axial compressive loading. Firstly, a finite element model is established on the basis of experiments available in the literature, and the finite element analytical results are compared with experimental values to verify the correctness of the finite element model. Subsequently, the finite element model is used in a parametric analysis to study the likely influence of concrete strength on the axial bearing capacity and ductility of rectangular short columns with 4 different stirrup configurations which are designed in a principle of equal material consumption. Finally, through analysis of influencing factors for the rectangular short RC columns confined with five-spiral stirrups, a calculation method of axial compression bearing capacity based on volume-stirrup ratio is proposed. The analysis results show that compared with those of RC columns with five-hoop stirrups, rectangular stirrups and rectangular-spiral stirrups, the average bearing capacity of the RC columns with the five-spiral stirrups is increased by 0.78%, 6.70%, and 13.73%, respectively; and the average ductility coefficient is increased by 2.00%, 10.32%, and 10.41%, respectively. It demonstrats that the five-spiral stirrup columns have higher bearing capacity and ductility. In addition, compared to the formulas recommended by codes of different countries, the calculation method of axial compression bearing capacity proposed in this paper is relatively simple, and the average error between calculation and experimental values is only 2.83%.

In order to study the applicability of steel slag fine aggregate in concrete and the Axial compression constitutive model of steel slag concrete, the physical and chemical properties of steel slag fine aggregate were tested and the stability of steel slag fine aggregate was tested and analyzed. Six groups of steel slag concrete cube and prism specimens with different steel slag fine aggregate content were prepared, and the uniaxial compression performance of steel slag concrete was studied. The results show that the free calcium oxide and autoclaved pulverization rate of steel slag selected in this paper meet the requirements of relevant specifications and are suitable for concrete aggregate. The brittle characteristics of steel slag fine aggregate concrete are more obvious, which is significantly improved than ordinary concrete. The ratio of prism compressive strength of steel slag concrete to cube compressive strength is 0.80−0.86. The constitutive relationship of steel slag concrete can be described in sections according to the Carreira and Chu model and Wee model. The sectional constitutive model is basically consistent with the measured stress-strain curve of steel slag fine aggregate concrete.

In order to reveal the time-varying laws of the axial compressive strength, elastic modulus and splitting tensile strength of the pumped self-compacting concrete (SCC) in super high-rise buildings during the construction period, and to provide a basis for mechanical performance analysis of super high-rise buildings in construction stage, 120 specimens, including 96 cylindrical specimens and 24 cube specimens, were fabricated by pumping the SCC of a high-rise building over 400 m, their mechanical properties were tested at different ages, and the stress-strain curves of the ultra-high pumped SCC were obtained. Based on the tested results, formulas for calculating the time-varying relationship of the mechanical properties were proposed. The results show that the ultra-high pumped SCC has good compactness under vibration-free conditions. With the increase of age, the peak stress of the ultra-high pumped SCC increases, and the peak strain is significantly greater than that of ordinary concrete. The period of early 14 d is the key stage for the growth of various properties; the elastic modulus tends to stabilize after 90 d, while the axial compressive strength and split tensile strength still increase significantly after 28 d. When the age T ≤ 60 d, the axial compressive stiffness of the ultra-high pumped SCC increases with time, while the relative toughness decreases. When T > 60 d, both of them change little and tend to be stable. The improvement of strength of the pumped SCC in super high-rise buildings can accelerate the development of its early performance, and increase the axial compressive stiffness and relative toughness. The proposed calculation formulas of time-varying mechanical properties can provide a reliable basis for prediction and evaluation of mechanical performance of the pumped SCC in super high-rise buildings.

