To improve the operation efficiency of wagon flow at railroad terminals with multiple marshaling stations, firstly based on the analysis of different types of wagon flow operations at the terminal, the mathematical model is constructed with the goal of minimizing the total costs of wagons entering and leaving a station, wagon flow reorganization, wagon assembly and transship. Secondly, according to the railroad direction and the capacity of receiving and dispatching trains at the station, an initial matching scheme of the train−marshaling station−shunting system is designed, and then the matching scheme is adjusted and tested by using the transship and disaggregation capacity. Thirdly, a double-layer coding scheme for the matching scheme of the train−marshalling station−shunting system is provided, and the group optimization of the matching scheme is completed with the asynchronous iteration updating process of the embedded replacement–disturbance–crossing operation. Finally, a simulation experiment is conducted to test the proposed model and algorithm. The experimental results show that the proposed method can complete the reasonable division of assignments for multiple marshalling stations. Five groups of test tests can obtain the matching scheme of train−marshalling station−shunting system in the calculation time of 369, 422, 516, 641 and 763 s, respectively. In the optimized division of assignments, the proportion of wagon transship in arrival wagons flow can be stabilized at 20%, so as to reduce the number of wagon transships and reduce the cost of traffic flow organization.

With the increasing penetration of carsharing, vehicle overflow and congestion propagation at the level of station and path tend to be serious. In order to describe the influence mechanism of congestion propagation on the operation of carsharing systems, firstly, a queuing network of the carsharing system is built with time-varying and state-dependence properties. Secondly, based on C# language and O²DES framework of discrete event simulation, a simulation model of the carsharing system under dynamic stochastic environment is proposed, which allows for the influence of vehicle–road interaction and congestion propagation. The influence of congestion propagation on the operation of the carsharing system is analyzed in terms of the station and path levels. Finally, a small-scale carsharing system, i.e., three stations in Chengdu, is exemplified. The proposed model and the infinite queuing model in virtual space are compared and analyzed under different transfer ratios, demands and road congestion scenarios. The results show that congestion propagation at the stations and paths will decline the system service rate by 9.3%–16.9%. Compared with the infinite queuing model, the proposed model can better reflect the actual operation of the carsharing system because of considering congestion propagation. When the occupancy rate of the road network reaches 70% (the road network is in moderate congestion), the proposed carsharing system can achieve maximum benefits. The introduction of the carsharing system will bring new changes to the dynamic allocation of road resources. When the proportion of users from public transportation to the carsharing system exceeds 70%, it will intensify the congestion of the road network, which is not conducive to the effective operation and sustainable development of the carsharing system.

A large number of large-span prestressed concrete (PC) continuous rigid-frame bridges (CRFBs) have been built in China with heights of piers up to 100 m or more. They are likely to suffer strong earthquakes, especially in high seismic risks areas of Western China. The main girder and piers are rigidly connected together in CRFBs, and jointly bear the seismic force in earthquakes. In order to promote the seismic research of CRFB, the seismic damage of several CRFBs which have undergone recent earthquakes at home and abroad was reviewed firstly. Then, from the aspects of seismic theory, model tests, seismic isolation (energy consumption) design and post-earthquake repair, hot issues of the main components such as pier, superstructure, foundation, and the seismic performance of the whole bridge are reviewed. The current research shows that the CRFBs have good seismic performance; the high-order effect and the longitudinal bending moment at the pier-beam consolidation have a greater impact on the seismic response of piers. In the model test and theoretical analysis, the cracking and damage of the main girder are easily ignored, and research on pier bottom and foundation isolation has been carried out for low-pier or double-column CRFBs. Finally, future research directions were explored, including the cracking mechanism and seismic control of box girder under strong earthquakes, the high piers composed of new materials and energy-consuming components, the practical isolation technology of the bottom and foundation of high-piers, the identification of earthquake damage and post-earthquake repair of box girders and hollow piers, the mechanism and prevention of the progressive collapse of rigid frame bridges under the action of (near-fault) cross-fault earthquakes.

In order to study the influence of transverse asymmetrical load on the mechanical characteristics of long-span cable-stayed bridges, this paper takes the Changtai Yangtze River Bridge as the engineering background and analyzes the mechanical characteristics of the bridge with a transverse asymmetrical load. In addition, reasonable control measures are put forward to realize the ideal state of the final bridge. Firstly, a fast optimization method of the cable force is proposed for the cable-stayed bridge with a transverse asymmetrical load. Secondly, asymmetrical upstream and downstream cable forces are used to control the main girder torsion caused by the transverse asymmetrical load. The influences of the unbalanced components along the cable-stayed bridge and the transverse direction of the bridge on the alignment of the main girder and the deviation of the main tower are analyzed. Finally, the method of reducing the secondary dead load of the railway is used to control the deviation of the main tower caused by the transverse asymmetrical load. The results show that the main girder torsion can be effectively controlled by the asymmetrical upstream and downstream cable forces. The transverse bending deformations generated by the unbalanced components along the cable-stayed bridge and the transverse direction of the bridge are basically consistent but opposite in direction, which is beneficial to reduce the transverse bending deformations of the main girder. The transverse deviation of the main tower can be effectively controlled by reducing the difference of the transverse asymmetrical load.

To study the influence of the maintenance rail position and the guide vanes on the vortex-induced vibration (VIV) performance of the wide flat box girder section, the Lingdingyang Bridge (a long-span suspension bridge with wide flat steel box girder) is taken as an example. The VIV of the main girder is studied using a 1∶25 scale section model wind tunnel test, and the two-dimensional flow field of the cross section is simulated using computational fluid dynamics. The test results show that increasing l (the distance between the maintenance rail and the bottom edge of the main girder) can significantly improve the VIV performance of a wide flat box girder. When $l\geqslant {W}_{{\rm{b}}}/6$ ($ {W}_{{\rm{b}}} $ is the bottom width of the main girder), the VIV of the girder can be completely eliminated under different wind attack angles. Installing a guide vane with a 17° angle inside or on both sides of the maintenance rail can significantly suppress the VIV of the girder, and the suppression effect is the same. When $l\geqslant {W}_{{\rm{b}}}/10$, the installation of the guide vane can completely eliminate the VIV of the girder. The numerical simulation results show that increasing the distance between the maintenance rail and the bottom edge of the girder and setting a guide vane can both significantly reduce the periodic vortex-induced force of the main girder by eliminating the wake vortex at the inclined web downstream of the section, thereby suppressing the VIV of the girder.

In order to study the seismic performance of low-yield-point corrugated steel plate shear walls (CSPSWs), a new type of lateral-resistant load system, the finite element software ABAQUS was used to numerically analyze the seismic energy dissipation behavior of 16 CSPSWs under lateral monotonic and cyclic loads. The yield strength and thickness of the infill CSPSWs were selected as the key parameters in the parametric study of the new wall system to assess their impact on the lateral resistance performance, hysteresis performance, stiffness degradation, ductility, and energy dissipation. The results show that the low-yield-point CSPSWs possess the same initial stiffness as that of the ordinary steel plate shear wall, but has a weaker lateral resistance than the latter. Compared with the ordinary-yield-strength CSPSWs, the low-yield-point CSPSWs is capable of developing a fuller hysteresis curve. The energy dissipation and ductility of the new wall system are also better than those of the ordinary-yield-strength CSPSWs. As the yield strength of the CSPSWs decreases, the ductility and energy dissipation of the low-yield-point CSPSWs increase, and the degradation of structural horizontal stiffness accelerates. As the thickness of the low-yield-point stress CSPSWs increases, the initial rigidity and structural energy dissipation increase, while the load-bearing capacity changes less.