The popular digital aggregate modeling technology has low efficiency and quality, with the parameters uncontrollable and not involving the shape, edges, texture and other geometric morphological parameters of the aggregate. Thus it is difficult to effectively study the effects of geometric-morphology parameters on the comprehensive performance of particle composites at a micro level. In view of this shortcoming, firstly, the evaluation method of the geometric morphology characteristics of a single aggregate is explored and a mathematical method is presented to evaluate the micro-texture of the aggregate surface, as well as the aggregate system. Secondly, with 3D Max, a novel digital model design method for a single aggregate is proposed to create a digital model with irregular shapes, disordered edges and corners, and fine surface textures. Finally, the particle-replacement method is used to create a digital model of particle composites with PFC 3D. Then the porosity difference between the physical and the digital models is analyzed and the method to solve the porosity difference is provided. On this basis, the influence of aggregate geometry on the peak compressive strength of particle composites is studied through uniaxial compression experiments. The results show that, 1) the mathematical evaluation method of aggregate texture can quantify the microstructure of the aggregate, and the mathematical model of aggregate system evaluation expands the geometric morphology evaluation indexes of the aggregate; 2) there is a big difference in the porosity between the physical and digital models of particle composites; 3) the geometric characteristics of the aggregate can improve the occlusal interlocking effect between aggregates, and the replacement of spherical particles with particle size ≥ 2.36 mm by irregular particles can increase the peak compressive strength of the composite by 20.7%.

Basalt fiber reinforced polymer (BFRP) has the advantages of light weight, high strength and good durability. Using the material as anchor can effectively solve the corrosion problem of traditional steel bar anchor and has a broad application prospect in engineering construction in harsh environment. Taking the collapse alluvial mixed soil which widely exists in the southwest mountainous area as the object, through the indoor pull-out test, the effects of anchor type, anchor diameter, anchorage length and grouting diameter on the ultimate pullout load and interface shear stress are studied. The failure mode and stress distribution law of the anchoring system are analyzed. The results show that the failure mode of BFRP anchor in mixed soil is shear failure along the interface between grouting body and soil, and the pull-out bearing capacity of BFRP anchor is basically the same as that of reinforced anchor, so BFRP anchor can be used to replace reinforced anchor directly in practical engineering. The pullout load-displacement curve of BFRP anchor is in the form of three stages, and the elastic critical load is 20%−28% of the ultimate load. Under the test condition, the ultimate bearing capacity of the bolt is proportional to the anchoring length and the diameter of the grout. And the circumferential crack in the grouting body makes the axial stress of the bolt in the form of a single peak and reduces the stress concentration in the front of the anchoring section. The larger the diameter of the grouting body in the mixed soil is, the lower the interface strength is, the diameter increases from 90 mm to 110 mm, and the interface strength decreases by about 8%.

Clay fines from subgrade would gradually intrude into the ballast layer under cyclic loadings of passing trains, which would reduce the bearing capacity and impede the free drainage of track beds. A series of large-scale direct shear tests were carried out to investigate the strength and deformation characteristics and stress-dilatancy relationship of the geogrid-reinforced and unreinforced ballast contaminated by clay fines. The results showed that the strength and normal displacement of ballast aggregate decrease with an increase in the contamination level. The stress ratio of clean ballast is linear with the dilatancy ratio, while the addition of clay fines would increase the plasticity of the ballast aggregate. For fouled ballast in the peak state of shear stress, the dilatancy ratio of aggregate increases while the shear strength decreases, and a second-order polynomial relationship between stress ratio and dilatancy ratio is observed. Under a higher normal pressure, the aggregates have a lower dilatancy ratio. The reduction in the dilatancy rate and the shear strength of clay-contaminated ballast can be remedied by the inclusion of geogrid in the aggregate.