Since U-shaped steel sheet piles (USSSPs) are prone to buckle, a dynamic coordination coefficient method for theoretically calculating the local stiffening effect of USSSPs is proposed, and a formula for calculating the critical buckling load of locally stiffened USSSPs is established. The influence of different stiffening areas, stiffening positions, and number of stiffened plates on the stiffening effect (i.e., critical buckling load) is analyzed. The study shows that when the total stiffening area (i.e., cumulative width of stiffened plates) is constant, for the USSSP with determined length, there is a stiffened plate arrangement scheme to maximize the critical load value of the component, and the local buckling law caused by too small width of a single stiffened plate when the number of plates is too large is revealed. As an example, the critical load value of the component of the USSSP with a length of 10 m can be increased by 13.55% with the change of the position of the single stiffened plate, and the optimal theoretical scheme of stiffened plate arrangement corresponding to different total stiffening areas of the USSSP with a length of 20 m is obtained. This paper provides a theoretical basis and reference for the design of the local stiffening scheme of USSSPs in practical engineering.

Due to the requirements for the appearance of civil buildings and the production process of industrial buildings, abnormal internal joints may be formed in the partially-encased steel-concrete composite structures due to misalignment and change of beams and columns. In order to investigate the shear resistance of this type of abnormal internal joints, low circumferential reciprocal load tests were conducted on one conventional internal joint and three profiled internal joints with 1∶2 scale-down model specimens. The damage morphology, hysteretic energy dissipation, load bearing capacity, and ductility properties of the internal joints were analyzed using the height of the beam dislocation on both sides of the column and the height of the single side beam section as variable parameters. The experimental results show that the hysteresis curves of all specimens present a symmetrical and full shuttle shape. The equivalent viscous damping coefficient is between 0.598 and 0.618 and the displacement ductility coefficient is between 3.28 and 4.96, showing good performance in energy dissipation and deformation. Compared to the conventional internal joint, the load bearing capacity of the three profiled internal joints formed due to misalignment and variable beams is increased by 6.1%, 14.0% and 15.0%, respectively, and their displacement ductility factor is increased by approximately −26.6%, 11.0% and −14.1%, respectively, with insignificant ductility performance patterns and little change in energy dissipation capacity, strength and stiffness degradation. For class Ⅰ heterogeneous internal joints with the same cross-sectional dimensions of the left and right beams but completely staggered (i.e., the dislocation height greater than the beam height), they can be designed according to T-shaped edge joints. Based on the joint domain force transfer mechanism, a shear calculation model for class Ⅱ heterogeneous internal joints was established, a formula for calculating the shear bearing capacity was proposed, and the test results agreed well with the theoretical calculation results.

In the discrete element numerical simulation, different sample preparation methods will lead to differences in soil void ratio and uniformity, which will affect the simulation results of bearing capacity of shallow foundation. Therefore, it is necessary to analyze the influence of different sample preparation on bearing capacity of shallow foundation. Four methods (e.g., particle amplification method, distribute method, grid method, and under compaction method) were used to prepare the samples of cohesionless sand, and the samples were balanced under the gravity field of 10g. The void ratio, horizontal stress and vertical stress of soil at different positions were monitored by measuring circle, and the average void ratio e and the lateral earth pressure coefficient K₀ value less than 1 were obtained. The influence of different sample preparation methods on the bearing capacity of shallow foundation was studied by placing rigid wall on the surface of the sample and loading at the same speed to simulate the bearing capacity test of shallow foundation. The results show that the porosity ratios of samples generated by the GM and under compaction method are closer to the original target porosity ratio with an error of about 3.5%. In comparison, the porosity ratios generated by particle amplification method and distribute method are smaller than the target porosity ratio, with an error of about 20.0%. Additionally, GM presents the most homogeneous sand samples, followed by the under compaction method, distribute method and particle amplification method, respectively. Due to the varying porosity ratios and K₀ of samples, the obtained bearing capacity of shallow foundation also changes. The relationship of bearing capacity obtained by different sample preparation methods in the simulation of bearing capacity of shallow foundation is : GM < underlayer compaction method < particle size amplification method < distribute method.

The coupling effect of slurry and soil stress has a significant influence on the slurry diffusion law of split grouting, which should be fully considered in the design phase of split grouting. The split grouting is regarded as a circular expansion process in an infinite plane domain. Based on the Newtonian constitutive equation, variation characteristics of the slurry flow field are analyzed. The sand layer under the splitting channel is regarded as a semi-infinite elastic body. The distribution equations of the splitting channel width and the slurry pressure under the uniform load are derived by elastic mechanics. By setting different slurry viscosity and sand elastic modulus parameters, the basic mechanism of split grouting in sandy soil under coupling effect is revealed. The results show that the slurry pressure decreases rapidly at the orifice and the far end, but changes slowly in the middle section. The width of the splitting channel is basically determined by the slurry pressure, and has a distribution trend the same as that of the slurry pressure. The viscosity of slurry and the elastic modulus of sand are important factors affecting the splitting diffusion radius , and both are positively related to the diffusion radius. The slurry viscosity is positively related to the splitting width, while the modulus is negatively related to the splitting width. Finally, a comparison was made through a project under construction, and it was found that the deviation between the theoretical calculation value and the actual excavation value was 12%−15%, which basically met the expected requirements.

To study the mechanical properties and energy evolution law of red-bed mudstone, five different confining pressure gradients (200, 250, 300, 350, and 400 kPa) and different rate stress paths between the axial and confining pressures are used to perform indoor cyclic loading and unloading tests, taking red-bed mudstone in the Suining area, Sichuan Province, as an example. The effects of the confining pressure on the mechanical properties, total strain energy density, elastic strain energy density, and dissipated energy density of the red-bedded mudstone and the evolution law between the stress-strain and energy are obtained. The results show that the total energy density, elastic strain energy density, and dissipated energy density of the axial stress input generally increase first and then decrease in a normal distribution from the initial loading and unloading to the end of the test. During the pre-peak stage, the difference between the elastic strain energy density and the dissipated energy density increases with increasing number of loading and unloading cycles. The confining effect of the confining pressure improves the bearing capacity of the samples, and the total energy density, elastic strain energy density, and dissipated energy density increase with increasing confining pressure. During the post-peak stage, when the confining pressure ≤ 300 kPa, the dissipated energy density is lower than the elastic strain energy density; meanwhile, when the confining pressure > 300 kPa, the dissipated energy density is higher than the elastic strain energy density, indicating that increasing the confining pressure further inhibits the release of elastic strain energy density after failure. Under the same confining pressure, the compressive strength and elastic strain energy density in the dry and saturated states are negatively correlated with the increase in the number of wet-dry cycles. For the same number of dry and wet cycles, the compressive strength and elastic strain energy density in the dry state are higher than those in the saturated state, showing a positive correlation.

In view of the risk of excessive surface uplift in rectangular pipe jacking construction under the condition of shallow-buried soil, the finite element software MIDAS GTS was applied to simulate the dynamic process of large-diameter rectangular pipe jacking, and to explore the law of surface uplift caused by different thickness of soil covering, different jacking pressure of excavation face and different friction between pipe joint and soil, and to compare and analyze with the field monitoring data. The results show that the uplifting rule of surface monitoring points during pipe jacking is basically consistent with the numerical calculation results. The surface uplift caused by the jacking pressure on the excavation surface of the pipe jacking and the frictional resistance between the pipe joints and the formation is superimposed to reach a peak value near the excavation surface. As the thickness of the overburden decreases, the peak value of the surface uplift increases linearly. When the jacking pressure of excavation face is 100 kPa and the friction resistance of the pipe joint is 35 kPa, the peak value of surface uplift increases about 6.00 mm when the soil jacking thickness decreases for 0.1H (H is the height of rectangular pipe jacking). The growth law of surface uplift in shallow-buried soil jacking pipe with the increase of jacking pressure and joint friction can be divided into two stages: slow growth stage and nonlinear rapid growth stage. In order to reduce the risk of overburden soil being broken in shallow-buried soil pipe jacking project, it is necessary to take reasonable measures to control the jacking pressure and its fluctuation on the excavation surface, and to form an antifriction slurry sleeve with uniform thickness around the pipe joint, so as to ensure the stability of stratum and the safety of pipe jacking project.