In order to improve the seismic safety of immersion joints, a new type of damping device for immersed tube tunnel joints was designed, and two groups of quasi-static tests were performed using a 1/4 scale shear key model with and without the damping device for comparative analysis. Through the tests, the mechanical behavior and seismic performance of the immersion joints under horizontal cyclic loading were revealed, and the feasibility of applying the new damping device in the vibration reduction of the immersed tunnel joint was verified. The results show that under the cyclic shear load, cracks first appear at the groove end of the traditional joint model, and then appear at the end of the shear key. As the loading displacement increases, the shear key experiences a large plastic deformation and then fails. Under the cyclic shear load, the damping device first suffers a local buckling, and the shear key is then subject to a shear failure. The damping device can delay the cracking time of the shear key. Compared with the traditional model without damping equipment, the cracking load, yield load, peak load and failure load of the damping model are increased by about 45.2%, 37.33%, 26.8% and 29.2%, respectively, under the same loading displacement. Meanwhile, the influence of the damping device on the initial stiffness of the model is relatively small, and it can meet the allowable displacement of the joint specified by the relevant code. The single-loop hysteretic energy consumption and the cumulative hysteretic energy consumption can be increased by 55.1% and 31.9%, respectively. Overall, the damping device can effectively improve the seismic performance of the shear key.

In order to improve the efficiency of the coupled train-bridge system reliability analysis, a train-bridge coupling vibration model is established, and the track irregularity is simulated by auto-regressive (AR) method. After basic principles of the (auto-regressive moving average exogenous (ARMAX) model are reviewed, the reliability analysis framework of the train-bridge coupling system based on the ARMAX surrogate model is proposed. The acceleration responses predicted by the surrogate model are then compared with those by the direct Monte Carlo method to examine the calculation accuracy and reliability analysis efficiency of the surrogate model in analyzing driving safety. The results show that the efficiency of the surrogate model in predicting the vertical and lateral acceleration responses of trains is significantly higher than that of the MCS method, by about 3 orders of magnitude; besides, acceptable solution accuracy of 98.66% and 86.55% can be achieved for vertical and horizontal car body acceleration prediction, respectively. Therefore, the surrogate model can significantly improve the reliability analysis efficiency of coupled train-bridge systems.

In order to improve the flutter performance of a long-span suspension bridge with a double-deck truss girder, the Yangsigang Yangtze River Bridge with a main span of 1700 m was taken as the engineering prototype to conduct section model wind tunnel tests, to study the effects of the upper central stabilizers, lower stabilizers, horizontal flaps and their combinations on the flutter performance of the bridge girder. Then, the effective aerodynamic measures were combined with the truss girder components to reduce the adverse effects of traditional aerodynamic measures. Finally, for the optimal aerodynamic scheme, the influence of damping ratio on the flutter performance of the optimized bridge girder was investigated. The results show that the single-degree-of-freedom torsional soft flutter with no evident divergent point has occurred to the bridge girder in the original design at attack angles of 0° and +3°, and the corresponding critical flutter wind speeds are 50.5 m/s and 31.2 m/s, respectively. The upper central stabilizers installed on the upper deck, the lower stabilizers installed below the lower deck, and the horizontal flaps installed at the level of the bottom of sidewalk can improve the flutter performance of the double-deck truss girder to varying degrees. The critical flutter wind speed of the main girder can be increased by over 34% by combining the horizontal flaps and lower stabilizers installed at the lower deck. On this basis, an optimal aerodynamic scheme is proposed to broaden the upper bracket and sidewalk plate, and combine the lower stabilizer with the track of the maintenance vehicle. Meanwhile, it is found that the critical flutter wind speed of the main girder can increase by 11.9% when the system torsional damping ratio increases from 0.37% to 0.52%. This indicates that the dampers may be efficient in suppressing the soft flutter of bridges with single-degree-of-freedom torsional vibrations.

Under the operation conditions of mixed passenger and freight railways as well as heavy haul railways, wheel/rail wear problem is particularly prominent. In order to slow down the growth of wheel/rail wear, the 60 kg/m rail profile is optimized. The objective function and constraint conditions are determined with the principle of optimizing the conformal degree of wheel-rail profiles under different contact conditions. Then, a nonlinear optimization model of the rail profile is established. The sequential quadratic programming method is adopted to solve this optimization model, and an optimization scheme of 60 kg/m rail profile is proposed. The effect of the optimized profile is compared and analyzed from the perspective of wheel-rail contact geometry, vehicle-track system dynamic interaction, and wear. The results show that: 1) the proposed optimized profile of 60 kg/m rail reduces the objective function value by 50%, compared with the original profile, and has higher conformal level with the LM wheel profile. 2) The distribution of wheel-rail contact points is more uniformly distributed on the optimized profile. The rolling radius difference is smaller when there is a small wheelset lateral displacement, and larger when there is a large wheelset lateral displacement. 3) The optimized profile has no significant influence on vehicle safety and comfort, and can effectively increase the wheel-rail contact area by 11.24%, reduce the contact stress by 20.42%, and slow down the occurrence and growth rate of wheel-rail wear.