In order to scientifically address the uncertainty caused by material properties and complex geological environments in tunnel engineering and make the support design more safe and reasonable, a tunnel reliability design and calculation method based on reliability theory is proposed. Firstly, according to the geometric meaning of the reliability index, an optimization model considering correlation and non-negativity of parameters is established, which is limited in the original space based on Nataf transformation and Cholesky decomposition. Secondly, the design points are obtained directly by using the optimization function embedded in the numerical software without the need to calculate the partial derivatives manually, and the probability of failure is calculated by the important sampling method at the design point, so as to form a real reliability calculation method considering convergence and accuracy. Then, highly nonlinear numerical examples and non-circular tunnel problems are presented to validate the effectiveness and adaptability of the proposed method. Finally, the proposed method is applied to design and calculate the tunnel support resistance based on the performance function concerning the deformation failure of surrounding rock. Results of the example analysis show that the proposed method can obtain design points at a low cost, and the relative error with the results of the Monte Carlo method is less than 1.0%. In addition, the proposed method can address the non-circular tunnel reliability without analytical performance function with the help of response surface function with cross terms. The proposed method can obtain accurate design values of support resistance with a relative error of less than 0.5% with the Monte Carlo method. Additionally, parameter analysis shows that the sensitivity of the internal friction angle is greater than that of the cohesion and deformation modulus.

In order to solve the key technology of oxygen supply in high-altitude tunnel construction, the influence of oxygen supply concentration and labor power on labor intensity in a high-altitude tunnel is studied. Through field measurement in Guigala Tunnel in Dazi District, Lhasa, Tibet, the average energy metabolic rate is used as the indicator to measure labor intensity. The lung flux data of six testers under different labor intensities (labor power of 50, 75, and 100 W) and oxygen supply concentrations (20.9%, 25.0%, and 29.0%) are collected by using the lung ventilation meter, and the data are then converted into the indicator of labor intensity. The mind evolutionary computation back-propagation (MEC-BP) neural network is used to fit the labor intensity indicator. The results show that the goodness of fit of the MEC-BP neural network is slightly higher than that of GA-BP and BP neural networks. Experimental tests and MEC-BP neural network fitting data show that under low labor power of 50 W, the average energy metabolic rate of construction personnel changes slightly with oxygen concentration, with a maximum value of about 0.1 kJ/(min•m²). Under high labor power of 100 W, an oxygen supply concentration of 25% can be used as the reference value of oxygen supply concentration at the plateau of 4 200 m.

In order to adapt to the large-scale construction of urban rail transit and meet the needs of vibration reduction, a new type of prefabricated ballastless track structure is proposed. The vehicle ballastless track coupling dynamic analysis model and indoor full-scale test model are established to study the mechanical properties of the vibration reduction prefabricated track structure. The results show that the system dynamic response indexes of vibration damping assembled track structure under the action of train dynamic load are within the safety limit, meeting the requirements of driving safety and comfort. The maximum compressive strain of track structure concrete in static load test is −129.6 με, which is far less than the limit strain of concrete. After unloading, the residual deformation is very small, the track elasticity is good, and the structural bearing capacity meets the requirements. Under the action of fatigue load, the overall deformation of the structure is small, no crack is found, and the anti fatigue performance is good. The track vibration decreases layer by layer from top to bottom, the frequency is between 100−125 Hz, the maximum vibration reduction effect can reach 12.4 dB, the structure vibration reduction effect is significant and the vibration reduction stability is good.

In order to explore the bonding and anchoring performance of the reinforced concrete-filled steel tube (R-CFST) under cyclic loading, monotonic loading and cyclic loading tests were carried out on R-CFST specimens. The monotonic loading test is used to compare and analyze the response relationship between the bonding force and slip. Based on the monotonic loading test results, the bond behavior was experimentally investigated for several R-CFST specimens under uniaxial cyclic loading. The factors considered include the stirrup arrangement, the bond length, the rebar diameter, and the number of loading cycles. The effects of these factors on the bond behavior were explored. The bonding force degradation mechanism and failure modes of different specimens were analyzed. The results show that the stirrups increase the ability of the specimens to withstand cyclic loading; the bonding force and ductility of the R-CFST specimens increase with the increase of the bond length and the diameter of the steel bars; and the cohesive force and ductility decrease with the increase of the loading times of the cyclic load in the early stage. When the reinforced steel tube concrete bond fails, the entire bond zone is relatively smooth, and the deformation rib imprint is not obvious.