The midpoint chord method can effectively control the track irregularity of the designated band that affects the driving safety and comfort. It is mainly used to measure the track static irregularity. However, its low measurement efficiency restricts the development of track ‘state-maintenance’. To solve the above problems, the track dynamic irregularity is output according to the midpoint chord. The correlation between the dynamic and static chord measured values with the chord length and the irregularity wavelength is analyzed. The dynamic track irregularity is outputed according to the midpoint chord measurement. A dynamic chord measurement method is proposed, that can evaluate the dynamic smoothness of the track, and studies the mapping relationship between dynamic irregularity and static irregularity. The results show that, the dynamic high-pass filter amplitudes of 42 m and 70 m are equivalent to the measured values of 10 m chord and 20 m chord respectively. When the irregularity wavelength is greater than 70 m, the 120 m dynamic high-pass filter amplitude basically corresponds to the variation law of 40 m chord measured value. The track dynamic irregularities with cut-off wavelengths of 42, 70 m and 120 m have the best correlation with the dynamic chord measurement waveforms with chord lengths of 20, 30–40 m and 30–60 m respectively. The maximum reasonable chord lengths of the dynamic chord measurement method are 20, 30 m and 40 m respectively. The adaptability of the dynamic chord measurement method has been verified by the measured data of subgrade and simply supported beam sections. In the subgrade settlement section, when the chord length is 60 m, the place where the static chord measurement value deviates significantly from the dynamic chord measurement value in the negative direction is the settlement point, and the places where the adjacent two sides deviate from the dynamic chord measurement value in the positive direction are the beginning and end of the settlement section.

To investigate the influence of variable switch rail cross-sections on wheel-rail contact behaviors and the wear distribution of the curved switch rail, a three dimensional (3D) asymmetric contact geometry method for the turnout area was proposed, which could calculate the real normal gap between the wheel and curved switch rail. The vehicle-turnout multi-body dynamics model was initially built by SIMPACK to obtain simulated results. And then the wear depth of the curved switch rail was calculated by the contact model considering variable cross-sections and the USFD wear model proposed by University of Sheffield. The results show that: 1) Taking the S1002CN wheel profile and No. 12 curved switch rail for examples, both the wheel-set yaw angle and variable cross-sections result in the asymmetric distribution of wheel-rail normal gap along the longitudinal direction within the contact patch. Therefore, the normal gap causes the contact patch shape and stress distribution asymmetric along the longitudinal direction within the contact patch. When the wheel-set yaw angle is 10 mrad and the lateral displacement is 7.5 mm, the area of contact patch obtained by the proposed method is 9.2% larger than that solved by the simplified method without considering the variable cross-sections and yaw angle. 2) Taking the CRH3 vehicle and No.12 curved switch blades as the research objects, the maximum wear depth calculated by the simplified method is 0.75 times as large as that according to the proposed method.

In order to reasonably optimize and match the suspension parameters to improve the dynamic performance of high-speed locomotives, the pseudo-excitation method was used to calculate the lateral riding quality index in the frequency domain for a high-speed locomotive, and a collaborative multi-parameter optimization method for the key suspension parameters was proposed considering the multi-objective performance of lateral riding quality in the frequency domain and lateral stability. Taking the operational scenarios as examples in which two yaw damper layouts and three wheel-rail contact conditions were considered, the improvement effect of this method on the lateral dynamic performance of the locomotive was illustrated. The results show that the primary hunting stability of the locomotive is poor in the low equivalent conicity condition. The lateral riding quality of the rear cab is significantly deteriorated, especially when the skewed symmetrical arrangement of the yaw damper is adopted. For the case of low equivalent conicity, it's necessary to regard improving locomotive lateral stability as the optimization objective, while for the case of high equivalent conicity, more attention should be paid to lateral riding quality. In order to give consideration to the dynamic performance of the locomotive under different wheel-rail contact states, thus improving the adaptability to track lines, the values of primary longitudinal stiffness, yaw damper damping and secondary lateral damping should be designed as small as possible in the given optimization range, it is recommended to choose them as 12 kN/mm, 600 kN·s/m, and 25 kN·s/m, respectively.

The post-construction settlement prediction of loess high fill site is an important basis for determining the time sequence of construction and the spatial layout of ground engineering. In order to accurately predict the post-construction settlement of loess high fill site, two post-construction settlement predictive models, one is a convergent model and the other is a divergent one, were proposed based on the characteristics of settlement data and the evolution laws of settlement curves of two typical loess high fill sites. The basic properties of the models, the solution to the parameters of the models, and the applicability of the proposed models in predicting the post-construction settlement of typical loess high fill site were all presented in detail.The results show that the proposed models are suitable for post-construction settlement prediction of loess high fill site due to its smaller fitting error and prediction error. Moreover, it is found that the divergent model is more suitable to the grounds whose settlement curves have S-shape, the mean absolute percentage error (MAPE) is 4.6%, which is 78.7%−95.8% lower than the prediction error of the traditional prediction models, while the convergence model is more applicable to the grounds whose settlement curves have J-shape, the mean absolute percentage error (MAPE) is 1.9%, which is 68.3%−84.4% lower than the prediction error of the traditional prediction models. Due to their good adaptability, generality and stability, the proposed models can provide more choices and references for the prediction and evaluation of post-construction settlement of loess high fill site in the future.

To improve the initial density of a paving mixture behind the paver, the research of the dynamic characteristics of tamper is needed. In the impact compaction, the vibration shock force will impact mixture. A simulation model reflecting the relationship between paving material compactness and vibration parameters, system parameters was established to analyze the relationship between the compactness and dynamic response of the tamper, and reveal compaction mechanism of tamper. By testing the paving compaction effect of the tamper on the mixture, the paving compactness-frequency curve was obtained. Through analysis of the model with different damping coefficients, the best vibration frequency corresponding to maximum density peak could be determined. Meanwhile the influence of vibration frequency on paving density was analyzed through simulation and experiment. The results show that, when the frequency ratio is less than the best vibration frequency ratio, mixture compactness is sensitive to vibration frequency, and the compactness changes gently with increasing frequency when vibration frequency is larger than the best vibration frequency ratio, so the simulation results are basically consistent with the experimental results. When the vibrating frequency is close to the best vibrating frequency, besides the steady-state response with the same excitation frequency, there is a harmonic vibration component whose frequency is equal to 2 times the excitation frequency. When the frequency ratio is greater than 0.45, the high compacting effect would be obtained for the paving mixture. The research results provide the basis for pavement effective compaction.

In order to study the influence of openings of door, window and roof on the tsunami force on low-rise houses, physical model experiments were conducted under different wave heights of tsunami bores which were generated by dam-break method. The influence mechanisms and rules of the opening rate and opening position were investigated, based on which, the influence coefficients of the opening rate and position are proposed. Results show that an opening on walls makes the peak value of the horizontal tsunami force occur at the fluctuation stage. The occurrence time is delayed and the peak value of the horizontal tsunami force becomes smaller when the opening rate of the walls increases. Also when the opening is closer to the high-energy zone of tsunami bore, the horizontal tsunami force becomes smaller. The openings on the roof can release the entrapped air, thus reducing the pressure on the roof bottom and finally reducing the positive vertical force in the quasi-stable stage. However, the openings on the roof can increase the horizontal tsunami force on structure by 20%, which should receive attentions.