To avoid frost damage problems in cold regions tunnels, a solar shed was introduced as one of the insulation measures. First, a three-dimensional transient heat transfer numerical model was established to obtain the tunnel temperature field under natural ventilation, and the reliability of the numerical calculation model was verified through on-site temperature monitoring. Then, the temperature rise effects of the solar shed were demonstrated through numerical methods, and the effects of the solar shed’s speed, length, span, and ambient temperature on the temperature rise effects were further investigated. Finally, a sensitivity analysis was conducted on these factors to study their impact on temperature rise. The study results show that the solar shed has a good temperature rise effect when the wind speed is lower than 1.0 m/s. The wind speed was negatively related to the temperature rise effects. The air temperature at the tunnel entrance increases by 0.71 ℃ for every 50 m increase in length. The influence of the span on temperature rise effects is not obvious. The temperature rise effects become more and more significant as the ambient temperature becomes lower. The sensitivity of the above factors to temperature rise effects is ranked as follows: wind speed (0.39) > length (0.30) > ambient temperature (0.19) > span (0.12), of which the wind speed is the most sensitive factor and should be considered in the application. Therefore, the recommendation for the use of a solar shed is in environments with low wind speed, long tunnel lengths and small tunnel spans.

To explore the influence of the braking structure/track structure on the frictional self-excited vibration of the wheelset−track−brake system on braking sections of high-speed trains, firstly, combined with on-site investigation, the finite element model of the wheelset−track−brake system of CRH3 high-speed trains is established. Then, the frictional self-excited vibration characteristics of the wheelset–track–brake system under the effects of wheel-rail stick-slip and brake roll-slip are analyzed with the complex eigenvalue method. Furthermore, the influence of the surface texture in the braking structure on the frictional self-excited vibration of the whole system is examined. The parameters of the fasteners in the track structure are parametrically analyzed. The least squares method and the particle swarm optimization algorithm are used to obtain the optimal solution of the fasteners parameters to suppress the rail corrugation. The results show that the main frequency in the frictional self-excited vibration of the wheelset–track–brake system caused by wheel-rail stick-slip and braking roll-slip is 526.75 Hz on the braking section, which is close to the characteristic frequency of the on-site corrugation. The frictional self-excited vibration of the whole system may be the main cause of rail corrugation. The rail corrugation on the braking section can be effectively suppressed by the brake pads or brake discs with surface texture, and the grooved brake pads have the best suppression effect. When the vertical stiffness of the fastener is 65.5 MN/m, the lateral stiffness is 46.0 MN/m, the vertical damping is 84.0 kN·s/m, and the lateral damping is 23.5 kN·s/m, the rail corrugation on the braking section of high-speed trains can be effectively suppressed.

Aiming at the increasingly serious problem of high-order wheel polygonal wear, based on the wheel-rail system rotor dynamics model, wheel-rail contact model and wear depth model, the generation and evolution model of wheel polygonal wear is established. By analyzing the variation of speed and wheel mass eccentricity, the regularity of generation and evolution of wheel polygonal wear are revealed, and the field tracking data are verified. The influence of system parameters on polygonal wear is studied by modal analysis and sensitivity analysis. The results show that the generation and evolution of high order wheel polygonal wear follows the law of “fixed frequency and divisible”. When the fixed frequency of 580 Hz divides wheelset rotation frequency, the wheel wear will evolve into 19^th order polygonal, otherwise it will tend to be uniform. This fixed frequency is mainly derived from the 2^nd-order bending mode of the wheelset and has the greatest sensitivity to the axle diameter. The fixed frequency of the wheel can be changed by running at constant rotating speed and increasing the axle diameter, which can effectively suppress the polygonal wear of the wheel.