Ethanol pool flame experiments disturbed with transverse low-frequency acoustic wave were carried out to understand the mechanism of acoustic fire suppression and flame dynamics under acoustic disturbance. The acoustic frequency range was 28–54 Hz, and the local acoustic pressure range at the flame was 0.10–1.25 Pa. The basic acoustic parameters, phenomenological characteristics of flame, flame height and width, and flame periodic pulsation were explored with the changing acoustic duct length and distance between acoustic duct and flame. The relation model of flame width and flame height coupled with acoustic parameters was established. The results show that, compared with free flame, the lower acoustic pressure disturbance makes the flame shape and time series more stable, and the larger acoustic pressure disturbance makes the flame more unstable. With increasing Reynolds number locally, the relative flame height is suppressed by acoustic wave and declined, and the flame width changes from being compressed to being lengthened. In addition, lower acoustic pressure will modulate flame to stable periodicity and regular phase. Higher acoustic pressure will disturb flame periodicity, resulting in flame pulsation disorder and phase chaos.

To reduce the time and economic costs of freight cableway path planning, according to factors such as geographical information, project requirements, and costs of cableway erection and operation, the requirements of cableway path planning are proposed, and the technical process of automatic path planning is developed. To meet the requirements of cableway erection, an automatic method for screening loading and unloading points based on clustering was presented to achieve the avoidance of cableway to highway, and by the use of three-dimensional terrain data, a search method for trestle positions based on terrain adaptation is proposed. The selected trestle positions can ensure that the distance between the bearing cable of the cableway and the ground meets the construction requirements, and in this way the number of trestles is optimized to reduce the cost of the cableway. The field applications of the proposed method demonstrate that it can provide thousands of cableway paths and trestle positions automatically in around half an hour, showing fast planning and comprehensive data analysis.

This study is aimed at risk assessments of railway schemes in moraine-dammed lake outburst risk areas in the southern Tibet Plateau. Based on the mechanism analysis of 74 outburst events in the study area, a multi-state Bayesian-Network prediction model is established considering the complex relationships among the influencing factors and high uncertainties of the outburst events. Taking the railway line engineering along rivers in moraine-dammed lake outburst risk areas as the hazard bearing body and combining the risk assessment of moraine-dammed lakes with geomorphological characteristics of the river valley, a calculation program of the railway scheme length in various risk areas and an evaluation system for railway schemes considering moraine-dammed lake outburst risk are proposed. In addition, the operation procedure of the proposed method is demonstrated taking the Zhangmu and Gyirong schemes of the China-Nepal railway as examples The results show that the Zhangmu scheme has a smaller route length in the moraine lake distribution area than the Gyirong scheme, and belongs to the short straight scheme from the perspective of route selection; however, the total length of the Zhangmu scheme in each risk area exceeds that of the Gyirong scheme by about 45%. Therefore, the Gyirong scheme is the better selection when considering the moraine lake outburst risk.

Traditional route planning scheme cannot meet the increasing travel demand of travelers in the process of multimodal transportation. To provide personalized route planning scheme based on various travel demands of travelers, public transport timetable is simulated with the integrated circuit card data, and a multimodal transportation network modal is established based on simulated schedule. A dynamic thresholding method is used to establish the personalized travel demand evaluation value model. The depth first search-genetic algorithm (GA-DFS) is designed, and the initial population generation strategy and two-point mutation method based on this combination algorithm are proposed. Finally, three scenarios with different travel demands are assumed, the example data of a multimodal transportation network in an urban area is applied to the modal and the solution algorithm, comparing with the simulated annealing-genetic algorithm (GA-SA) which is widely used. The results show that compared with GA-SA, the proposed algorithm reduces the average number of iterations by 42%, improves the optimization ability by 50% and provides a route planning scheme based on multiple travel demands of passengers.

To make full use of the low-rank characteristics and local neighbor relationship of the traffic data, and accurately recover the missing data in traffic data acquisition system, firstly, the traffic data matrix is pre-interpolated by the low-rank matrix completion model based on kernel norm to obtain the initial estimate of the missing data. Based on this, a graph model that characterizes the local neighbor structure of the data is constructed. Then, a missing value imputation model combining graph regularization and Schatten-p norm minimization is proposed. Furthermore, an optimization algorithm based on alternating direction multiplier framework is proposed to solve the optimization of missing value imputation, so as to obtain the final imputation result. Finally, the real expressway traffic volume and speed data are used to compare the imputation errors of several methods, and the parameter sensitivity of the proposed method is analyzed. The experimental results show that compared with local least squares, probabilistic principal component analysis and low-rank matrix completion, the proposed method reduces the error of traffic data imputation by 3.02%−28.49% when the missing rate is 10%−50% in missing completely at random mode, missing at random mode and mixed missing mode.

High-speed railway signal system is a key infrastructure to ensure the safety operation of trains. Once the equipment or system functional failure happens, it could easily lead to safety accidents. To this end, a fuzzy comprehensive evaluation method is proposed for risk assessment of the high-speed railway signal system from the perspective of functional safety. Based on the improvement of fuzzy comprehensive evaluation and analytic hierarchy process (AHP), this method strengthens the system in terms of threat scenario analysis, and establishes the coupling relationship of risk factors in respective scenarios. Firstly, forty-four threat scenarios of five categories are analyzed, which affects traffic safety in railway signal systems. A hierarchical structure is constructed with system functional safety accidents as the analysis criterion layer and threat classification as the factor layer. Then the weight of each element in the structure is determined according to the subjective evaluation, and a comprehensive risk assessment is performed with the expert system. In addition, the weights of each level and factor can be adjusted dynamically according to the change of risk value of each scenario, making the evaluation result more practical and forming a mapping relation between security risks and signal services. In the evaluation of real operation data of a certain passenger dedicated line, the signal system risk level is relatively low, and evaluation results are basically consistent with other method, which verifies the effectiveness of the proposed method.

Multiple vehicle bus (MVB) is employed to transmit important train operation control instructions and monitoring information, and accurate diagnosis of the fault types of MVB network is the basis of the intelligent operation and maintenance system. To this end, a fault diagnosis method for MVB network is proposed, which combines the active learning and deep neural networks. It adopts the stacked denoising autoencoder to automatically extract physical features from the electrical MVB signals; then the features are used to train a deep neural network classifier for identifying MVB fault classes. An efficient active learning method based on uncertainty and credibility can solve the problems of insufficient labeled samples and high costs of manual labeling in practical application. It can build a competitive classifier with a small number of labeled training samples. Experiment results demonstrate that to achieve a high accuracy above 90%, the proposed method requires 600 labeled training samples, which is less than 2800 labeled training samples required by random sampling method. With the same number of labeled samples, the proposed method can achieve the better performance as to three different metrics than traditional methods.

Train on-board vision system is an important guarantee for the safety of future urban rail transit operations. The detection effect of the on-board vision system will be seriously affected by the low-light environment when the train operates in a closed environment or at night. To this end, a real-time visual enhancement algorithm is proposed for low-light images in a closed railway environment or night driving environment. The algorithm uses a densely connected network (DenseNet) structure as the backbone network to establish a feature-size invariant network. The network extracts image illumination, color, and other information and predicts the light enhancement rate images. These rate maps adjust the light intensity of each pixel on the basis of the nonlinear mapping function. The network enhances the exposure rate of low-light input images through a hierarchical structure from low level to high level. The developed deep learning network model uses self-supervised learning to train the network parameters. The chracteristics of the low-light image and the prior knowledge are utilized to construct the loss function, which consist of three components: exposure loss, colour constancy loss and illumination smoothness loss. The experimental results of low-light enhancement in multiple scenes show that the algorithm can adapt to the exposure value of input images, dynamically adjust the exposure rate for low-exposure and high-exposure regions to improve the visualization of low-light images, and the processing speed can reach 160 fps to meet the requirements of real-time processing. The comparative experiments of railroad segmentation and pedestrian detection before and after low-light enhancement prove that the proposed algorithm can improve the visual detection in a low-light environment. As for testing on the RSDS (railroad segmentation dataset) datasets, the F-value of railroad segmentation is increased by more than 5%, and the false detection rate and missed detection rate of pedestrians in multiple railroad scenes are effectively reduced.