To realize the condition-based maintenance of rail corrugation in a data-driven way, a time-space intensive measurement method of rail corrugation is proposed. First, the intelligent terminal is used to detect vibration and noise at a car body of the train formation, the acceleration waveforms in three directions of different car bodies were matched to obtain the estimated time delay, and the error of the train speed and mileage estimation was corrected. Second, the car voiceprint data is analyzed with the voiceprint spectrum energy method, and the corrugation-noise ratio index is defined to quantify the ratio of rail corrugation noise energy and its high-order harmonic energy to the total noise energy, as the basis for automatic identification of rail corrugation. Finally, the inverse mapping relationship between the train response and rail corrugation sate is established to obtain the rail corrugation wavelength and mileage information when the corrugation-noise ratio exceeds the limit. The field test in metro line is taken as an example, where a rail corrugation measurement trolley was used to measure the shortwave irregularity of the rail surface on 1.6 km rail, and the measured peak-to-peak value in the wavelength range of [0,50] mm was compared to the corrugation-noise ratio of voiceprint data. The results show that, when the threshold of the corrugation-noise ratio is set as 0.2, the rail corrugation identified by the voiceprint data is consistent with the line distribution, which verifies that this method can enhance data evidence in evaluating rail corrugation state.

The product quality of batch process is closely related to the process characteristics or the process reaction principle. In order to solve the problems of multistage, time sequence and dynamics of batch process data, a multistage quality prediction method based on nonlinear autoregressive with exogeneous inputs (NARX) neural network is proposed. First, the batch process data are divided into stages by the K-means, and the data dimension is reduced using the kernel entropy component analysis (KECA). These works can improve the prediction efficiency of the subsequent process while ensuring the stage characteristics of the input data. Then, the NARX prediction model is constructed in each stage, and the number of hidden layer nodes of the network is optimized by using the improved bat algorithm (MBA) to realize the online quality prediction of batch processes. The penicillin simulation data is used to verify the effectiveness of the proposed method. The results show that the open-loop structure of the NARX neural network has a good prediction effect, and the data dimensionality reduction method of the KECA is more conducive to subsequent quality prediction research. MBA optimizes the number of hidden layer nodes in the network with higher efficiency and better stability. The stage division can improve the prediction performance of the batch process to a certain extent. The proposed multi-stage quality prediction model has higher prediction performance, its root means square error and the mean absolute percentage error reached 0.02 and 1.48%.

The implementation of the national strategy, actively addressing population aging, has underscored the importance of paying attention to the older adult population, which has become a crucial demographic group. Previous research has predominantly assumed that the effect of the built environment on the travel behavior of older adults is spatially fixed, failing to account for spatial heterogeneity. Therefore, to address this research gap, the propensity to travel, a travel behavior indicator, is analyzed using data from the 2011 Hong Kong Travel Characteristics Survey, and built-environment attributes are assessed using geo-data. A three-level random-intercept binary logistic regression model (level 1: individual, level 2: household, level 3: street block) and a geographically weighted binary logistic regression model are then developed to establish the complex relationship between the built environment and the propensity to travel of older adults, and the association is visualized with the help of ArcGIS platform. The findings demonstrate that population density, land use mix, intersection density, and streetscape greenery have a positive association with the propensity to travel of older adults, while accessibility to the metro and parks does not significantly affect this propensity. Moreover, the effects of all built-environment attributes on the propensity to travel vary across space. Specifically, the local effects of land-use mix on the propensity to travel are bi-directional: positive in the western part of the city and negative in the eastern part.

To enhance the ability of automated theorem provers to select relevant premises from large-scale premises of first-order logic problems, a symbol weight calculation formula is first proposed, obtaining various weights corresponding to different symbols based on the frequency of symbols in the problem. Secondly, the correlation calculation formula is proposed, using assigned symbol weights to compute the correlation between a premise and the conjecture in a problem. At the same time, the adaptive correlation boundary is studied, which is used to determine whether the premise is correlated with the given conjecture. Finally, both processes of premise selection and automated reasoning are interactively combined in automated theorem provers, achieving the goal of stopping the premise selection process in time when the relevant premises are fully selected. The experimental results show that in the optimal case, the proposed premise selection method can reduce the average number of premises involving in proving process from 1876 to 174, and compared with widely used premise selection methods E-SInE and Vampire-SInE in automated theorem provers, the propoed method can help the automated theorem prover E improve the proof rate by 19.49% and 10.49% respectively on the MPTP2078 benchmark.