With the continuous development in miniaturization, lightness, high efficiency and invisibility for shipboard integrated power systems, the energy storage component gradually become an important part of power systems. For the three-phase isolated inverter in the power distribution system, the challenges of wide DC voltage range and high AC power quality arise, resulting the need to increase power density as much as possible and reduce noise. A three-phase dual inverter isolated of line frequency topology is proposed, which possesses high power density and low noise. Firstly, the use of multiple technologies significantly increases the equivalent switching frequency to achieve the effects of four times the phase voltage and five levels, which reduces acoustic noise, vibration noise, and the volume and weight of the vibration isolation and noise reduction device. Secondly, the phase-shift modulation strategy on carriers reduces the common-mode electromagnetic noise and the volume and weight of the EMI (electromagnetic interference) filter. Finally, the use of magnetic integration technology and the three-dimensional wound core structure reduces the volume and weight of the isolation transformer and output filter inductor. The experimental results show that compared with the traditional topology, the acoustic noise of the new topology is reduced by 8 dBA, and the vibration noise in whole frequency band is lower than the traditional topology. The volume power density reaches 136 kW/m³, the weight power density reaches 116 kW/t, and the overall efficiency reaches 93.5 %.

In order to improve the ability of automated theorem provers, the premise selection task emerges. Due to the directional nature of formula graphs, most current graph neural networks can only update nodes in one direction, and cannot encode the order of sub-nodes in the formula graph. To address the above problems, a bidirectional graph with edge types to represent logical formulae and a edge-weight-based graph neural network (EW-GNN) model to encode first-order logic formulae are proposed. The model firstly uses the information of connected nodes to update the feature representation of the corresponding edge type, then calculates the weight of the adjacent node to the central node with the updated edge type feature, and finally uses the information of the adjacent node to update the target node in both directions. The experimental results show the edge-weighted-based graph neural network model performs better in the premise selection task, which can improve the classification accuracy by 1% compared to the best model on the same test dataset.

In order to study the influence of facility relationship on layout in corridor allocation problem, An integer programming model considering location and relationship constraints is constructed, and a hybrid clonal selection algorithm based on clonal selection algorithm is proposed to solve the problem. Before clonal operation, a new 2-opt operation based on problem characteristics is added. Then, tabu search operation is carried out for the optimal individuals in the generated population, mutation operation is carried out for other individuals and adaptive mutation probability is set. The model is accurately solved to verify the correctness of the model and the solution results provide a theoretical basis for the algorithm. Applying the proposed algorithm to test 42−49 scale examples of constrained corridor allocation problem and basic corridor allocation problem respectively. The results are compared with clonal selection algorithm, genetic Algorithm, Scatter Search, flower pollination algorithm and fireworks algorithm. The results show that the hybrid clone selection algorithm can achieve the solution effect of the current advanced algorithm and perform better in the examples sko-42-04 and sko-49-03.

As a key enabling technology for the cyber-physical fusion of intelligent manufacturing, the digital twin has drawn extensive concern. And how to build a digital twin model has become a current research hotspot. At present, digital twin models are mostly focused on conceptual abstraction or specific engineering applications, and seldom consider how to construct and apply digital twin models step by step from the perspective of construction methods and processes. Therefore, this paper proposed the digital twin evolution model (DTEM), which divides the construction and application process of the digital twin into three evolution stages, namely digital model, digital shadow, and digital twin. Then, the application methods, key technologies and tool platforms of each evolution stage were discussed. And the typical applications of DTEM were explored, including intelligent equipment, intelligent production, and intelligent operation and maintenance. The applications show that the proposed model provides a feasible technical route and useful application reference for the step-by-step implementation of digital twins in intelligent manufacturing.